SAB Review Draft

Valuing Mortality Risk Reductions
for Environmental Policy:
A White Paper

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

DRAFT
December 10, 2010

For consultation with the Science Advisory Board-Environmental Economics Advisory Committee

1


-------
SAB Review Draft

1	Introduction	3

1.1	Key topics	3

1.2	Roadmap	5

2	Background	7

2.1	The valuation challenge	7

2.2	Existing EPA Guidance	10

2.3	Recommendations from prior expert committees	11

3	Key Issues for EPA	14

3.1	Fundamental Concepts and Recommended Terminology Changes	14

3.1.1	Fundamental Valuation Concept	14

3.1.2	Change in metric and terminology	15

3.2	Altruism and willingness to pay for mortality risk reductions	17

3.3	Valuing cancer risks	20

4	Review of stated preference and hedonic wage studies	26

4.1	Stated preference studies	28

4.1.1	Recent meta-analyses of SP studies	29

4.1.2	A new meta-analysis dataset	31

4.2	Hedonic wage studies	35

4.2.1	Data sources	36

4.2.2	Estimation issues	37

4.2.3	Recent meta-analyses of hedonic wage studies	38

4.2.4	A new meta-analysis of hedonic wage studies	41

5	Methods for Combining Data	46

5.1	Meta-analysis	47

5.1.1	Parametric distribution	47

5.1.2	Classical econometrics	48

5.1.3	Bayesian estimation	51

5.2	Structural benefit transfer	53

5.2.1	Static preference functions	55

5.2.2	Life-cycle preference functions	57

6	Conclusions	59

6.1	Addressing key issues: terminology, altruism, cancer valuation	59

6.2	Longer term analytical directions	60

6.2.1	Meta-analysis	60

6.2.2	Structural Benefit Transfer	61

6.3	Other research directions	61

References	63

Tables and figures	74

Appendix A	89

1

2


-------
2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

SAB Review Draft

1 Introduction

The valuation of human health benefits is often a crucial, but sometimes controversial, aspect of
the application of benefit-cost analysis to environmental policies. Valuing the reduced risks of mortality,
in particular, poses a special set of conceptual, analytical, ethical and empirical challenges for economists
and policy analysts. This white paper addresses current and recent U.S. Environmental Protection
Agency (EPA) practices regarding the valuation of mortality risk reductions, focusing especially on
empirical estimates of the "value of a statistical life" (VSL) from stated preference and hedonic wage
studies and how they might be summarized and applied to new policy cases using some form of benefit
transfer. Benefit transfer concepts will be highlighted throughout the paper, since any application of
existing empirical estimates of values for health risk reductions to new policy cases is inherently a benefit
transfer problem.

The main intended audience for this paper is EPA's Science Advisory Board-Environmental
Economics Advisory Committee (EEAC). The main objectives of the paper are to highlight some key
topics related to the valuation of mortality risks, and to describe several possible approaches for
synthesizing the empirical estimates of values for mortality risk reductions from existing hedonic wage
and stated preference studies for the purpose of valuing mortality risk reductions associated with future
EPA policies. Some of these approaches could be implemented in the short term, but others will likely
require longer term research. We are soliciting general feedback and specific recommendations from the
SAB-EE AC on each of these key topics and approaches.

1.1 Key topics

We highlight several issues in this paper, offering preliminary recommendations where we feel
conclusions can be supported by existing data and methods. In other cases we describe alternative
methods, data and data gaps, and possible future directions, with the intention of soliciting meaningful

3


-------
25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

SAB Review Draft-

feedback from the EEAC. The key topics addressed in this paper—loosely ordered from short- to longer-

term tasks—include:

•	Improving communication by reporting value estimates in terms of risk changes rather than "statistical lives."
We fear, as do others, that the prevalence of such terms of art as "the value of a statistical life" has
contributed to unnecessary confusion and consternation among decision-makers and members of the
general public. We aim to ease these communication difficulties by replacing the VSL terminology
with the straightforward term "value of mortality risk" (VMR). The "units" associated with the
mortality risk change must be clearly delineated and in this paper we report the units in terms of
willingness to pay for a reduced risk of 1/1,000,000 or a "micro-risk," following Cameron (2008) and
Howard (1989). We believe that this term provides a more accurate description of the fundamental
valuation concept that underlies the marginal willingness to pay for risk reduction, and that this
choice of measurement unit is a more natural one considering the typically small (relative to the full
suite of risks from all hazards) changes in individual-level risks resulting from most environmental
policies.

•	Alternative approaches for updating EPA's best central estimate, or range of estimates, of the willingness-to-
pay for mortality risk reductions for use in regulatory impact analyses. EPA is interested in updating its
guidance to better reflect the existing estimates of mortality risk reduction values in the revealed and
stated preference literatures. Specifically, how can the empirical results (described below in Section
4) be used to revise EPA's mortality risk valuation guidance in the form of a revised point estimate or
range or benefit transfer function?

•	Incorporating a cancer differential into mortality risk valuation guidance. We discuss the possibility of
adding a "cancer differential" (often called a "cancer premium" in the literature) to the standard
(non-cancer) estimates of mortality risk reduction values, specifically for use in analyzing policies
expected to reduce carcinogenic pollutants. EPA first raised the issue of a cancer premium with the

4


-------
49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

SAB Review Draft:

EEAC in 2000 (USEPA 2000b), but the literature has developed considerably since that time. Given
its importance for the valuation of environmental health risks in particular, we review the current
literature and recommend including a cancer differential in future guidance.

•	The role of altruism in valuing risk reductions. The role of altruistic motives for improved health and
safety is typically ignored in most benefit-cost analyses but may have important implications for
estimating individuals' willingness to pay for environmental improvements. We review several
recent studies that examine the role of altruism in benefit-cost analysis and highlight the potential
relevance of these findings for the valuation of mortality risk reductions, in particular their
implications for interpreting and transferring stated preference estimates of "public" versus "private"
risk reductions.

•	Toward functional benefit transfer. We discuss specific issues that we expect to arise in applying both
classical and Bayesian meta-regression techniques to new datasets of stated preference and hedonic
wage value estimates described in this paper, as possible approaches for developing a benefit transfer
function. We also discuss the structural benefit transfer approach, which involves specifying a direct
or indirect utility function, including parameters that can describe the relevant attributes of the risk to
be evaluated, and then deriving analytical expressions for observable economic variables that can be
used to calibrate the parameters of the preference function. Developing a valid benefit transfer
function, using either meta-regression or a structural approach or some combination of these, is a
longer-term task than the others mentioned above, but EEAC feedback on these issues would be very
helpful in shaping EPA's research agendas in these areas.

1.2 Roadmap

The remainder of this white paper is organized as follows. Before we address our key topics in
more detail, Section 2 provides background discussion that (1) describes the valuation challenge facing
the Agency and the differences in the contexts underlying existing mortality risk reduction value

5


-------
73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

SAB Review Draft

estimates and the policy scenarios we seek to analyze; (2) briefly summarizes EPA's most recent
guidelines for valuing mortality risk reductions (USEPA 2008);1 and (3) recaps the main
recommendations from several recent expert advisory committees to EPA on the valuation of human
health risk reductions and the use of meta-analyses for combining estimates from different studies.

With this context in mind, in Section 3 we describe and discuss three of the key topics of this
whitepaper: terminology and metrics, cancer risk valuation, and altruism. In Section 4, we review the
empirical mortality risk value estimates from the stated preference and hedonic wage literatures,
including recent meta-analyses of these literatures. The discussion of the stated preference literature
includes a newly assembled database of stated preference estimates of mortality risk reduction values in
anticipation of an updated meta-analysis. We also review and extract value estimates and other
attributes from hedonic wage studies that have provided estimates of the VSL, with selected studies
spanning 1974 to the present. We discuss strengths and weaknesses of these studies for application to
environmental policies.

In Section 5 we discuss alternative approaches for synthesizing the estimates from these
literatures as a necessary step for updating EPA guidance. A longer term goal is to develop a benefit
transfer function for valuing mortality risk reductions, rather than relying on the current practice of
transferring a single central point estimate. We discuss two basic approaches for developing such a
benefit transfer function: meta-analysis and structural benefit transfer. Meta-analysis uses statistical
regression techniques to quantify the influence of study, policy, demographic, and possibly other
variables on the willingness to pay for health risk reductions. The structural benefit transfer approach
involves specifying a direct or indirect utility function and then deriving analytical expressions for
observable economic variables that can be used to calibrate the parameters of the preference function.

1 These are reflected in EPA's revised Guidelines for Preparing Economic Analyses (2008).

6


-------
95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

SAB Review Draft

Section 6 concludes with summaries of the key topics and needs for both short-term guidance and longer-
term research.

2 Background

2.1 The valuation challenge

Benefit cost analysis is a useful tool that provides detailed information on a wide variety of
consequences associated with environmental policies. Benefits are based on what individuals would be
willing to pay for risk reductions or for other improvements from pollution reduction. Costs are
determined using the value of the resources directed to pollution reduction. As safeguarding human
health is among the EPA's primary goals, to develop more complete and more accurate benefit-cost
analyses of its policies, EPA must estimate individuals' willingness to pay for reductions in health risks
from environmental harms. Ideally, benefit-cost analysis of policies that reduce health risks would
account for all of the factors that may cause willingness to pay to vary across different types of policies
and individual characteristics and circumstances. The literature has indicated that these factors may
include the sources of risk affected by the policy (e.g., hazardous air pollutants, water contamination,
etc.), the resulting health conditions (e.g., cancer, cardio-respiratory diseases, gastro-intestinal diseases,
etc.), how the policy affects the timing of morbidity and mortality risks across each individuals' life span
(i.e., how it shifts the "survival curve"), the income and other personal characteristics of the affected
individuals, and how the changes in risks are perceived by those individuals. While addressing all of
these factors simultaneously is currently empirically infeasible, there are three challenges that we
highlight for their direct relevance to EPA.

First, fundamental to this valuation challenge is that the risk reductions provided by EPA policies
are inherently public in nature, unlike, for example, private purchase decisions. The distinction is
important because individuals may reasonably value risk reductions from public policies differently than

7


-------
118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

SAB Review Draft

those from private actions even if their own mortality risks are affected in a quantitatively identical
manner. Such differences could be due to differences in "controllability," "dread," or other tangible or
intangible factors (e.g., Slovic 1987, Savage 1993, Chilton et al. 2006). Furthermore, public policies raise
issues about altruistic values for risk reductions to others, something that may be of particular relevance
for environmental risks. EPA would like to use the existing literature to evaluate the extent and nature of
altruistic values and consider how to formulate mortality risk valuation guidance accordingly. We
address altruism in greater detail in Section 6.3.

A second major challenge for the valuation of mortality risk reductions for environmental
policies is the intertwined nature of morbidity and mortality risks. Environmental policies generally do
not reduce the risks of fatal workplace or automobile accidents, for example, which provide the context
for many of the mortality valuation estimates in the literature and generally have little or no
accompanying morbidity or period of illness. Ideally, we would use an integrated model that could
estimate willingness to pay for mortality and associated morbidity risk reductions simultaneously.
Developing such a model is beyond the scope of this white paper and current guidance development
effort, and is near the frontier of the empirical valuation literature. Nevertheless, to the extent possible
with currently available data and models, we would like to account for how individuals consider
morbidity in existing estimates of mortality risk reduction values when they always occur together. It
also is important to capture some related losses that may not be reflected in willingness to pay estimates,
depending on context in which they were estimated. For example, reduced health from illness preceding
death is certainly a loss to an individual and his or her quality of life, but may not be reflected in VSL
estimates from the hedonic wage literature, which are based on the risks of workplace injuries that lead to
death. Society also is worse off because of the illness due to the individual's lost productivity, something
that may not be reflected in revealed or stated willingness to pay estimates, depending upon the type of
insurance held by the individual and possibly the scenario description.

8


-------
142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

SAB Review Draft

This issue is of particular relevance to EPA when addressing reductions in cancer risks since
many EPA policies focus on reducing exposure to carcinogens. Ten years ago EPA reviewed the
economic literature on valuing fatal cancer risk reductions and discussed a number of risk characteristics
that may influence people's values, including but not limited to the timing of the risks (USEPA 2000b,c).
The committee recognized many of the issues reviewed by EPA as theoretically valid but empirically
ambiguous, and therefore recommended that "the only risk characteristic for which adjustments to the
VSL can be made is the timing of the risk" (USEPA 2000c p 1). In particular, this recommendation
advised against the application of any differential to reflect preferences for reducing cancer risks relative
to other types of risk because of dread or other factors. With an additional decade of valuation literature
to draw upon, EPA is seeking to re-examine this question using data from the stated and revealed
preference studies described below, as well as other relevant empirical results. We will discuss cancer
valuation in more detail in Section 6.4.

Finally, the empirical literature may allow us to account for the extent to which individuals value
different categories of risks differently in a systematic transfer of benefits. For example, if environmental
risk reductions are valued differently from workplace or auto accidents, regardless of whether the
mitigation is from private or public actions, our guidance should reflect this difference.

It is important to keep the overarching valuation challenge in mind as we begin discussing recent
studies and value estimates. Each study reflects an attempt to measure the value of a reduction in
mortality risk from a specific cause (or small set of causes), in a specific context, among a specific
population. By now there is ample theoretical and empirical evidence to indicate that values for health
risk reductions are not "one-size-fits-all"—that is, they are "individuated" (e.g., Sunstein 2004, Evans and
Smith 2008, Scotton and Taylor 2009). For this reason, we believe that there is great scope for improving
upon the point value benefit transfer approach that has traditionally been applied to mortality risk
reductions based on a central estimate of the VSL. Therefore, we ultimately are seeking both short-term

9


-------
166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

SAB Review Draft

recommendations as well as advice on a longer-term research agenda on how these heterogeneous
studies can best be synthesized for systematic benefit transfers to improve the application of benefit-cost
analysis to future environmental policies.

2.2 Existing EPA Guidance

EPA's draft Guidelines for Preparing Economic Analyses (2008) (hereafter, the draft Guidelines)
retains the recommendation from the 2000 version, a default central VSL value $4.8 million in 1990 real
dollars. This estimate, after adjusting for inflation and real income growth, is to be applied to mortality
risk reductions for all types of policies, no matter the source of the risk.2 The estimate is based on the
mean of a probability distribution fit to twenty-six published VSL estimates. The draft Guidelines also
indicates that the distribution itself can be used for formal uncertainty analysis. The underlying studies,
the probability distribution parameters, and other useful information are available in Appendix B of the
draft Guidelines (USEPA 2008).

The draft Guidelines also retains the 2000 version recommendation that the VSL for mortality risk
reductions should not be adjusted for differences in sources of risk or population characteristics—rather,
these factors should be examined qualitatively. In some cases, the analysis may include a quantitative
sensitivity analysis. Analysts should account for timing when valuing mortality risk reductions, and
should discount the benefits of future risk reductions at the same rate used to discount other costs and
benefits. Because the VSL represents the marginal willingness to pay for contemporaneous risk
reductions, this is typically done by estimating the lag between reduced exposure and reduced mortality
risks, calculating willingness to pay in all future periods when mortality risks are reduced, and
discounting back to the present.

Finally, EPA's draft Guidelines also recommends accounting for increases over time in average
income. This is done by using projections of real GDP per capita and applying an income elasticity
2 We report all estimates in 2009 US dollars unless otherwise noted.

10


-------
189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

SAB Review Draft

estimate. The resulting future (real) VSL will therefore reflect the idea that health risk reductions are
normal goods and so willingness to pay will increase with income.

2.3 Recommendations from prior expert committees

This white paper is one stage in a detailed process that EPA has undertaken with the SAB-EEAC
to improve the Agency's ability to value health risk reductions. Since its review of EPA's Guidelines for
Preparing Economic Analyses (USEPA 2000a) the SAB has offered several specific sets of recommendations
on valuing risk reductions, particularly for mortality risks.

In July 2000 the SAB-EEAC released an advisory report in response to EPA's white paper, Valuing
the Benefits of Fatal Cancer Risk Reduction, which focused on benefit transfer issues associated with using
existing mortality risk values to estimate the benefits of EPA actions on carcinogens, including potential
adjustments that could be made to existing risk values to account for this category of benefits (USEPA
2000b). As noted earlier, after reviewing the white paper and current economics literature, the SAB
concluded that, while many of the issues raised in the white paper were theoretically valid and
potentially important, the empirical literature supported only accounting for latency and for income
growth over time. The SAB-EEAC did not consider other adjustments to EPA's default mortality risk
value to be appropriate for the Agency's primary analyses, but could be addressed separately using
sensitivity analysis.

An August 2001 SAB report, Arsenic Rule Benefits Analysis: An SAB Review (USEPA 2001),
generally supported EPA's estimate of the marginal willingness to pay for mortality risk reductions. The
SAB also offered additional recommendations to account for the time between reduced exposure and
reduced mortality risks. This report coined the term "cessation lag" for this concept and offered specific
recommendations for estimating cessation lags based on the types of risk data available. The SAB review
also clarified that reductions in exposure to carcinogens—that is, exposure per se, aside from the increased
cancer risks that the exposure causes—are not a separate benefit category under a damage function

11


-------
213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

SAB Review Draft

approach to valuing reduced risks. The board noted that it is possible that there is an existence value for
protected drinking water; however, without sufficient empirical evidence to estimate the magnitude of
this value, it cannot be included in the quantitative benefits analysis. Finally, the report indicated that it
is appropriate to add the costs of illness to the willingness to pay for mortality risk reductions when
estimating the benefits of reduced cancer mortality.

EPA further consulted with the SAB-EEAC on additional mortality risk valuation issues in 2004,
developing a strategy to gather additional information on meta-analysis to inform both the SAB-EEAC
and EPA (USEPA 2004b). In 2006, EPA returned to the SAB-EEAC with two documents for formal
review: a white paper addressing how remaining life expectancy affects willingness to pay for mortality
risk reductions, and an expert report on the use of meta-analysis for combining existing mortality risk
value estimates. A 2007 report, SAB Advisory on EPA's Issues in Valuing Mortality Risk Reduction,
responded to both topics (USEPA 2007).

On the subject of life expectancy, the SAB-EEAC noted that there was theoretical ambiguity on
how willingness to pay might change with age (and, hence, remaining life expectancy). The committee
concluded that the existing economics literature does not provide clear theoretical or empirical support
for using different values for mortality risk reductions for differently-aged adults or a constant "value of
statistical life year" (VSLY). Thus, the SAB-EEAC recommended that EPA continue using its traditional
assumption of an age-independent willingness to pay for mortality risk reductions.

To address meta-analysis, EPA assembled a work group of expert statisticians in December 2005
to discuss the meta-analysis of VSL estimates and to examine three existing meta-analyses: Mrozek and
Taylor (2002), Viscusi and Aldy (2003), and Kochi et al. (2006). While the expert workgroup did not
endorse any one of these studies, the panel did encourage the use of meta-analytic techniques for the
analysis of the existing literature on VSL. The workgroup recommended analyzing stated preference and

12


-------
236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

SAB Review Draft

hedonic wage data separately, and offered a set of principles that should be followed in conducting such
an analysis (USEPA 2007).

The SAB-EEAC review of the Meta-analysis workgroup's report stated that meta-regression is "a
useful statistical technique for identifying various aspects of study design or population characteristics
that are associated with differences in VSL," but concluded that meta-regression is "not appropriate [for]
combin[ing] VSL estimates" into a summary measure (USEPA 2007 p i). Rather, the SAB-EEAC
suggested using meta-regression to examine how study design characteristics influence the VSL estimates
and relying on other statistical techniques to determine a central estimate or range of estimates for use in
benefit transfer to new policy cases.

Based on these expert recommendations and other considerations, we believe that updated
reviews and meta-analyses of the stated preference and hedonic wage literatures could help refine the
Agency's central estimate(s) or range of estimates of the marginal willingness to pay for mortality risk
reductions. Studies have shown that values for health risk reductions may depend on differences among
policies and the affected individuals. These factors include the sources of risk affected by the policy (e.g.,
hazardous air pollutants, water contamination, etc.), the resulting health conditions (e.g., cancer, cardio-
respiratory diseases, gastro-intestinal diseases, etc.), as well as how the policy affects the timing of
morbidity and mortality risks across each individuals' life span (i.e., how it shifts the "survival curve").
Therefore, as is widely recognized in most other contexts where some form of benefit transfer is used for
policy analysis, we believe a functional benefit transfer approach should be more accurate than a single
point estimate applied in all circumstances. Consequently, we are interested in exploring approaches for
developing benefit transfer functions that can account for some or all of these factors.

13


-------
257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

SAB Review Draft:

3 Key Issues for EPA

3.1 Fundamental Concepts and Recommended Terminology Changes

3.1.1 Fundamental Valuation Concept

We begin by identifying the fundamental valuation concept that economists aim to estimate
using non-market valuation methods and apply in benefit-cost analyses of policies that reduce human
health risks. Consider a general utility function for an individual i with income X and some health risk

R, among the arguments: L/ = U Y, R. , Z . The vector Z; is included to emphasize that, in addition

to income and risk, the individual's utility (and therefore the willingness to pay for health risk
reductions) also may be influenced by many other factors specific to the case at hand. We will highlight
several of these factors throughout this white paper. The individual's marginal rate of substitution between
income and risk is:

eu ,v dUn dX dU/dR,
dU:=—dY- +—dR = 0 => —L =	L.

8Yi dRi	dRi dU/dY,

This marginal rate of substitution, dYJdR., also can be interpreted as the individual's marginal

willingness to pay (wtp) for a change in risk—that is, the amount of money the individual would be willing
to swap for a small change in risk on the margin.3 This is the fundamental value concept that must be
estimated for use in benefit-cost analyses of policies that may improve human health. With estimates of
these quantities, conditioned as necessary on possibly many observable characteristics of the policy and
the affected individuals, it is straightforward to calculate the total willingness to pay for the risk
reductions that are expected to be produced by the policy: ^. wtp{ x AR;, where i indexes all individuals
affected by the policy, and wtp{ and AR, are the estimated marginal willingness to pay and risk

3 Throughout this white paper, we will use "wtp" to refer to marginal willingness to pay, which will have units of
$/change in risk, and we will use "WTP" to refer to discrete willingness to pay amounts, which will have units of $.

14


-------
277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

SAB Review Draft-

reduction for individual i, both of which may depend on individual-level characteristics and
circumstances.4

It is important to emphasize that this is a marginal value concept—a dollar value per unit change in
risk. These values should be thought of as the slope of a curve at a point, rather than the height of the
curve.5 For practical purposes, the units used to report estimates of these slope values are of no
consequence. They could be reported as dollars per nano-risk (10 9), or micro-risk (10 ' ), or mili-risk (
10 ), etc. As long as the measurement units are known, then the risk changes to be valued can be
expressed in the same units and the correct total value can be calculated. The conventional measurement
units used for reporting these slope estimates are (effectively) "dollars per mortality" risk changes,
usually simply written as "$," where "per mortality" is understood (or misunderstood, depending on the
audience). This quantity was often referred to as the "value of life" in the early literature on the subject
(e.g., Rice and Cooper 1967). While the terminology varies, the quantity is now typically called the "value
of a statistical life," or VSL, where "statistical" has been added to emphasize that valuation is based on
changes in risk rather than the loss of life with certainty.6

3.1.2 Change in metric and terminology

Despite its widespread usage, this particular selection of measurement units for the denominator
of the marginal rate of substitution between income and risk, and the VSL label that has been attached to

4	For ease of exposition we ignore the time dimension here. We will allude to some of the complications that arise in
the more realistic dynamic case, using a life-cycle model, in Section 6.2.2 and Appendix A.

5	Also note that if the risk changes to be valued are large, then the slope of the willingness to pay function may
change over the relevant range and so the marginal willingness to pay ' the change in risk may not give an accurate
estimate of total willingness to pay. For the most part in this white paper we will ignore this complication, though
we do come back to it in an illustrative example in Section 5.2.1.

6	A common way of explaining the meaning of the VSL is based on a population's aggregate willingness to pay for an
aggregate risk reduction. For example, suppose in a town of 1,000 people a policy is enacted that reduces each
person's risk of dying by 1 in 1,000 in a year. Lhen the expected number of avoided deaths (lives saved) by the policy
for the year would be equal to one—a so-called "statistical life." Suppose further that we know (from a survey or
other study) that the average amount that people in the town would be willing to pay for the risk reduction of 1 in
1,000 was $8,000. We then know that the aggregate willingness to pay is $8,000,000 for saving the one statistical life,
so the "value of a statistical life" would be $8,000,000.

15


-------
294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

SAB Review Draft

it, have caused or contributed to needless confusion and controversy, especially among non-economists
(Cameron 2009). Most economists recognize that the "units" associated with the VSL reflect the
aggregation of the small risk reductions across many individuals until that aggregate reflects a total of
1.0, or one statistical life. However, for non-specialists this potentially subtle point is often lost; the
addition of the word "statistical" to the terminology does not seem sufficient to clarify the concept.7

To help reduce the misconceptions that seem to be inspired or aggravated by the VSL
terminology, we propose a change in EPA standard practice such that estimates of health values will be
referred to as the "value of mortality risk" (VMR), and report the associated units using standard metric
prefixes to indicate the size of the risk change and the associated time scale, e.g., $/|jr/person/yr (dollars
per micro[10'6]-risk per person per year) (Howard 1989, Cameron 2009).8

As noted earlier the choice of risk increment for aggregating and reporting risk changes is mainly
one of convenience. However, we believe that explicitly labeling the units of the VMR in this way more
clearly emphasizes that these values refer to small changes in individual-level risks over a definite time
span rather than how much money any single individual or group would be willing to pay to prevent the
certain death of any particular person. It also should be emphasized that the use of a standardized

7	A recent example of the confusion surrounding this concept in the popular press can be found in an AP story titled,
"American Life Worth Less Today" (Bornstein 2008) that opened by saying "[EPA] has decided that an American life
isn't worth what it used to be." The story was referring to an alternate analysis in some air regulatory impact
analyses that used a more recent review of the literature to report a lower VSL than is reflected in EPA's 2000
Guidelines. This story quickly spread throughout the media even appearing on the Colbert Report as EPA's efforts
to "devalue life." Video clip at http://www.colbertnation.com/the-colbert-report-videos/176175/july-14-2008/the-
word—priceless (04:06) Posted on 7/14/2008.

8	Other alternatives to the VSL to better describe marginal wealth-risk tradeoffs have been used or proposed as well.
For example, the UK government uses the term "value of prevented fatality (VPF)," but as described by Wolfe (2007)
this designation confronts the same misinterpretations as VSL. Cameron (2009) suggests a greater departure from
standard terminology not only to communicate that "lives" are not being valued, but also to clarify that "value" itself
should be understood in terms of opportunity costs. After considering several alternatives, the term suggested is
"willingness to swap (WTS) other goods and services for a micro-risk reduction," abbreviated WTS (|-ir). In recent
empirical work, Cameron and DeShazo (2008) report results in terms of micro-risk reductions. Scotton and Taylor
(2009) use the term "value of a risk reduction" (VRR), noting that "explicit consideration of the heterogeneous values
for heterogeneous risks underscores the importance of moving the policy discussion from 'a VSL' to valuation of
marginal changes in fatality risks specific to the type of the risk affected by the policy" (p 23).

16


-------
309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

SAB Review Draft

measurement unit for reporting values for health risk reductions should neither be taken to imply that
the values themselves are invariant across individuals or contexts, nor that these marginal values will be
constant across the full range of relevant risk changes.

For the remainder of this paper we will use the general term "value of mortality risk" whenever
possible. We will report estimates as VMRs, as defined above, to the extent possible, using the VSL
terminology only as necessary in discussing the previous literature.

3.2 Altruism and willingness to pay for mortality risk reductions

We now turn to an overarching conceptual issue that may affect the conduct of benefit-cost
analysis more generally: altruism. The default assumption for most applications of revealed and stated
preference methods for non-market valuation is that individuals' (or households') well-being depends on
their own consumption (interpreted broadly to include market and non-market goods and services) and
is not directly influenced by the consumption or well-being of others. If this assumption is invalid, we
may be concerned that our standard methods of estimating willingness to pay assuming "atomistic"
individuals or households may give misleading results in benefit-cost analysis.

There are at least two ways that altruism may be relevant for the valuation of mortality risk
reductions. First, some stated preference studies are based on surveys that make a distinction between
"public" and "private" risk reductions.9 The difference, if any, between WTP for public versus private
risk reductions may be partly due to altruism, but other factors could be at work as well. For example, a
distrust of government may lead some respondents to express a lower WTP for public risk reductions
provided through government programs compared to those provided through private initiatives. While
stated preference studies may in principle be able to distinguish altruistic preferences from other

9 Few studies explicitly address the public versus private issue. However, for most of the studies it is possible to
assign the estimates to one category: estimates that accrue to an individual only, such as an individual health risk
reduction or the decision to wear a seatbelt or purchase a health care treatment, are "private" and estimates that can
accrue to the individual and others, such as reductions in highway safety-related deaths, are "public." See section 6.1
for more details on the stated preference studies.

17


-------
330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

SAB Review Draft

confounding factors, it is difficult to draw clear conclusions from the existing literature because most
studies that have been conducted to date were not designed to examine altruism per se.10 Therefore, the
proper application of the results of these stated preference studies may depend in part on how altruism
should be treated in benefit-cost analyses. Second, since hedonic wage studies are focused on
compensation received by individual workers for taking on private, job-related risk, the mortality risk
values from hedonic wage studies do not incorporate altruism. Therefore, if (some forms of) altruistic
preferences should be included in benefit-cost analysis, then hedonic wage-based estimates of mortality
risk values may need to be supplemented with separate value estimates that capture altruistic preferences
alone. On the other hand, if (some forms of) altruistic preferences should be excluded from benefit-cost
analyses, then this may influence whether (or how) some stated preference studies should be used for
benefit transfers.

EPA's Guidelines for Preparing Economic Analyses (USEPA 2000a) discussed the role of altruism in
estimating the total benefits of public actions, and noted the key distinctions between paternalistic (or
"safety focused") and non-paternalistic (or "preference respecting") forms of altruism.11 If altruistic
motives are non-paternalistic, then individuals care not only about the benefits others receive, but also
the costs they bear, and most economists who have studied this issue have concluded that it is generally
inappropriate to add these altruistic values for benefits others receive to total willingness to pay. Doing
so could lead to "double-counting" some of the benefits and/or costs. Paternalistic altruism, on the other
hand, should be included in the calculation of total benefits. EPA's Guidelines (USEPA 2000a p 61)
describes the issue as follows:

10	Stated preference studies and the treatment of altruism also may hold promise for identifying preferences related
to equity or environmental justice (EJ) concerns. For example, preferences for reductions in risks for others,
particularly those who may be disproportionately exposed to pollutants (which are often low income and minority
groups typically associated with EJ) could be identified through a well designed stated preference study.

11	Formally, the utility function of non-paternalistic altruists includes others' utility, while the utility function of
paternalistic altruists includes others' consumption of one or more types of private or public goods or services.

18


-------
SAB Review Draft

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

While benefits are generally calculated by summing each individual's WTP for his or her own
welfare, there are conditions under which it is appropriate to include altruistic values, or individuals'
WTP for the welfare of others. Economic theory concludes that if one cares about a neighbor but
respects the neighbor's preferences, and if the neighbor would have to pay for the policy action being
analyzed, then altruistic benefits should not be counted in a benefit-cost analysis. The intuition
behind this result is that, if one respects the neighbor's preferences, one cares about both the benefits
and the costs the neighbor faces. It is therefore inappropriate to add the value one attaches to the
neighbor's benefits without considering the cost implications of doing so. Comparing individual
benefits and costs in this case is the appropriate decision rule.

Altruistic benefits may be counted either when altruism toward one's neighbor is paternalistic or
when one will in fact bear the costs of the project but the neighbor will not. In the first case
(paternalistic altruism), one cares about the benefits the neighbor will enjoy, e.g., from a health or
safety project, but not about the costs the project will impose on him. An example of the second case
would be a project whose costs are borne entirely by the current generation; i.e., the project imposes
no costs on future generations. In this case, altruism toward future generations by the current
generation could legitimately be counted as a benefit.

The conclusions in the Guidelines were based largely on Bergstrom (1982) and McConnell (1997)

369	who demonstrated that the optimal provision of public goods based upon selfish preferences is a

370	necessary and sufficient condition for the optimal provision based on social preferences (including

371	altruistic preferences). However, since the publication of the Guidelines, Flores (2002) has challenged the

372	conventional wisdom that (non-paternalistic) altruism should be excluded from benefit-cost analysis.

373	Flores showed that passing a private values benefit-cost test is a sufficient but not a necessary condition

374	for non-marginal policies to be potentially Pareto improving, except under special circumstances. That is,

375	even if all altruism is non-paternalistic, failure to include altruistic values may lead to the rejection of

376	policies that are potentially Pareto improving. Flores concluded that "benefit-cost analysis with altruism

377	cannot simply be conducted independent of who pays." In a more recent study, Bergstrom (2006)

378	concluded that "The assumptions under which the private values benefit-cost test is necessary for

379	potential Pareto improvements need not always be satisfied;" nevertheless, "Despite these

380	qualifications... for a broad class of economies, a comparison of the sum of private values to the cost of a

381	project is the appropriate test for determining whether it can lead to a Pareto improvement" (p 348-349).

19


-------
382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

SAB Review Draft

Bergstrom's conclusion seems to summarize the prevailing view regarding non-paternalistic
altruism in benefit-cost analysis, especially for policies that would cause marginal changes in
environmental quality (since Flores' counter-examples involved non-marginal changes). Therefore, the
main relevance of altruism for mortality risk valuation lies in the distinction between the paternalistic and
non-paternalistic forms. Including the former but excluding the latter may require supplementing
revealed preference estimates of health risk valuations with a careful selection of results from previous
stated preference studies. Stated preference surveys that elicit only private willingness to pay would
exclude both forms of altruism. One way to include paternalistic but exclude non-paternalistic altruism
would be to design a survey that would inform respondents about health improvements that others
would experience from the policy, but also ask each respondent to assume that all others would be taxed
an amount equal to their private willingness to pay for the policy, so that their utility remains unchanged
(Johansson 1994). It is not clear which if any of existing stated preference studies (many of which are
reviewed below in Section 6.1) were designed this way, so the current body of empirical results cannot
support the separation of paternalistic from non-paternalistic altruism. We recommend additional
research in this area to help estimate paternalistic willingness to pay for environmental policies that
reduce health risks. Additional examination of existing studies may shed light on this issue in the
relative short-term, and we are interested in feedback on this issue.

3.3 Valuing cancer risks

As noted in our description of EPA's valuation challenge, willingness to pay for cancer risk
reductions may be systematically different than that for workplace or auto accidents or other risks not
associated with a lengthy and painful illness. This difference is sometimes referred to as a a "cancer
premium," but we will use the more general term "cancer differential." While not often defined
precisely, the differential is posited as capturing elements of dread and fear of cancer, as well as the pain

20


-------
405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

SAB Review Draft:

and suffering from the period of illness preceding death. It might also include income and household
productivity losses over this period of morbidity.

Several authors have recommended accounting for this differential in benefit-cost analysis of
policies that reduce exposure to carcinogens (e.g., Revesz 1999, Sunstein 2004). To the extent that existing
policy guidance on valuing mortality risk reductions is based on non-cancer risk-wealth tradeoffs, this
would involve an "adjustment" to the default (generic, non-cancer) mortality risk reduction value.
Governmental analyses in the UK have adopted this approach, applying a 100% differential for cancer
risks (HM Treasury 2003).12 In addition, the European Commission has recommended a 50% differential
for carcinogenic pollutants over its default value of preventing a fatality (European Commission 2000).

For the purpose of developing guidance, we are interested in assessing the valuation literature on
cancer risks and any cancer risk differential, both in the short-term and the longer term. Ultimately, this
literature could inform the development of a benefit transfer function, in combination with the stated
preference and hedonic wage estimates described in greater detail below. While such longer-term
research is being conducted, we believe it is reasonable that evidence of systematically different
preferences for cancer risk reductions be part of any recommended short-term guidance.

To inform this discussion, this section contains a somewhat more detailed assessment of the
empirical literature on cancer risk valuation, with a particular emphasis on studies that examine risks in
both cancer and non-cancer contexts. These studies are described in Table 1 in the following categories:

•	studies comparing values for cancer and non-cancer fatal risk reductions

stated preference studies that estimate willingness to pay
risk-risk studies

•	stated preference studies of cancer risks without internal comparisons, and

•	related hedonic property and hedonic wage studies.

12 Specifically, this adjustment is applied for the benefits from asbestos proposals by the UK Health and Safety
Executive (HSE).

21


-------
428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

SAB Review Draft

The first of these categories contains the most direct evidence on any cancer differentials.

Note on Cessation Lag and Latency

Reduced exposure to carcinogens results in reduced cancer incidence after a period of time that
EPA has referred to as "cessation lag," a term originally coined by the SAB in its review of the Agency's
arsenic in drinking water benefits analysis. Cessation lag addresses only reduced risks from reduced
exposure and thus applies best to populations currently at risk. The time between initial exposure and
increased cancer incidence is referred to as "latency" in recent EPA analyses, but it is often used in the
literature in a broader sense to refer to the time difference between a change in exposure and a change in
risk.

Prior SAB-EEAC advice and agency practice has been to estimate cessation lag and latency from
available epidemiologic data, apply a value of statistical life estimate at the time at which cancer mortality
reductions occur, and discount this value back to the present at the rates prescribed in Agency guidance.
The practice has generally been supported by research findings suggesting that individuals discount over
these lag times at rates generally consistent with market rates, although some recent stated preference
studies find near-zero discount rates over latency periods (Hammitt and Haninger, 2010; Alberini and
Scasny, 2010a).

An important issue in estimating a cancer differential is the potential need to consider
differences in the time profile of mortality risks between cancer and non-cancer cases. Earlier studies
were often silent on the issue, but more recent ones have attempted to address it explicitly. Our focus in
this section is on a potential cancer differential that captures the difference in marginal willingness to pay
for reduction of cancer mortality risks relative to that of a non-specific mortality risk holding timing
equal. That is, the differential, in principle, compares a contemporaneous non-cancer risk reduction with
a contemporaneous cancer risk reduction. We recognize that timing may be intertwined with how people

22


-------
451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

SAB Review Draft

perceive and value risk reductions, something that should be considered more fully in any rigorous,
systematic benefit-transfer exercise as we develop guidance.

Stated Preference studies including cancer and non-cancer risks

Several stated preference studies have estimated willingness to pay for both cancer and non-
cancer risks, in large part to examine a possible cancer differential. A few studies have focused only on
cancer risk reductions without an internal comparison to other types of risk. The results of these studies
are somewhat mixed—some have found evidence of a cancer differential (Hammitt and Liu 2004, Tsuge
et al. 2005, Alberini and Scasny 2010a, and Alberini and Scasny 2010b), while a few others found no such
evidence (Hammitt and Haninger 2010, Adamowicz et al. 2008) when looking at whole-household or
public risks. Cameron and DeShazo (2008) found evidence of a differential for some cancers (breast and
prostate) over other cancers (colon, lung, and skin), but not over other health endpoints (heart attacks and
disease).

There have been two risk-risk tradeoff studies specifically examining how preferences for cancer
risk reduction compare to those for automobile accident risk reductions. By asking respondents to choose
among different bundles of risks, these simplified choice experiments aim to estimate the relative values
of various types of risk reductions. They do not, however, provide a willingness to pay for either risk
type and therefore are not included in our reviews of the willingness to pay literature above. Van
Houtven et al. (2008) found a strong preference for avoiding cancer risks relative to automobile accidents
even after controlling for latency and morbidity periods. With a 5-year latency, values for reductions in
fatal cancer risk were approximately three times larger than those for immediate accident risks, declining
to fifty percent larger for a 25-year latency. By contrast, in a study by Magat et al. (1996), the median
respondent was indifferent between fatality risk from auto accidents and lymphoma, suggesting that
cancer mortality is no more 'dreaded' than accidental mortality. It is difficult to draw firm conclusions,
however, because the study did not specify the timing of the risks, and, in particular, any latency

23


-------
475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498

SAB Review Draft

associated with cancer. Therefore, if respondents assumed that cancer risks would be realized after a
latency period then the results suggest that any preference for cancer reductions was approximately
offset by discounting future risks.

Three additional stated preference studies focus on WTP for cancer risks without direct
comparisons to other risks. These do not internally address the question of how cancer risks are valued
differently from non-cancer risks, but may be combined with the results from other studies to address
this question. Focusing on cancer risks from hazardous waste sites Alberini, et al. (2010) estimated a
cancer VSL of approximately $5.6 million (2009 dollars) using the results of choice experiments in Italy.
Carson and Mitchell (2006) examined willingness to pay for installing a water filtration system to remove
trihalomethanes (THM) in public drinking water. Estimated values depend upon an assumed latency
and discount rate, as well as the specific risk reduction, but generally range from $3.4 to $8.0 at the
smallest risk changes for a 25-year latency. Buzby et al. (1995) used a telephone-mail survey to examine
the value of reduced fatal cancer risk from exposure to pesticides in grapefruit, and estimated a value of
statistical cancer fatality at $6.99 million based on exposure assumptions.

Related Hedonic Property and Wage Studies

There are a small number of studies that have estimated WTP for reduced cancer risks using
revealed preference approaches. The results have generally shown that the value of a statistical cancer
case is similar to prevailing VSL estimates from hedonic wage studies. Direct comparison, however, is
difficult without additional assumptions about latency or cessation lag and cancer fatality rates, as noted
for each study.

In the context of hazardous waste, Gayer et al. (2000) and Gayer et al. (2002) employed a hedonic
property framework to estimate the implicit value of a statistical cancer case from surrounding
Superfund sites. In the first study, the value of a statistical cancer case was approximately $5.5 million,
but did not include any assumptions or information on latency or fatality. The 2002 study calculated

24


-------
499

500

501

502

503

504

505

506

507

508

509

510

511

512

513

514

515

516

517

SAB Review Draft

estimates under a variety of latency and discounting assumptions with results ranging from $5.2 million
to $10.0 with no latency, and from $6.2 to $11.8 million using a 3% discount rate and 10-year latency
period.

Davis (2004) used housing price responses to an observed cancer cluster in Nevada to estimate
marginal willingness to pay for a change in lifetime pediatric leukemia risk ranging from $3.7 million to
$11.1 million, which is generally consistent with the Gayer et al. studies, although the leukemia values are
specific to children. Ho and Hite (2008) included risks from air toxics and hazardous waste sites in a
hedonic property model and estimated the implicit value of cancer mortality to be $6.0 million. Finally,
Lott and Manning (2000) explored the presence of compensating wage differentials for carcinogenic
exposures in the workplace using the hedonic wage framework, finding that workers were being
compensated for carcinogenic exposures. By making assumptions about the proportion of cancer deaths
that arise from occupational exposures they calculated a cancer-specific VSL of $12.4 million.13

Because reducing environmental cancer risk is an important part of EPA's mission to protect
human health, a key question is how the results from the empirical literature summarized here, along
with other literature described in this report, can be systematically synthesized to account for individuals'
preferences for reducing cancer risks relative to other types of health risks. As a first-cut, the simple
average of the central estimates of the cancer differential from the subset of studies in Table 1 that
reported values for both cancer and non-cancer risks is 52%.14 This is a preliminary estimate and should
be refined or replaced with a more systematic synthesis of the literature, possibly incorporating results

13	As stated earlier, all figures have been updated to 2009 dollars using the Consumer Price Index, unless otherwise
noted.

14	Specifically, the summary point estimates that we drew from each of the nine studies in Table 1 that
reported results pertaining directly to the cancer differential (i.e., VSLcancer / VSLnon-cancer -1) are: 0
(Hammit & Hanninger 2010), 0.5 (Alberini & Scansy 2010a), 0.85 (Alberini & Scansy 2010b), -0.15
(Adamowiz et al. 2008), 0 (Cameron & Deshazo 2008), 0.2 (Tsuge et al. 2005), 0.3 (Hammitt & Liu 2004), 3
(Van Houtven et al. 2008), and 0 (Magat et al. 1996). The average of these figures is 0.52.

25


-------
518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

534

535

536

537

538

SAB Review Draft

from other relevant studies. In the meantime, a cancer differential of 50% might be a reasonable
placeholder value for use in upcoming RIAs.15

4 Review of stated preference and hedonic wage studies

Our reviews of the literature in the sections that follow focus on results from stated preference
and hedonic wage and studies. This reflects where the majority of potentially relevant empirical
estimates are found and is consistent with prior consultations and advisory reports. The hedonic wage
approach is well-established and vetted and remains influential in informing guidance across the federal
government. However, the approach is limited to work-related risks and the associated risk
characteristics, many of which differ from EPA policy scenarios, as has been detailed many times in the
economics literature.

There has been a tremendous growth in the number of stated preference studies to estimate
values for mortality risk reductions in recent years; certainly there is now a far larger and more
sophisticated body of literature to draw upon than was available at the time of EPA's last revision of its
guidance. These developments potentially allow for an examination of important valuation dimensions
including risk source (e.g., environmental, traffic-related); type of illness (e.g., any cancer differential or
associated morbidity); and altruism. Our review of the empirical literature and how it can be synthesized
attempts to address these issues.

However, additional studies exist that may supplement the reviews of the stated preference and
hedonic wage literatures below. First, some stated preference studies do not seek to estimate willingness
to pay or accept, but rather relative preferences for different types of mortality risk reduction. Two
examples addressing cancer risks are described more completely above (Magat et al. 1996 and Van

15 Another possible way to represent the cancer differential would be to estimate the absolute (rather than fractional)
increment of the cancer mortality risk values over the values for non-cancer risks (i.e., VSLcancer - VSLnon-cancer).
This would require an additional step of estimating the income elasticity of this absolute cancer differential.
Estimating the fractional cancer differential implicitly assumes that the income elasticity of the absolute cancer
differential equals that for the non-cancer VSL.

26


-------
539

540

541

542

543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

558

559

560

561

562

SAB Review Draft

Houtven et al. 2008). The study results do not estimate willingness to pay, but it may be possible to
combine the estimates from the studies on relative tradeoffs with the willingness to pay literature to
refine our benefit transfers.

Another segment of the literature that we do not examine in detail here includes studies that
evaluate only public preferences for risk reducing policies. Examples from this literature include
Cropper et al. (1994) and Subramanian and Cropper (2000), who used survey methods to examine how
respondents would allocate a given public budget to public programs for lifesaving and risk reduction;
and Bosworth et al. (2009) who assessed community-level preferences for public programs to improve
health and safety. The SAB previously concluded that these studies can be informative in their own right,
but cannot be directly related to individual willingness to pay and used directly for benefit-cost analysis
(USEPA 2001). EPA is open to suggestions on whether and how this literature may be effectively and
appropriately synthesized with the results of other studies for the development of guidance on mortality
risk valuation.

The hedonic property method has been used to estimate the value of environmental amenities
and disamenities including mortality risks. A major challenge has been to limit the analysis to risk
reduction rather than more comprehensive measures or indicators of environmental quality, such as air
quality (e.g., Chay and Greenstone 2005) or the presence of or distance to hazardous waste sites (e.g.,
Greenstone and Gallagher 2008). These studies can be useful for evaluating some policies directly, such
as the remediation of hazardous sites, but cannot be directly informative for mortality risk valuation.
Willingness to pay for reduced mortality risks have been estimated in hedonic property studies, as first
described and demonstrated in Portney (1981), who examined the relationship between housing prices
and mortality risks from air quality. Four other studies, described more completely above in this paper,
estimate marginal willingness to pay for cancer risk (Gayer et al. 2000, 2002; Davis 2004; and Ho and Hite
2008).

27


-------
563

564

565

566

567

568

569

570

571

572

573

574

575

576

577

578

579

580

581

582

SAB Review Draft

Finally, implicit values for risk reductions can be estimated in "averting behavior" studies,
wherein an individual or household uses the good as an input into the production of health or safety.
Blomquist (2004) conducted an extensive review of this literature and concluded, with some caveats, that
the findings are broadly similar to hedonic wage estimates. Recent additions to the literature are
generally consistent with this conclusion (e.g., Andersson 2005, 2008 (automobile risks); Hakes and
Viscusi 2007 (seatbelt use)). Key concerns about averting behavior studies include issues of risk
perception and the separability of joint benefits and costs (USEPA 2000b). Viscusi (1992) explicitly
excluded these studies from consideration in his meta-analysis of VSL estimates. Further, the lack of
available studies on environmentally-related risks limits the usefulness of this class of studies for the
present purpose of developing guidance for mortality risk valuation.16

4.1 Stated preference studies

Stated preference (SP) is a survey-based method for estimating willingness to pay or accept for
non-market goods or services. SP methods are widely used to value environmental amenities or
improvements in human health endpoints that may be difficult or impossible to estimate using revealed
preference methods because of long lag times, unclear causality, or other factors. For example, SP studies
have been used to elicit willingness to pay for reductions in the risks of dying from cancer and cardio-
vascular disease. SP studies vary widely in terms of the types of risk considered, payment vehicles,
latency periods, mode of survey administration, etc. The number of and variation among existing SP
studies is now large enough that the variation in their results can be analyzed statistically, although this
involves a number of data collection and model estimation challenges.

16 Note that there are some studies that relate averting behaviors to environmental quality or even related risks (e.g.,
Dickie and Gerking, 2009; Um, Kwak, and Kim, 2002), but, as documented in Blomquist, 2006, relatively few studies
estimate WTP for reduced mortality risks in an environmental context.

28


-------
583

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

SAB Review Draft

4.1.1 Recent meta-analyses of SP studies

Three recent meta-analyses examined the stated preference literature using statistical methods.
Kochi et al. (2006) used both stated and revealed preference studies in an empirical Bayes framework.
Dekker et al. (2008) focused exclusively on stated preference studies, also with Bayesian methods.
Braathen et al. (2009) conducted a meta-regression analysis of a wide variety of stated preference studies
using classical econometric tools. Each of these studies is discussed in more detail below.

Kochi et al. (20061 used an empirical Bayes estimation method to generate predicted VSL
estimates using multiple estimates from both stated preference and hedonic wage studies. Here we focus
on the analysis and results for the stated preference data in their study. Study selection criteria were
similar to those used by Viscusi (1992), including the use of studies for the general population and those
conducted in high income countries only, and a minimum sample size.17 Another important criterion was
the use of estimates for immediate risk reductions; specifically, estimates for risks involving a latency
period were excluded.

Kochi et al. analyzed 45 VSL estimates drawn from 14 stated preference studies. The authors
recorded all estimates from each study and then separated them into "homogeneous subsets."

Specifically, they grouped estimates by lead study author and used a Q-test for homogeneity to
determine whether the estimates within a group are homogenoeous. After completing the separation of
the estimates into homogenous subsets, they recalculated the VSL for the subset to create a unique VSL
for that author. The recalculated mean reflects a weighted VSL of the estimates in the homogeneous
subset, where the weights are based on the standard errors for the estimates.18 This technique is intended
to address the troubling issue of choosing among multiple estimates from each study when those

17	Viscusi (1992) excluded two studies with sample sizes of around 30. Kochi et al. (2006) chose a minimum sample
size of 100 for their analysis.

18	Another implicit selection criterion in this study was the use of estimates with reported standard errors. In the
assembly of our new meta-analysis dataset, described in Section 4.1.2 below, we find that this may be a highly
constraining selection criterion.

29


-------
604

605

606

607

608

609

610

611

612

613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

SAB Review Draft

estimates may be based on overlapping samples. The process of creating homogeneous subsets resulted
in 18 stated preference VSL estimates with a mean of $3.5 million and a standard error of $0.67 million (in
2009 dollars).

Dekker et al. (20081 examined the influence of risk context (i.e., deaths from automobile-related
accidents, air pollution, and all causes) on willingness to pay estimates from SP studies. The authors
discussed the benefits transfer challenge associated with applying estimates from one context (e.g., auto
risks) to another (e.g., air pollution), particularly when there is limited empirical evidence on the size and
direction of the effects. Employing Bayesian techniques in a meta-regression, they compared willingness
to pay or accept estimates in three different risk contexts—air pollution, traffic safety, and
environment/general—while attempting to control for the size of the risk change and other respondent
and study characteristics. Several study design decisions by Dekker et al. were based on
recommendations from the EPA meta-analysis work group (USEPA 2006).

The authors used existing meta-analyses and additional literature searches to identify stated
preference studies for auto, air pollution, or context-free (unspecified) mortality risk reductions. After
searching the literature and applying screening criteria, a final database was assembled containing 98
VSL estimates from 27 studies, including three studies from the U.S. Seventy-one of the estimates were
based on studies of road safety, seven on studies of air pollution, and twenty on studies of "general
mortality" (presumably deaths from all, or unspecified, causes). The authors drew multiple estimates
from each study, although it appears that they attempted to ensure that those estimates were from non-
overlapping subsamples. Because of the small sample size that results from this approach they use
Bayesian techniques suitable for these situations.

The analysis by Dekker et al. focused on explaining variation in willingness to pay for discrete
changes in mortality risk reductions rather than the VSL and therefore includes as an independent
variable the magnitude of the risk change associated with each estimate. They found that willingness to

30


-------
628

629

630

631

632

633

634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

SAB Review Draft

pay estimates are lower when the commodity is described as a public good and that there is a premium
for risk reductions from air/general context over automobile risks.

Braathen et al. (2009) reviewed and conducted a meta-analysis of 75 studies with 900 estimates
from developed and developing countries. The authors recorded a variety of attributes for each estimate:
type of risk, country, survey mode, type of study, etc. The purpose of the study was to examine how
these attributes influence the resulting VSL estimates. Using classical econometric techniques, their
results show that methodological variables (i.e., type of payment questions, survey mode) explain 70
percent of the variation in the estimates. Of particular relevance to EPA, the authors found that health
risks are valued lower than traffic and environmental risks, in contrast to the results of Dekker et al.
However, risks to individuals are valued higher than risks to the public, similar to the results of Dekker et
al. (2008). The work of Braathen et al. still is preliminary and, like the Dekker et al. meta-analysis, it
includes studies from both developed and developing countries.

4.2.2 A new meta-analysis dataset

In an effort to both update the estimate or range of estimates used by EPA, we have constructed a
new dataset containing information from a set of studies reflecting the current literature appropriate for
application to U.S. environmental policy.19 We used EconLit, conference proceedings, published and
unpublished meta-analyses, working paper series, and personal contacts to identify and generate a
comprehensive list of stated preference mortality risk valuation studies from 1974 and later.20

Each study was screened to ensure that it provided empirical estimates of the value of mortality
risk reductions (i.e., purely theoretical studies and those that only examined morbidity were not
included). Following the advice from the SAB-EEAC (USEPA 2007), we established a set of selection

19	There is substantial overlap between our data set and those reflected in the meta-analyses reviewed in this section.
Differences are due to different selection criteria and new studies that have appeared since the other meta-analysis
studies were conducted.

20	The earliest study that forms the basis of the recommendations of the existing EPA Guidelines (2000a) was
conducted in 1974. Therefore, we limited our search for relevant literature to this starting date, assuming that the
earlier literature had been vetted and judged to be obsolete prior to the release of the 2000 Guidelines.

31


-------
649

650

651

652

653

654

655

656

657

658

659

660

661

662

663

664

665

666

667

668

669

SAB Review Draft-

criteria that determined which studies to include in our final data set. These criteria are based on
information from other meta-analyses, as well as our own best judgment regarding study features
necessary for application to valuing mortality risk reductions when analyzing U.S. environmental
policies. The criteria we applied are as follows:

•	minimum sample size of 100,

•	sample frame based on general population,

•	conducted in a high-income country,21

•	results based on exclusive dataset,

•	written in English,

•	provides enough information to calculate a WTP estimate if one is not reported in the paper,

•	provides estimates for willingness to pay (willingness to accept estimates were not included),22
and

•	provides estimates for willingness to pay for risk reductions to adults (estimates for risk
reductions to children are not included).

We focus on studies with a sample size of at least 100 because smaller samples tend to suffer from
small sample size problems (e.g., less precision) and are less likely to be representative of the general
population. Because the purpose of this exercise is to determine an estimate or range of estimates for use
in environmental policy, we limit our studies to those of the general population as opposed to specialized
subgroups, like students or business owners. In addition, because our focus is on U.S. environmental
policy we choose to limit our studies to those conducted in high-income countries. Socio-economic and
cultural differences between the U.S. and most developing countries may be too large for reliable

21	High-income countries are defined as having a gross national income per-capita of $11,906 (2008 US dollars)
according to the World Bank reports (www.worldbank.org). The most recent World Bank data is for 2008.

22	Three studies report willingness to accept estimates. These studies also report WTP estimates so we do not reject
any study based solely on this criterion.

32


-------
670

671

672

673

674

675

676

677

678

679

680

681

682

683

684

685

686

687

688

SAB Review Draft

transfers of value estimates. Our own language limitations required that we restrict ourselves to studies
written in English. Finally, we limit our investigation to willingness to pay estimates for adults only.

Thirty-three studies published between 1988 and 200923 meet the selection criteria described
above, yielding nearly 450 willingness to pay estimates. For each of the studies we recorded all
willingness to pay and value of statistical life estimates that were reported in the study, as well as those
we could calculate based on information available in the study.24 The meta-analyses using stated
preference studies we described earlier draw multiple estimates from each study, and each has a different
way to address the fact that these estimates are almost always drawn from overlapping samples (e.g.,
authors report multiple results from different estimation exercises or sub-samples within their data).
However, we believe that the issues associated with using multiple estimates from each study are
sufficiently problematic to warrant selection of independent estimates from each study.25 Table 3 reports
selected data for each study with detailed footnotes to describe the decisions to support the selected
estimates.26 This exercise results in 40 independent estimates. We report select characteristics for each
estimates along with the willingness to pay and standard errors (reported in $/|jr). The willingness to
pay for micro-risks are either directly extracted from the underlying studies (when the information was
reported in the papers) or calculated by dividing the VSL estimates by 10 6 when the WTP estimates are
not reported.

All estimates were recorded in the currency and dollar year presented in the study. If the dollar
year was not noted or could not be gleaned from other information in the study then we assumed that it

23	While we set a start date of 1974 for inclusion in our data set, only studies published after 1988 met our selection
criteria.

24	For the most part, all possible estimates were calculated or recorded for each study. We did not, however, record
or calculate estimates for various levels of confidence respondents had in their responses, passing/failing quizzes
about risk, and various forms of scenario rejection. We felt that these estimates were designed mainly to test the
validity of the survey instrument and not to produce central estimates of mortality risk valuations per se.

25	Later we discuss in detail the various issues associated with using multiple estimates and how this can be
addressed econometrically.

26	In general we opted for the estimate(s) that were the most inclusive of all the data in the study. Alternatively, we
could select more estimates from each study - for example, by including estimates by age group - if this was
determined to be an important dimension to the analysis.

33


-------
689

690

691

692

693

694

695

696

697

698

699

700

701

702

703

704

705

706

707

708

709

710

711

712

SAB Review Draft

was the year prior to the release or publication of the paper. All estimates are for individuals; when it
was clear that an estimate reflected a household willingness to pay, we divided those estimates by the
average household size for the country and year when the study was conducted. We then converted all
estimates to U.S. dollars using the Purchasing Power Parity Index for the dollar year of the estimates.
Next, all estimates were converted to 2009 dollars using the Consumer Price Index (CPI) and adjusted for
income growth over time assuming an income elasticity of 0.5.

In addition to the willingness to pay estimates and standard errors (when available), we
quantified and recorded as much information as we could for each study. Our data set includes whether
or not the study was published in a peer-reviewed journal, the year it was conducted and published or
released, the country where the study was conducted, sample characteristics, risk reduction information
(e.g., magnitude, type of risk), scope tests, public versus private risk reductions, etc. See Table 2 for a
description of many of the variables in our data set. Much of this information is only available for a
subset of studies, particularly information on the demographic characteristics of the sample.

Twenty-two studies were published in journals, with 13 published in the Journal of Risk and
Uncertainty. Six of the remaining studies are unpublished reports or working papers and five are book
chapters. We identified nine different sources of mortality risk represented in the studies, including
automobile accidents, air pollution, drinking water, hazardous waste sites, and food. The studies were
predominantly conducted in the U.S. and Europe. Other countries represented in the data include
Canada, Japan, Taiwan, and New Zealand.

Most of the studies are contingent valuation studies where the choice question involves stating a
response (e.g., yes/no to a dichotomous choice question, open-ended response) to a scenario with a fixed
set of attributes. Several studies are choice experiments in which respondents choose one option from
several in which the attributes, including the magnitude of the risk reductions and the cost, vary across
the options.

34


-------
713

714

715

716

717

718

719

720

721

722

723

724

725

726

727

728

729

730

731

732

733

SAB Review Draft

The average sample size for the estimates is 814 observations with a range of 13 to over 2,000.27
Most studies were conducted with a self-administered mode via web-TV or a centralized computer
facility. The second most common mode is an in-person survey. Other modes represented in the data
include mail, telephone, and a combination of the two. A scope test was performed or calculated for
about half of the estimates, and of those about 90 percent passed a weak form of the test (i.e., willingness
to pay estimates exhibited a statistically significant increase with the size of the risk reduction, but was
not necessarily proportional). Fifteen percent passed a strong form of the scope test (i.e., willingness to
pay was proportional or nearly proportional to the size of the risk reduction).

4.2 Hedonic wage studies

In their recommendations to EPA, the SAB-EEAC and the Meta-Analysis workgroup clearly
stated that both revealed hedonic wage and stated preference studies should be considered when
deriving estimates of mortality risk values (USEPA 2006, 2007). Both groups also recommended that the
two segments of the literature be analyzed separately. In this section we focus on the hedonic wage
literature.

Hedonic pricing models use statistical methods to measure the contribution of a good's
characteristics to its price. As applied to the labor market, hedonic wage studies (also known as
compensating wage studies) are based on the premise that heterogeneous goods and services can be
viewed as "bundles" of attributes and are differentiated from each other by the quantity and quality of
these attributes. Fatal and nonfatal risks are among the many attributes that differ across jobs. All else
equal, we would expect riskier jobs to pay higher wages. Therefore, it should be possible to estimate the
value associated with reduced occupational fatality risk using data on wage and risk differentials among

27 This is the sample size for the recorded estimates. Most studies used a subset of the data when recording different
estimates (e.g., males only, younger respondents only). All studies meet the criteria of a minimum sample size of 100
respondents.

35


-------
734

735

736

737

738

739

740

741

742

743

744

745

746

747

748

749

750

751

752

753

754

755

SAB Review Draft-

jobs, controlling for other factors that might influence the wage. A standard regression equation in the
hedonic wage literature is

In w{ = X§ + 
-------
756

757

758

759

760

761

762

763

764

765

766

767

768

769

770

771

772

773

774

775

776

777

778

SAB Review Draft

classifications (assigning all occupations within an industry the same risk) or by broad occupational
classification (ignoring potential differences within an occupation across industries).

A number of recent studies, however, have turned to the Bureau of Labor Statistics' Census of
Fatal Occupational Injuries (CFOI) as the source for workplace risk characteristics. The CFOI data are
considered the most comprehensive data on workplace fatalities available (Viscusi 2004), compiling
detailed information since 1992 from all states and the District of Columbia. Not only are the counts of
these fatal events reported by 3-digit occupation and 4-digit industry classifications, but the
circumstances of the fatal events as well as other characteristics of the workers involved (e.g., age, gender,
race) also are recorded.28 To ensure the veracity and completeness of the reported data, multiple sources
are consulted and cross-referenced, including death certificates, workers' compensation reports and
Federal and State administration reports. To form a complete dataset for estimation, these data still must
be paired with worker samples drawn from another source (often the Current Population Survey) and
fatality rates still must be constructed by the researcher using estimates of the number of workers, as with
the other BLS data.

4.2.2 Estimation issues

Recently, EPA funded a study to examine the hedonic wage methodology and to provide a
quantitative assessment of the robustness of the resulting value estimates for mortality risk reductions.
The results of this research are summarized in Black et al. (2003) and were subsequently published in
Black and Kniesner (2003). These studies examined the roles of the functional form of the estimating
equation, measurement error, and unobservable characteristics using various commonly used data sets.
Their findings highlighted a number of potential problems with previous hedonic wage studies. First,
they found that estimates of the value of risk reductions can be very sensitive to seemingly minor changes
in the specification of the regression equation. In fact, many specifications lead to negative estimates,

28 More information on the CFOI data is available at: http://www.bls.gov/iif/oshfatl.htm.

37


-------
779

780

781

782

783

784

785

786

787

788

789

790

791

792

793

794

795

796

797

798

799

800

801

802

SAB Review Draft:

which would suggest that people would be willing to accept lower wages for jobs with higher risks. They
were unable to alleviate this problem using more flexible functional forms, so they concluded that this
instability is not due to equation mis-specification. Instead, they found strong evidence that the job risk
estimates contain considerable measurement error.

Black and Kniesner (2003) examined both the BLS SWC and NIOSH data sets (the CFOI dataset
had not been widely used by that time). Their results indicate that, while both datasets have advantages
and disadvantages, they both also are subject to considerable measurement error. They identified three
sources of measurement error in the two data sets:

•	sampling variation within industry and occupation cells given the small size of some of the cells
(in recognition of this problem, BLS and NIOSH suppress data when the number of fatalities is
low),

•	heterogeneity in job risks and non-random assignment of those risks within occupations (e.g., late
night convenience store clerks tend to be male and older), and

•	industry and occupation are not measured accurately, especially at the three-digit level.
Moreover, they found that the measurement error is correlated with covariates commonly used in the
wage equations and is likely correlated with the regression error as well. They concluded that studies
that do not control for measurement error suffer from attenuation bias, resulting in under-estimates of
mortality risk values. They also concluded that the NIOSH data produce results most consistent with
economic theory.

4.2.3 Recent meta-analyses ofhedonic wage studies

In addition to the methodological assessment conducted by Black and others, several meta-
analyses of the hedonic wage literature have been conducted in recent years. We focus here on four
recent studies, three of which were reviewed by the Meta-analysis workgroup convened by EPA. The
fourth was published after their deliberations.

38


-------
803

804

805

806

807

808

809

810

811

812

813

814

815

816

817

818

819

820

821

822

823

824

825

826

SAB Review Draft

Mrozek and Taylor (20021 used multiple observations from 47 hedonic wage studies. Variables
included in their meta-regressions were of three types: (1) those which may influence wage/risk tradeoffs
(e.g., mean hourly earnings), (2) those describing the sample, and (3) methodological choices of the
original researchers (e.g., if a risk-squared term was included in the estimating equation).

The authors used weighted least squares where the weights were the number of estimates
provided by the study. This ensured that each study was weighted equally, regardless of the number of
observations drawn from it. Four meta-regression models were estimated, each using log(VSL) as the
dependent variable. All four models indicated a positive and significant relationship between the mean
risk and VSL. The authors used the meta-analysis results to "predict" the VSL as if the original studies
had all followed a set of "best practice" assumptions. The predicted values range from $1.78 million to
$15.4 million (2009 dollars). Those assuming the use of National Institute for Occupational Safety and
Health (NIOSH) data are higher than those assuming use of Bureau of Labor Statistics (BLS) data. The
authors concluded that the evidence best supports an estimate of $2.69 million at the average
occupational risk level of 0.5 per 10,000 (2009 dollars).

While this study provides a comprehensive overview of the hedonic wage literature, it includes
studies using older (and possibly unreliable) occupational risk data. In addition, the authors excluded
estimates in original studies that were statistically insignificant or negative.

Viscusi and Aldv (20031 conducted a review of more than 60 hedonic wage studies of values for
mortality risk reductions across 10 countries (including 52 from the U.S.), examining a number of
econometric issues, the effects of unionization on risk premiums, and the effects of age and income on
VSL estimates. No studies were eliminated from the sample, and no attempt was made to modify the
original VSL estimates. Point estimates extracted from each study were those based on the "whole
sample" and the original authors' preferred model specification. Viscusi and Aldy generated summary
VSL estimates by using the estimated coefficients from the meta-analysis to predict the natural logarithm

39


-------
827

828

829

830

831

832

833

834

835

836

837

838

839

840

841

842

843

844

845

846

847

848

849

SAB Review Draft

of VSL for each original study, then study-specific predicted-VSLs were averaged to produce an overall
mean estimate. Predicted U.S. mean values were constructed based on regression samples using all
countries, but with averaging across U.S. studies only. The predicted values in the study for the U.S.
range from $6.85 million to $9.47 million (2009 dollars), and the median predicted values were generally
very close to the means.

Kochi et al. (2006) used an empirical Bayes estimation method to generate predicted VSL
estimates based on previous hedonic and stated preference studies. Here we focus on the analysis and
results for the hedonic wage data. Using selection criteria similar to those from Viscusi (1992), the
analysis included 162 VSL estimates from 31 hedonic wage studies. All possible VSL estimates and
associated standard errors for each included study were re-estimated based on information provided in
each original study. Estimates without standard errors were not included. The homogeneous subsetting
method described earlier also was applied to the hedonic wage estimates (the hedonic and stated
preference data were analyzed together), resulting in 42 VSL estimates from hedonic wage studies with a
mean of $11.96 million and a standard error of $0.62 million (2009 dollars). Because of the subsetting
technique employed to pool the estimates, Kochi et al. could not explicitly account for study design and
population characteristics in their analysis.

Bellavance et al. (2009) is the most recent meta-analysis of the hedonic wage literature. The
authors' principle objective was to better understand the variability in VSL estimates from hedonic wage
studies, which is described as ranging from $0.5 to $50 million. Thirty-nine VSL estimates from 37
studies were assembled based on those used in prior meta-analyses and further searches of several
economics databases. The resulting dataset contains sixteen studies from the U.S., seven from Canada,
and three or fewer from each of several other countries. The earliest study is from 1974 and the most
recent is Viscusi (2004).

40


-------
850

851

852

853

854

855

856

857

858

859

860

861

862

863

864

865

866

867

868

869

870

871

872

SAB Review Draft

The authors draw only one VSL estimate from each study. Standard errors were recorded or
computed for 32 of the 39 estimates. Criteria were established to chose the specification within each
study, including: (1) no interaction terms between the probability of death and other explanatory
variables (in order to more easily compute the standard error), (2) similarity of specification to other
included studies, (3) larger samples with characteristics most similar to other studies, and (4) the
recommendations of authors of prior meta-analyses. Bellavance et al. acknowledged that the source for
U.S. risk data varies and has evolved over time from early BLS surveys to NIOSH to BLS' Census of Fatal
Occupational Injuries (CFOI). However, their analysis did not control for the data source other than for
the use of Society of Actuaries (SOA) data, which was found to have a significant impact on the estimated
VSL. Sensitivity analyses were conducted with and without studies using SOA risk data.

Using a mixed effects model (random intercept with fixed effects for study characteristics), the
authors regressed the VSL estimates on average income, probability of death, and several study design
variables. The mean weighted average VSL is approximately $7.23 million (2009 dollars). Other key
findings include that the VSL is significantly higher for studies that treat risk as endogenous, and there is
some evidence that the VSL declines with the baseline risk.

4.2.4 A new meta-analysis ofhedonic wage studies

Using Appendix 1 from Bellevance et al. as a starting point, we constructed a new data set of
hedonic wage studies, augmenting the information contained therein with data from Kochi et al. (2006)
and Viscusi and Aldy (2003). We also conducted a full text search in JSTOR for "Census of Fatal
Occupational Injuries" and "CFOI" in order to develop a comprehensive list of studies using these data.
A total of 14 CFOI studies were reviewed, with those actually using the CFOI data in an original, hedonic
wage analysis retained for further assessment. These seven studies were further augmented with an
unpublished manuscript using the CFOI data, for a total of eight additional studies.

41


-------
873

874

875

876

877

878

879

880

881

882

883

884

885

886

887

888

889

890

891

892

893

894

895

896

SAB Review Draft

Additional searches were conducted in JSTOR for studies published in 2000 or later using the key
words "hedonic wage" and "compensating wage." We also conducted a search in the Social Science
Citations Index for studies citing Viscusi (2004), a paper that derives mortality risk valuation estimates
controlling for occupation and industry using the CFOI data.

In constructing our data set, we generally employed the same selection criteria used in Bellavance
et al. (2009), with some exceptions based on our own judgment and to ensure consistency with the criteria
used for the stated preference data set. First, we limited our data to those studies with a sample size of
100 or more. We also retained only those studies conducted in a high-income country as defined by the
World Bank. Third, we omitted studies that rely on Society of Actuaries data as the source of risk
information as these data are thought to reflect broader risks than those experienced on the job (Viscusi
1992, Kochi et al. 2006). We further limited our data by excluding those studies that focus on extremely
dangerous jobs (e.g., police officers), since the risk preferences of individuals who take these jobs may
differ substantially from those of the general public. We do, however, apply the other selection criteria
employed by Bellavance et al., including retaining only those studies using a model specification similar
to that given near the beginning of this section, excluding studies based on specific causes of death,
excluding studies using the same samples as other studies, and excluding studies failing to report enough
information to calculate the value of mortality risk reductions and/or the average probability of death.
Applying all of these criteria resulted in the selection of 37 studies.

For each of our selected studies we recorded the following key variables: year of publication, the
country in which the sample was drawn, sample size, average income, average annual probability of
death, source of risk information, the estimated coefficient on the risk variable, whether the sample was
exclusively male, manufacturing, blue collar and white, as well as whether the regression controlled for
nonfatal risks, union status, and worker compensation. We calculated VMRs for each study by deriving
the VSL and dividing these estimates by 106. As with our stated preference data, all estimates are

42


-------
897

898

899

900

901

902

903

904

905

906

907

908

909

910

911

912

913

914

915

916

917

SAB Review Draft

reported in 2009 dollars after adjusting for inflation using CPI and accounting for income growth over
time assuming an income elasticity of 0.5.

Similar to the stated preference data, we capture only one specification per study in our database,
following the criteria established by Bellavance et al.29 Because the hedonic studies are more
homogeneous in their design than the stated preference studies, we are able to be more selective in which
specifications to include. Although one motivation here is to minimize the influence of each individual
study, it does not necessarily rid us entirely of the problem of overlapping subsamples as many of the
studies draw their samples from the same source.

Table 4 lists key characteristics for our selected studies.30 A total of 24 studies out of 37 were
conducted in the U.S. with 3 using NIOSH data, 13 using BLS data and 8 using CFOI data as the source of
occupational risk. Seven of these twenty-four studies rely on the Panel Study of Income Dynamics (PSID)
as a source for worker characteristics with another 11 using CPS data. Twenty-six studies included
women in their samples and 7 focused on blue collar workers only. Three studies restricted their samples
to union members only. Average sample size across studies was 17,741, and the average income was
$40,508 per year (2009 dollars). The mean probability of occupational death across studies was 0.00014.

5 Income Elasticity Considerations

EPA first attempted to address the income elasticity of VSL issue in its analysis of The Benefits and
Costs of the Clean Air Act, 1990 to 2010 (US EPA, 1999), which made a distinction between application of
income adjustments for longitudinal changes in income over time and cross-sectional income differences
for benefit transfer. The report applied a range of VSL income elasticities in a sensitivity analysis to
project the value of reduced mortality risks in the year 2010.

29	Note that some hedonic studies report results for multiple non-overlapping subsamples (e.g., male vs. female,
union vs. non-union) within the study. Rather than capture these multiple observations, we have elected to
implement the selection criteria used by Bellavance et al.

30	Information reported in the table was adapted from Bellavance et al. (2009).

43


-------
918

919

920

921

922

923

924

925

926

927

928

929

930

931

932

933

934

935

936

937

938

SAB Review Draft

The issue was further developed in EPA's White Paper Valuing the Benefits of Fatal Cancer Risk
Reductions, where income was one of the many benefit transfer issues to be addressed. The SAB-EEAC
review of the White Paper concluded: "With regard to population characteristics, the Committee believes
that it is appropriate to adjust the value of the projected statistical lives saved in future years to reflect
higher incomes in those years, but not for cross-sectional differences in income, because of the sensitivity
of making such distinctions."31 The SAB-EEAC recommended that any appropriate adjustments for
income growth should be part of the Agency's main analysis.

Based on a review of the empirical literature on the cross-sectional income elasticity of VSL
literature originally developed for use in The Benefits and Costs of the Clean Air Act, 1990 to 2010 report,
EPA analyses have typically applied a range of estimates with a low end of 0.08, a central value of 0.4,
and a high end of 1.0. Many analyses characterize this range with a triangular distribution with a
resulting mean estimate of approximately 0.48. Income elasticity is then typically paired with projections
of growth in real US GDP per capita.

More recent information on the income elasticity of VSL has come primarily from meta-analyses
of hedonic wage studies. The results in Mrozek and Taylor (2002) suggest income elasticities ranging
from 0.37 to 0.49, although the authors note that these results should be interpreted with caution because
of measurement error in the income variable and the functional form used by many hedonic wage studies
included in their meta-analysis. As described earlier in this paper, more recent work from Viscusi and
Aldy (2003) estimates the income elasticity of the VSL in the range of 0.5 to 0.6, slightly higher than the
mean value used in many EPA analyses. None of the 95 percent confidence bounds on the Viscusi and
Aldy estimates include a VSL income elasticity as high as 1.0. The Bellevance et al. (2009) meta-analysis,

31 "An SAB Report on EPA's White Paper Valuing the Benefits of Fatal Cancer Risk Reduction," US EPA, 2000, page 7. A
2007 SAB review also noted the empirical difficulties of accounting for differences in real income and wealth across
populations due, in part, to "uncertainty about the value(s) of income elasticity and very little empirical evidence
concerning the relationship between wealth and mortality valuation." US EPA 2007, page D-7.

44


-------
939

940

941

942

943

944

945

946

947

948

949

950

951

952

953

954

955

956

957

958

959

960

SAB Review Draft-

also described earlier, predicts somewhat higher elasticity estimates ranging from 0.84 to 1.08 depending
upon the model.

Some recent theoretical research has examined the relationship between the income elasticity of
the VSL and the coefficient of relative risk aversion and noted that these two quantities should be very
close in magnitude. This can be seen most easily in a simple two-period model. Let "lifetime" utility be
the expected discounted sum of utility in both periods: U = Ux 4- p/3u2, where ut is utility in period t,
p is the probability of survival between periods 1 and 2, and is the utility discount factor. Also assume
that ut depends on income in period t and takes the standard "constant relative risk aversion" (CRRA)
form: ut = C',' " J — 77, where t) is the coefficient of relative risk aversion. The VSL is the marginal rate
of substitution between the individual's first period income and her probability of survival to the second
period, i.e., VSL = / dP~2^U/ dyl j= yl (3 Q2'1 2 C —y and so the income elasticity of the VSL is
CVSL/dyl ~jf. ^'|/VSL j= 77 . Kaplow (2005) examined a more realistic version of this model by allowing
for self-defensive expenditures that could increase the individual's survival probability. Using that
elaborated model, Kaplow showed that the income elasticity of the VSL should be at least as large as 1
when 0 < rj < 1, and at least as large as rj when 77 > 1.

Empirical estimates of the coefficient of relative risk aversion span a wide range—from around
0.5 to 1 at the lower end (e.g., Shepard and Zeckhauser 1984, Eeckhoudt and Hammit 2001, Chetty 2006)
to 10 or more at the high end (e.g., Kocherlakota 1990)—but most estimated or assumed values for rj
seem to fall in the range of 1 to 3. For example, Hall and Jones (2008) and Hall (2010) estimated rj to be
around 2, based on the recent trend of income growth and the more rapid growth in health care
expenditures in the United States. Szpiro (1986), Feldstein and Ranguelova (2001), Barro (2006), and
Layard et al. (2008), among others, also estimate or use values of T] in this range. And in the

45


-------
961

962

963

964

965

966

967

968

969

970

971

972

973

974

975

976

977

978

979

980

981

982

983

984

SAB Review Draft-

contemporary climate change economics literature, the most commonly used values of // are 2 to 3 (e.g.,
Arrow 2007; Nordhaus 2008; Dasgupta 2008; Weitzman 2009, 2010a,b).

The theoretical considerations combined with (most of) the empirical estimates of relative risk
aversion cited above are at odds with the early estimates of the income elasticity of the VSL in the
neighborhood of 0.5 cited above. In a more recent study, Kneisner et al. (2009) applied a quantile
regression approach to a dataset assembled from the Panel Study of Income Dynamics (PSID) and the
Census of Fatal Occupational Injuries (CFOI). Their preferred regression model produced estimates of
the income elasticity of the VSL between 1.23, for the lowest quantile, to 2.24, for the highest quantile.
Kneisner et al. note that "Our estimates of a large income elasticity of VSL are consistent with the simple
theoretical models that have been developed [by Kaplow (2005)]," and "With recent estimates of the
coefficient of relative risk aversion being around 2 based on the labor supply analysis of Chetty (2006)
and the consumption analysis of Kneisner and Ziliak (2002), one would expect the VSL to be income
elastic, which is what the results above indicate."

Based on theoretical considerations such as those examined by Kaplow (2005) and the new
empirical results of Kneisner et al. (2009), EPA believes that its recommended estimate of the income
elasticity of the VSL appears to be on the low end of the range of estimates and may need to be updated
to a higher value or range of values.

6 Methods for Combining Data

The values for mortality risk reductions estimated in the stated preference and hedonic wage
studies described above constitute a current empirical summary of the literature, which can be used to
inform the revision of EPA's mortality risk valuation guidance. These studies could be combined or
synthesized in a number of ways, from a simple point estimate to range, distribution, or systematically
combined in a more rigorous meta-analysis. Our objective in this section is to outline analytical options

46


-------
985

986

987

988

989

990

991

992

993

994

995

996

997

998

999

1000

1001

1002

1003

1004

1005

1006

1007

SAB Review Draft-

that can be implemented in the longer term for updating the estimate or range of estimates used by EPA
in our guidance on valuing morality risk reductions. We begin with meta-analysis methods, including
methods similar to those used in our current guidance and extending to more rigorous application of
meta-regression techniques. This is followed by the structural benefit transfer approach, which involves
calibrating a direct or indirect utility function so that it is consistent with summary estimates of values for
health risk. Our goal is to provide enough information on the analytical options and key issues to receive
clear recommendations from the SAB-EEAC on an approach to implement for updating our guidance and
on future research directions.

6.1 Meta-analysis

There are several options for obtaining simple summary statistics or ranges from the existing
data. We outline these options and key issues in order of increasing complexity.

6.1.1 Parametric distribution

EPA's current guidance took one best estimate from each of five stated preference and twenty-
one hedonic wage studies and then fit a parametric distribution to the values. The resulting mean and
distribution has become EPA's default estimate for valuing mortality risk reductions. To replicate this
approach we could use the databases of SP and HW studies discussed above and then separately
characterize the resulting distributions in a curve-fitting exercise. Based on these distributions we could
define a range of default values for the value of mortality risk for EPA policies. Key choices and
principles are:

• Use all "independent estimates" from the studies rather than one estimate per study. Because many studies
provide estimates for different subpopulations or other treatments, we can often include multiple
study estimates without gross violations of independence. An alternative is to rely upon a single
estimate per study, which has been done for several meta-analysis.

47


-------
1008

1009

1010

1011

1012

1013

1014

1015

1016

1017

1018

1019

1020

1021

1022

1023

1024

1025

1026

1027

1028

1029

1030

1031

SAB Review Draft:

•	Update all study estimates to a common year, including the effect of real income (GDP per capita) growth over
time and the estimates income elasticity of the VSL. The review of the literature in the prior section
already includes this update.

•	Limit SP study estimates to those that are non-cancer and non-latent. In so doing, we will produce a "base
value" that should be more consistent with estimates stemming from the hedonic wage literature.
We will attempt to address any systematic difference in value between reduced cancer risks and
other types of risk separately. In part, this is simply recognizing that EPA policies affect both cancer
and non-cancer mortality risks and different values for each may be appropriate. Similarly, EPA
policies address risk reductions varying from the near-immediate to those delayed over many years,
a benefit-transfer aspect that we address by discounting over estimated latency periods. Including
latent risks in this simple aggregation would double-count the effects of timing on value.

•	Include public-risk studies or rely only on private-risk SP studies. Most EPA regulations result in public
risk reductions. To avoid under-counting benefits, we would want to err toward inclusion, basing
guidance on the full set of relevant studies including those that incorporate altruism even if we
cannot distinguish whether it is paternalistic or non-paternalistic. On the other hand, to avoid
double-counting of benefits we would want to use only those studies that capture private willingness
to pay for mortality risk reduction. Clear recommendations from the EEAC on this issue in particular
would be very helpful.

6.1.2 Classical econometrics

A second approach to combining the information from multiple studies—to determine the
characteristics of the studies that influence the value estimates or to generate a benefit transfer function-
is to perform a meta-regression using classical econometrics. Two issues arise when considering this
approach. First, the analyst must decide which observations to include in the analysis. Some previous
meta-regression studies have used all relevant observations in the analysis (e.g., Nelson and Kennedy

48


-------
1032

1033

1034

1035

1036

1037

1038

1039

1040

1041

1042

1043

1044

1045

1046

1047

1048

1049

1050

1051

1052

1053

1054

1055

SAB Review Draft

2008, Braathen et al. 2009, Mrozek and Taylor 2002). This approach incorporates all available
information, but runs the risk of including estimates from overlapping samples (and therefore non-
independent observations). For example, the same individual(s) may be represented multiple times in
the data when a paper reports multiple estimates using different modeling assumptions. Restricting the
data to non-overlapping samples is a non-trivial exercise because choosing the most appropriate
estimate(s) from each study involves subjective judgment. In addition, small sample size problems —
already a hurdle in meta-analysis—are exacerbated when the sample is limited in this way. The stated
preference and hedonic wage meta-analysis datasets described in Section 4 draw independent samples
based on procedures outlined above. However, a very strict interpretation of the requirement for non-
overlapping subsamples for the hedonic wage studies could result in just a handful of estimates for use in
a meta-analysis given the reliance by authors on the same sources of data.

Second, there are econometric issues to consider when analyzing these data. Nelson and
Kennedy (2008) discuss "factual" versus "methodological heterogeneity." Factual heterogeneity arises
because of real differences in what the primary studies are measuring. For example, the wtp for auto risks
may factually differ from that for cancer risks. Similarly, the wtp for occupational risks for male blue-
collar workers may factually differ from that estimated for a more inclusive sample. Methodological
heterogeneity arises because of different study design choices, such as the use of different models to
estimate willingness to pay. When these sources of heterogeneity are unobserved, errors may be
correlated. It also is likely that estimates produced by different surveys and designed by different
authors have different variances, making heteroskedasticity a concern. Classical econometrics provides
several approaches for dealing with correlated errors and heteroskedasticity. A fixed effects model
assumes that the unobserved heterogeneity among studies can be captured with an intercept shift. By
including a dummy variable for all but one of the studies, the intercept shift is estimated directly. This
approach can result in low degrees of freedom if each study contributes a small number of estimates. An

49


-------
1056

1057

1058

1059

1060

1061

1062

1063

1064

1065

1066

1067

1068

1069

1070

1071

1072

1073

1074

1075

1076

1077

1078

SAB Review Draft-

alternative approach that does not require a new independent variable for each study is the random
effects model. Using the "composite error" exposition of the random effects model, the estimating
equation is

ytj

where yi}- is WTP estimate j from study i, x;/ is the row of data for that estimate, and P is a vector of

coefficients. The error term sj}- has the following structure

Etj =ui + Vy, where u ~ N 0,
-------
1079

1080

1081

1082

1083

1084

1085

1086

1087

1088

1089

1090

1091

1092

1093

1094

1095

1096

1097

1098

1099

1100

1101

SAB Review Draft-

presents a few, select estimates. Mrozek and Taylor (2002) used this approach, as discussed above. This
insures that each study is given equal weight, as opposed to each estimate. The sample size for each
estimate also could be used to generate weights. Observations that arise from larger samples should be
more precise, all else equal. However, sample sizes are not available for all observations in our meta-
analysis datasets. Mrozek and Taylor (2002) used the level of significance of the VSL estimate to create a
f-statistic weight in an appendix to their paper. The estimating equation for this approach is:

1 1

— yV =— XyP
//; nt

where nt is the number of estimates or sample size from the /"' study. This technique provides more

efficient estimates than unweighted estimation of the analogous model.

Considering the data issues common to meta-analyses of willingness to pay estimates for
mortality risk reductions, we propose two classical approaches meant to address both heteroskedasticity
and correlated errors arising from unobserved study heterogeneity when multiple estimates are drawn
from each study. Weighted least squares estimation, as discussed above, can correct for
heteroskedasticity. However, relevant statistics may not be reported to construct the ideal weights. If
weighted least squares is used, we suggest testing for heteroskedasticity and using standard errors that
are robust to clustering. Alternatively, one could estimate a study-level panel model to account for
unobserved heterogeneity and calculate standard errors that are robust to heteroskedasticity. Since many
studies provide just a few estimates, a fixed effects model may not be feasible while a random effects
model would preserve degrees of freedom. We are particularly interested in EEAC comments on these
alternatives.

6.1.3 Bayesian estimation

In the previous section we discussed how classical estimation techniques could be used to
estimate a meta-regression of values for reductions in mortality risks while addressing heteroskedasticity

51


-------
1102

1103

1104

1105

1106

1107

1108

1109

1110

1111

1112

1113

1114

1115

1116

1117

1118

1119

1120

1121

1122

1123

SAB Review Draft

and correlated errors. However, if we use data sets with non-overlapping estimates—as has been
recommended by the Meta-analysis workgroup, and as is reflected in the summary of stated preference
and hedonic wage estimates in Tables 3 and 4—our data selection criteria leave us with relatively small
samples for meta-regression. The combination of small sample size and non-spherical errors presents a
particular problem for classical approaches to estimation. Specification tests, including those for
heteroskedasticity, and calculations of robust standard errors rely on asymptotic relationships and
therefore may not be reliable when the sample size is small (Moeltner and Woodward 2009). Bayesian
estimation has desirable small sample properties and can more easily accommodate general error
structures.

Bayesian analogs to the classical approaches discussed above have been developed and can be
used to estimate a meta-regression model to improve value estimates and provide richer inference into
the results. Koop (2003 p 124-129) presented a Bayesian pooled regression model with an error structure
general enough to be robust to correlated errors and heteroskedasticity even when the form of
heteroskedasticity is unknown. Moeltner and Woodward (2009) use this model to estimate a meta-
regression of wetland valuation estimates from a sample of just 12 values from 9 studies. They use Gibbs
sampling to estimate the model

yj=xB + sj with Cj N 0,cr2fc>; , and IG

fv v

.2 1,

where y^ is WTP reported in study j, X;- is a row vector of population and other characteristics
associated with study j, (3 is a vector of regression coefficients, Ł. is a zero mean regression error with

variance <7 COj, and IG denotes the inverse-gamma distribution. This approach allows the authors to
estimate study-specific variances by estimating a single parameter v and drawing ok in a data
augmentation step. Moeltner and Woodward (2009) showed that Bayesian estimation can be used to

52


-------
1124

1125

1126

1127

1128

1129

1130

1131

1132

1133

1134

1135

1136

1137

1138

1139

1140

1141

1142

1143

1144

1145

1146

1147

SAB Review Draft

conduct meta-regression on small heteroskedastic samples and produce consistent and efficient
parameter estimates.

A Bayesian analogue to the study-level panel model is also developed by Koop (2003 p 149-157).
Bayesian estimation of a study-level panel model with a non-hierarchical prior is analogous to the fixed
effects model in classical econometrics because the unobserved heterogeneity between studies is
attributed to a constant (intercept shift) for each study. If the number of studies is large relative to the
number of estimates from each study then, just as would be the case under classical assumptions, the
high-dimensional parameter space can be problematic. In these cases it may be beneficial to use a
hierarchical prior which places more structure on the unobserved heterogeneity by assuming the study-
level effects can be drawn from a distribution, thus only the parameters of that distribution, and not the
individual effects themselves, need to be estimated. Bayesian estimation of a panel model with a
hierarchical prior is analogous to the classical random effects panel model. In both cases the error
structure imposed on the model is general enough to be robust to non-spherical errors due to correlation
within studies and heteroskedasticity.

6.2 Structural benefit transfer

Thus far we have discussed meta-analysis, including classical and Bayesian approaches to
estimating a meta-regression model, which then could be used for functional benefit transfers. Using
meta-regression, the form of the estimating equation, and therefore the transfer function, typically would
be based on a combination of statistical tests and qualitative theorizing about the important variables to
include in the model. The resulting function can be viewed as a low-order Taylor series approximation to
the "true" preference function within the range of the data used to estimate it.

In contrast to the meta-regression approach, structural benefit-transfer (also known as preference
calibration) involves first specifying a direct or indirect utility function for a representative individual,
then deriving analytical expressions for observable economic outcomes from the utility function (Smith et

53


-------
1148

1149

1150

1151

1152

1153

1154

1155

1156

1157

1158

1159

1160

1161

1162

1163

1164

1165

1166

1167

1168

1169

1170

SAB Review Draft

al. 2002, 2006). Such observable outcomes could include labor-leisure tradeoffs, demand for related
market commodities, equilibrium wage schedules for jobs with differing risk or other characteristics,
responses to stated preference survey questions, etc. The parameters of the utility function are calibrated
using data on such outcomes, and the calibrated model then can be used to predict willingness to pay or
accept for any policy changes that can be described by variations in one or more of the parameters that
appear in the calibrated preference function.

The key advantages of the structural benefit transfer approach are that it provides a means of
combining estimates from separate studies that use different benefit concepts (e.g., marginal or non-
marginal willingness to pay or accept, consumer surplus, compensating or equivalent variation, etc.), and
it assures the economic consistency of transfers (Smith et al. 2002, 2006). In this context "economic
consistency" means, for example, that estimated willingness to pay will never exceed income, that value
estimates will always be responsive to scope (the size of the postulated change in quantity or quality),
that WTP and WTA will always stand in the proper relationship to each other, and so forth. The way that
such consistency is achieved is through the ex ante imposition of a specific form for the utility function,
from which all subsequent value estimates and behavioral responses are then derived. One way to think
about the contrast between meta-regression and structural benefit transfer is that the former uses
relatively more data and fewer theoretical assumptions, while the latter uses relatively fewer data (or
more highly aggregated data) and stronger theoretical assumptions. Therefore, the meta-regression
approach may give more accurate value estimates within the range of the data used to estimate the
function, while the structural benefit transfer approach may be more accurate in out-of-sample transfers.
Thus, the choice of one approach over the other may depend in part on whether the policy case(s) to be
examined fall largely within or largely outside of the range of data available for a meta-regression
transfer function.

54


-------
1171

1172

1173

1174

1175

1176

1177

1178

1179

1180

1181

1182

1183

1184

1185

1186

1187

1188

1189

1190

SAB Review Draft:

6.2.1 Static preference functions

A simplistic example may help clarify the structural benefit transfer approach. Here we follow
Smith et al. (2003, 2006) and use a static model of the tradeoff between income and survival. (In the next
sub-section we will consider a more general dynamic life-cycle model.) Assume that utility conditional
on survival is proportional to the log of scaled income, so expected utility is U — pin aY , where p is the
individual's survival probability. Using this functional form, the marginal willingness to pay for an
increase in the probability of survival is wtp = dll/dp / dU/dY =Yln aY /p. Next suppose that,

based on a comprehensive review of the hedonic wage literature, wtp is estimated to be $8/|jr (i.e., the
VSL is $8,000,000) for individuals with average annual income 35,000 $/yr and average annual survival
probability p = 0.984. This allows calibration of the single unknown parameter of the utility function:
In a = pVSL / Y — In Y = 214.5, which gives a function that can be transferred to individuals with different
background mortality risk levels. This function could vary by age and other personal and environmental
characteristics, and/or different income levels by adjusting p and/or Y, respectively. Using this functional
form, wtp is inversely proportional to the baseline survival probability (and therefore increases with the
background mortality risk) and is (nearly) proportional to income.

We also can use the calibrated utility function to calculate willingness to pay for changes in
mortality risks of any magnitude, rather than relying on the first-order approximation represented by the
wtp. In this case the willingness to pay function is WTP — Y — ex p | pin aY / p + Ap / a. Note that
for large enough Ap's the marginal approximation may exceed total income while the actual WTP
cannot.32 As noted by Smith et al. (2006), this is one of the key advantages of a structural benefit transfer

32 Letting A p go to its maximum value 1 - p gives WTP = Y |jl -1 / a Y 1 p J, which is necessarily less than Y. Also

note that, in this model, the smaller is p the larger is WTP, approaching Y asp goes to zero. This gives a simple
illustration of the "dead-anyway effect" (Pratt and Zeckhauser 1996).

55


-------
1191

1192

1193

1194

1195

1196

1197

1198

1199

1200

1201

1202

1203

1204

1205

1206

1207

1208

1209

1210

1211

1212

SAB Review Draft-

approach: it can produce more realistic predictions of WTP well outside of the range of data used to
estimate marginal willingness to pay. (Additional numerical examples are provided in Appendix A.)

Another advantage of the structural approach is that it can help to account for potential
behavioral responses. We can illustrate this by extending the simple model given above. Again
following the hedonic wage literature, suppose that wages, W, are an increasing function of job-related
mortality risk, m. Specifically, suppose that W = W0 + amp . Total income is comprised of wages plus

non-wage income, y. With this extension, expected utility is U - p0 - m In a y + W0 + am" , where

p0 is the background (non-job related) survival probability. Now suppose that after careful examination
of the hedonic wage literature we estimate that, for a sample of individuals of prime working age (say,
around 40 years old), y = 5,000 $/yr, W = 30,000 $/yr, pQ = 0.99, m = 0.006, and wtp = dW / dm = $8/|jr. So,

for example, if (3 = 0.5, then 8W / dm — pam13'1 => a = 8 / 1CT6 / 2 / 0.006 = 6.67 x 108 and

W0 = W — am1' = -8.6 x 109. (Note that with two estimates of wtp at two levels of job risk, we could
calibrate a and (i simultaneously.) Now recall the standard assumption underlying the hedonic wage
literature that the individual has chosen her job-risk level optimally, and assume she is able to adjust that
level to re-optimize her expected utility after a policy intervention changes p0 by some amount Ap . To
determine the maximum willingness to pay for an exogenous change in mortality risk, we must solve the
two-equation system comprised of (1) the equality between expected utility with and without the policy,
and (2) the first-order condition for maximized expected utility with respect to job-risk with the policy
and a reduction in income equal to WTP.

Results from some simple numerical experiments with this model are given in Appendix A. The
main lesson from these examples is that if individuals are able to adjust their job risk level, then WTP
generally will be higher and the total number of "statistical lives saved" will be lower than otherwise

56


-------
1213

1214

1215

1216

1217

1218

1219

1220

1221

1222

1223

1224

1225

1226

1227

1228

1229

1230

1231

1232

1233

1234

1235

SAB Review Draft

predicted under the assumption of no behavioral response. The numerical examples in Appendix A are
not intended to represent any specific real-world case; nevertheless, they clearly illustrate that the
structural benefit transfer approach is able to capture these effects.

6.2.2 Life-cycle preference functions

The structural benefit transfer function illustrated above was based on the simplifying
assumption that the representative individual looks ahead only one period at a time—that is, utility
depends only on the probability of survival to the next period and expected consumption in the next
period. A more realistic framework would account for expectations of survival and consumption in all
future periods. This brings us to the life-cycle consumption modeling approach. A life-cycle
consumption model represents consumption-versus-saving (and possibly other) choices by an individual
over the course of her lifetime. Life-cycle models are inherently dynamic, with age-specific mortality
probabilities included as key parameters. Individuals are assumed to maximize the expected present
value of discounted utility, where the expectation is conditional on the probabilities of living to all
possible future ages (e.g., Yaari 1964, Shepard and Zeckhauser 1984, Rosen 1988, Cropper and Sussman
1990, Ehrlich 2000, Johansson 2002, Aldy and Smyth 2006, Murphy and Topel 2006, Hall and Jones 2007,
USEPA 2007 p. 14-16).

A life-cycle consumption modeling framework could be used as the basis for a generalized
structural benefit transfer function. Such a transfer function would allow calculation of willingness to
pay for any marginal or non-marginal changes in the individual's mortality profile (i.e., "survival curve")
at any point in the life cycle. As emphasized by Hammit (2007 p. 232), "the survival curve and how it
shifts are the fundamental concepts; the number of life-years saved and lives saved in a specified time
period are the alternative and partial summary measures of the shift." The life-cycle consumption
framework is tailor-made to account for shifts in the survival curve, and it can easily account for the age

57


-------
1236

1237

1238

1239

1240

1241

1242

1243

1244

1245

1246

1247

1248

1249

1250

1251

1252

1253

1254

1255

1256

1257

1258

1259

SAB Review Draft

and lifetime income profile of the individual and the latency and cessation lag characteristics of the
policy.

As in any structural benefit transfer application, it may be necessary to calibrate the parameters
of a life-cycle consumption model using only a few aggregate data—for example, summary statistics on
labor-leisure tradeoffs, average rates of saving over a representative individual's life span, average
market wage differentials for more versus less risky jobs, summary results from stated preference surveys
on risk tradeoffs, etc. Thus, like other structural-benefit transfer functions, one based on the life-cycle
consumption framework would necessarily sacrifice statistical sophistication for theoretical consistency,
so many of the advantages and disadvantages of structural benefit-transfer functions discussed by Smith
et al. (2002, 2006) will apply to life-cycle models as well.

An important potential advantage of using a life-cycle consumption framework for structural
benefit transfers is that it could help to avoid the transfer errors that may arise from using a single VSL
point estimate for all varieties of mortality risk reductions. As shown in Appendix A, the life-cycle
framework allows calculation of the marginal willingness to pay at any age a for risk reductions at any
later age b, wtpa h. VSL estimates from hedonic wage studies may be most plausibly interpreted as the

marginal willingness to pay for contemporaneous mortality risk reductions for adults of prime working
age, e.g., wtp40 40. It may be inaccurate to use such estimates to calculate the willingness to pay for, say, a
20 year-old who will experience mortality risk reductions at ages 55 through the end of life. In contrast, a
schedule of wtya h estimates based on a calibrated life-cycle consumption model would give a ready

means of calculating total willingness to pay for any exogenous shift in the survival curve for individuals
of any age. Furthermore, this approach can properly account for all latency and cessation lag effects
associated with the specific pattern of mortality risk changes caused by the policy, without the need for
possibly inaccurate transfers of a VSL point estimate to earlier and later ages and across individuals with
different levels of wealth and income.

58


-------
1260

1261

1262

1263

1264

1265

1266

1267

1268

1269

1270

1271

1272

1273

1274

1275

1276

1277

1278

1279

1280

1281

1282

SAB Review Draft

Implementing such a structural life-cycle benefit transfer function would be challenging.
Estimating or calibrating such a model would require specifying or solving for the life-cycle pattern of
consumption, and specifying a functional form for the utility function as well as calibrating or estimating
its parameters. Any structural benefit transfer approach—whether based on a life-cycle consumption
framework or something else—would represent a significant departure from the traditional point
estimate transfer approach typically used for mortality risk valuations, mainly based on the VSL. To
accelerate the development of such an approach, we recommend conducting additional case studies
applying existing structural benefit transfer functions (e.g., Smith et al. 2002, 2003, 2006) to a wider range
of illustrative policy scenarios, and additional research aimed at expanding and refining the calibration of
existing benefit transfer functions or developing new ones for potential use in future policy analyses. The
scholarly research on structural benefit transfer methods is still in an early stage, so we are especially
interested in EEAC recommendations in this area.

7 Conclusions

EPA continually strives to improve the quality of its economic analyses of proposed
environmental policies. This is especially important in the area of human health valuation, in particular
the value of mortality risk reductions, since such a large fraction of the (monetized) benefits of EPA rules
are based on this category of impacts. This white paper represents the latest round of literature review
and study by EPA's National Center for Environmental Economics on this topic, submitted to the SAB-
EEAC for feedback. Advice from the committee will be carefully considered as EPA updates its
Guidelines for Preparing Economic Analyses.

7.1 Addressing key issues: terminology, altruism, cancer valuation

EPA plans to change its metric and terminology for mortality risk valuation in benefit-cost
analysis to better reflect the risk-dollar tradeoffs faced by individuals as evaluated in the economics

59


-------
1283

1284

1285

1286

1287

1288

1289

1290

1291

1292

1293

1294

1295

1296

1297

1298

1299

1300

1301

1302

1303

1304

SAB Review Draft

literature, and risk reductions provided by environmental policies. As detailed in section 3.1.2 of this
white paper, for valuation purposes we will report changes in risk reductions valued in terms of the
value of mortality risk (VMR), scaled to micro-risk reductions. This is consistent with recent suggestions
in the economics literature and is aimed at reducing confusion about how mortality risks are evaluated in
benefit-cost analysis.

A second key issue for EPA is the valuation of cancer risk reductions and how these risks are
valued systematically differently from the more immediate risks typically considered in WTP studies.
Our review of the cancer literature, while not conclusive, suggests a "cancer differential" of roughly 50%
over immediate accidental or "generic" risk valuation estimates. We recommend including a differential
of this general magnitude as part of Agency benefits analyses for reduced cancer risks. Specific guidance
on the application of this differential will be developed by the Agency at a later date.

7.2 Longer term analytical directions

In the longer term, EPA plans to perform analysis to better and more rigorously synthesize the
existing mortality risk valuation literature. Two key directions include meta-analysis and structural
benefit-transfer.

7.2.2 Meta-analysis

Section 5.1 described simplified approaches to aggregating the existing empirical valuation data,
along with some key issues to consider in this process. These include whether to (i) use multiple
estimates from studies, (ii) update all studies to a common year accounting for real income growth, and
(iii) limit SP studies to avoid double-counting the effects of cancer risks and latency or cessation lag. The
suggested approach evaluates the RP and SP studies separately, from which EPA would develop a range
of default values for the value of mortality risk (VMR).

60


-------
1305

1306

1307

1308

1309

1310

1311

1312

1313

1314

1315

1316

1317

1318

1319

1320

1321

1322

1323

1324

1325

1326

1327

SAB Review Draft

Alternatively, a new addition to the discussion of mortality risk meta-analysis with the SAB-
EEAC is the potential for Bayesian meta-regression, and we are particularly interested in the SAB-EEAC
comments on the potential advantages and disadvantages of this approach. Another key question to
consider is how the results of any meta-regression would be used to inform guidance, and the merits of
developing a statistical benefit-transfer function from these results.

7.2.2 Structural Benefit Transfer

An alternative to meta-regression and other largely statistical approaches to synthesizing
literature results for policy, is to impose more structure on the benefit-transfer problem and then calibrate
a preference function based on a specified utility function and data on observable outcomes. This is a
relatively new approach that has been developed and demonstrated in only a few previous studies. We
recommended conducting additional scoping studies and further research to develop structural benefit
transfer functions, possibly based on a life-cycle consumption framework, suitable for application in
benefit-cost analyses of future EPA policies.

7.3 Other research directions

We see three other areas where more research would be valuable in developing guidance for
mortality risk valuation, and we welcome SAB-EEAC comment on these (as requested in the
accompanying charge questions).

First, additional applied research on the altruistic components of WTP for public risk reductions
would be a valuable contribution, potentially allowing EPA to rigorously include theoretically-
appropriate altruistic values and better reflect the public value of environmental policies. We
acknowledge that this is a difficult task. The economics literature on the proper treatment of altruism in
benefit-cost analysis is well-developed, enumerating the conditions under which it is appropriate to
include altruistic values in evaluating the benefits of public programs. EPA programs are inherently

61


-------
1328

1329

1330

1331

1332

1333

1334

1335

1336

1337

1338

1339

1340

1341

SAB Review Draft

public and ideally should include paternalistic altruism. However, while the empirical literature has
been able to capture some altruistic values for public risk reductions, it has not generally been able to
distinguish among types of altruism sufficiently well for the values to be included neatly in applied
analysis.

Second, more and more research reflects the general understanding that value of reducing
mortality risks is not "one-size-fits-all." Rather, these values are heterogeneous, or "individuated," and
depend upon a wide array of individual and risk characteristics. More detailed research in this area also
will provide data needed for developing more general and more accurate benefit-transfer functions.

Third and finally, most of the valuation literature, and many theoretical frameworks, have
treated mortality and morbidity risks separately, focusing on just one of these endpoints at a time.
However, some recent work also suggests that changes in health risks may be best framed as changes in
health risk profiles that include both mortality and morbidity. Individuals may value different
combinations of changes in risk or illness and risk of death in complex ways. Systematic empirical work
to evaluate these relationships could lead to much more robust and complete benefits analysis.

62


-------
1342

1343

1344

1345

1346

1347

1348

1349

1350

1351

1352

1353

1354

1355

1356

1357

1358

1359

1360

1361

1362

1363

1364

1365

1366

1367

1368

1369

1370

1371

1372

1373

1374

1375

SAB Review Draft

References

Adamowicz W, Dupont D, Krupnick A, Zhang J. 2008. Valuation of cancer and microbial disease risk
reductions in municipal drinking water: an analysis of risk context using multiple valuation methods.
Working Paper.

Alberini A, Scasny M. 2010a. Labels and perceptions in mortality risk reduction valuations. Working
Paper.

Alberini A, Scasny M. 2010b. Context and the VSL: Evidence from a Stated Preference Study in Italy and
the Czech Republic. Working Paper.

Alberini A, Scasny, M, Guignet D, Tonin S. 2010. The benefits of contaminated site cleanup revisted: the
case of Naples and Caserta, Italy. Working Paper.

Alberini A, Chiabai A. 2007. Urban environmental health and sensitive populations: how much are the
Italians willing to pay to reduce their risks? Regional Science and Urban Economics 37: 239-258.

Alberini A, Cropper M, Krupnick A, Simon NB. 2004. Does the value of a statistical life vary with age and
health status? Evidence from the US and Canada. Journal of Environmental Economics and Management
48(1): 769-792.

Alberini A, Cropper M, Krupnick A, Simon NB. 2006a. Willingness to pay for mortality risk reductions:
does latency matter? Journal of Risk and Uncertainty 32: 231-245.

Alberini A, Hunt A, Markandya A. 2006b. Willingness to pay to reduce mortality risks: evidence from a
three-country contingent valuation study. Environmental and Resource Economics 33:251-264.

Alberini A, Scasny M, Kohlova MB, Melichar J. 2006c. The value of a statistical life in the Czech Republic:
evidence from a contingent valuation study. In Menne B, Ebi K, eds. Climate Change Adaptation
Strategies for Europe. Darmstadt, Germany: Springer.

Alberini A, Tonin S, Turvani M, Chiabai A. 2007. Paying for permanence: public preferences for
contaminated site cleanup. Journal of Risk and Uncertainty 34(2):155-178.

Aldy JE, Smyth SJ. 2006. A numerical analysis of the value of life. Working paper,
http://cbey.research.yale.edu/uploads/File/aldy_35.pdf

Aldy JE, Viscusi WK. 2007. Age differences in the value of statistical life: revealed preference evidence.
Resources for the Future Discussion Paper, http://www.rff.org/documents/RFF-DP-07-05.pdf

Aldy JE, Viscusi WK. 2008. Adjusting the value of a statistical life for age and cohort effects. The Review of
Economics and Statistics 90(3):573-581.

Andersson H, Lindberg G. 2009. Benevolence and the value of road Safety. Accident Analysis and
Prevention 41:286-293.

Andersson H. 2005. The value of safety as revealed in the Swedish car market: an application of the
hedonic pricing approach. Journal of Risk and Uncertainty 30(3):211-239.

63


-------
1376

1377

1378

1379

1380

1381

1382

1383

1384

1385

1386

1387

1388

1389

1390

1391

1392

1393

1394

1395

1396

1397

1398

1399

1400

1401

1402

1403

1404

1405

1406

SAB Review Draft

Andersson H. 2008. Willingness to pay for car safety: evidence from Sweden. Environmental and Resource
Economics 42(4):579-594.

Arabsheibani RG, Marin A. 2000. Stability of estimates of the compensation for danger. Journal of Risk and
Uncertainty 20(3):247-269.

Arrow KJ. 2007. Global climate change: a challenge to policy. The Economist's Voice June 2007 p 1-5.

Ashenfelter O, Greenstone M. 2004. Estimating the value of a statistical life: the importance of omitted
variables and publication bias. The American Economic Review 94(2):454-460.

Ashenfelter O. 2006. Measuring the value of a statistical life: problems and prospects. The Economic Journal
116(510):C10-C23.

Barro RJ. 2006. Rare disasters and asset markets in the twentieth century. The Quarterly Journal of
Economics 121(3):823-866.

Bellavance F, Dionne G, Lebeau M. 2009. The value of statistical life: a meta-analysis with a mixed effects
regression model. Journal of Health Economics 28:444-464.

Berger MC, Gabriel PE. 1991. Risk aversion and the earnings of U.S. immigrants and natives. Applied
Economics 23: 311-318.

Bergstrom TC. 1982. When is a man's life worth more than his human capital? In: Jones-Lee MW, ed. The

Value of Life and Safety: Proceedings from a Conference held by the Geneva Association. Amsterdam: North
Holland Press, p 3-25.

Bergstrom TC. 2006. Benefit-cost in a benevolent society. American Economic Review 96(1):339-351.

Bergstrom JC, Taylor LO. 2006. Using meta-analysis for benefits transfer: theory and practice. Ecological
Economics 60:351-360.

Black D, Gal do J, Liu L. 2003. How Robust are Hedonic Wage Estimates of the Price of Risk? Final Report to the
USEPAR 829-43-001.

Black DA, Kniesner TJ. 2003. On the measurement of job risk in hedonic wage models. Journal of Risk and
Uncertainty 27(3)205-220.

Blomquist GC. 2004. Self-protection and averting behavior, values of statistical lives, and benefit-cost
analysis of environmental policy. Review of Economics and the Household 2(1):89-110.

Bornstein, Seth. "American Life Worth Less Today." Associated Press 10 July 2008.

Bosworth R, Cameron TA, DeShazo JR. 2009. Demand for environmental policies to improve health:
evaluating community-level policy scenarios. Journal of Environmental Economics and Management
57(3):293-308.

64


-------
SAB Review Draft

1407	Braathen NA, Lindhjem H, Navrud S. 2009. Valuing lives saved from environmental, transport and

1408	health policies: a meta-analysis of stated preference studies.

1409	http://www.oecd.org/dataoecd/20/48/43809818.pdf

1410	Buzby JC, Ready RC, Skees JR. 1995. Contingent valuation in food policy analysis: a case study of a

1411	pesticide residue risk reduction program. Journal of Agriculture and Applied Economics 27(2):613-625.

1412	Cameron TA. 2009. The value of a statistical life: [they] do not think it means what [we] think it means.

1413	University of Oregon Working Paper, http://www.uoregon.edu/~cameron/vita/

1414	REEP_VSL_102509.pdf

1415	Cameron TA, DeShazo JR 2008. Demand for health risk reductions. Working Paper,

1416	http://www.uoregon.edu/~cameron/vita/Manuscript_20051269R2.pdf

1417	Cameron TA, DeShazo JR, Johnson EH. 2008. Willingness to pay for health risk reductions: differences by

1418	type of illness. Working paper, http://www.aere.org/meetings/documents/ 0608Erica_Johnson.pdf

1419	Carson RT, Mitchell RC. 2006. Public preferences toward environmental risks: the case of

1420	trihalomethanes. In Alberini A, Bjornstad D, Kahn JR, eds., Handbook of Contingent Valuation.

1421	Northampton, MA: Edward Elgar.

1422	Chay KY, Greenstone M. 2005. Does air quality matter? Evidence from the housing market. Journal of

1423	Political Economy 113(2):376-424.

1424	Chetty R. 2006. A new method of estimating risk aversion. The American Economic Review 96(5):1821-1834.

1425	Chilton S, Jones-Lee M, Kiraly F, Metcalf H, Pang W. 2006. Dread risks. Journal of Risk and Uncertainty

1426	33:165-182.

1427	Corso PS, Hammitt JK, Graham JD. 2001. Valuing mortality-risk reduction: using visual aids to improve

1428	the validity of contingent valuation. Journal of Risk and Uncertainty 23(2):165-184.

1429	Cropper ML, Sussman FG. 1990. Valuing future risks to life. Journal of Environmental Economics and

1430	Management 19:160-174.

1431	Cropper M, Aydede SK, Portney PR. 1994. Preferences for lifesaving programs - how the public

1432	discounts time and age. Journal of Risk and Uncertainty 8(3):243-265.

1433	Cousineau J-M, Lacroix R, Girard A-M. 1992. Occupational hazard and wage compensating differentials.

1434	Review of Economics and Statistics 74(1):166-169.

1435	Dasgupta P. 2008. Discounting climate change. Journal of Risk and Uncertainty 37:141-169.

1436	Davis LW. 2004. The effect of health risk on housing values: evidence from a cancer cluster. The American

1437	Economic Review 94(5):1693-1704.

1438	Dekker T, Brouwer R, Hofkes M, Moeltner K. 2008. The effect of risk context on the value of a statistical

1439	life: a Bayesian meta-model. IVM Working Paper.

65


-------
SAB Review Draft

1440	Desaigues B, Rabl A. 1995. Reference values for human life: an econometric analysis of a contingent

1441	valuation in France. In Schwabe Christie NG, Soguel NC, eds., Contingent Valuation, Transport Safety

1442	and the Value of Life. Boston, MA: Kluwer Academic Publishers.

1443	Dickie M, Gerking S. 2009. Family behavior: implications for health benefits transfer from adults to

1444	children. Environmental and Resource Economics 43:31-43.

1445	Dillingham AE. 1985. The influence of risk variable definition on value of life estimates. Economic Inquiry

1446	24:277-294.

1447	Dillingham AE, Smith RS. 1984. Union effects on the valuation of fatal risk. In Proceedings of the Industrial

1448	Relations Research Association 36th annual meeting, San Francisco, CA.

1449	Dorsey S, Norman W. 1983. Workers' compensation, job hazards and wages. Industrial and Labor Relations

1450	Review 36(4): 642-654.

1451	Ehrlich I. 2000. Uncertain lifetime, life protection, and the value of life saving. Journal of Health Economics

1452	19:341-367.

1453	European Commission. 2000. Recommended interim values for the value of preventing a fatality in DG

1454	environment cost benefit analysis, http://ec.europa.eu/environment/enveco/others/pdf/

1455	recommended_interim_values.pdf

1456	Evans MF, Smith VK. 2006. Do we really understand the age-VSL relationship? Resource and Energy

1457	Economics 28(3):242-261.

1458	Evans MF, Smith VK. 2008. Complementarity and the measurement of individual risk tradeoffs:

1459	accounting for quantity and quality of life effects. NBER working paper 13722.

1460	http://www.nber.org/papers/wl3722

1461	Federal Register. 1993. Executive Order 12866: Regulatory Planning and Review. Federal Register

1462	58(190):51735-51744.

1463	Feldstein M, Ranguelova E. 2001. Individual risk in an investment-based social security system. The

1464	American Economic Review 91(4):1116-1125.

1465	Flores NE. 2002. Non-paternalistic altruism and welfare economics. Journal of Public Economics 83(2):293-

1466	305.

1467	Garen J. 1988. Compensating wage differentials and the endogeneity of job riskiness. The Review of

1468	Economics and Statistics 70(1):9-16.

1469	Gayer T, Hamilton JT, Viscusi WK. 2000. Private values of risk tradeoffs at Superfund sites: housing

1470	market evidence on learning about risk. Review of Economics and Statistics 82:439-451.

1471	Gayer T, Hamilton JT, Viscusi WK. 2002. The market value of reducing cancer risk: hedonic housing

1472	prices with changing information. Southern Economic Journal 69:266-289.

66


-------
SAB Review Draft

1473	Gegax D, Gerking S, Schulze W. 1991. Perceived risk and the marginal value of safety. Review of Economics

1474	and Statistics 73(4):589-596.

1475	Gerking S, De Haan M, Schulze W. 1988. The marginal value of job safety: a contingent valuation study.

1476	Journal of Risk and Uncertainty 1(2):185-199.

1477	Greene WH. 1997. Econometric Analysis, Third Edition. Upper Saddle River, NJ: Prentice-Hall.

1478	Greenstone M, Gallagher J. 2008. Does hazardous waste matter? Evidence from the housing market and

1479	the Superfund program. Quarterly Journal of Economics 123(3):951-1003.

1480	Gunderson M, Hyatt D. 2001. Workplace risks and wages: Canadian evidence from alternative models.

1481	The Canadian Journal of Economics 34(2):377-395.

1482	Gyrd-Hansen D, Halvorsen PA, Kristiansen IS. 2007. Willingness-to-pay for a statistical life in the times of

1483	a pandemic. Health Economics 17(l):55-66.

1484	Hakes JK, Viscusi KW. 2007. Automobile seatbelt usage and the value of statistical life. Southern Economics

1485	Journal 73(3):659-676.

1486	Hall RE, Jones CI. 2007. The value of life and the rise in health spending. The Quarterly Journal of Economics

1487	122(l):39-72.

1488	Hammitt JK. 2007. Valuing changes in mortality risk: lives saved versus life years saved. Review of

1489	Environmental Economics and Policy l(2):228-240.

1490	Hammitt JK, Graham JD. 1999. Willingness to pay for health protection: inadequate sensitivity to

1491	probability. Journal of Risk and Uncertainty 8:33-62.

1492	Hammitt JK, Haninger K. 2007. Willingness to pay for food safety: sensitivity to duration and severity of

1493	illness. American Journal of Agricultural Economics 89(5):1170-1175.

1494	Hammitt JK, Haninger K. 2010. Valuing fatal risks to children and adults: effects of disease, latency, and

1495	risk aversion. Journal of Risk and Uncertainty 40:57-83.

1496	Hammitt JK and Liu J-T. 2004. Effects of disease type and latency on the value of mortality risk. Journal of

1497	Risk and Uncertainty 28:73-95.

1498	HM Treasury. 2003. The green book: Appraisal and evaluation in Central Government, http://www.hm-

1499	treasury ,gov.uk/d/green_book_complete.pdf

1500	Ho C-S, Hite D. 2008. The benefit of environmental improvement in the Southeastern United States:

1501	evidence from a simultaneous model of cancer mortality, toxic chemical releases and house values.

1502	Papers in Regional Science 87(4):589-604.

1503	Howard RA. 1989. Microrisks for medical decision analysis. International Journal of Technology Assessment

1504	in Health Care 5:357-370.

67


-------
SAB Review Draft

1505	Hultkrantz L, Lindberg G, Andersson C. 2006. The value of improved road safety. Journal of Risk and

1506	Uncertainty 32:151-170.

1507	Itaoka K, Krupnick A, Akai M, Alberini A, Cropper M, Simon N. 2007. Age, health and willingness to pay

1508	for mortality risk reductions: a contingent valuation survey of Shizuoka, Japan residents.

1509	Environmental Economics and Policy Studies 8:211-237.

1510	Johannesson M, Johansson P-O, O'Connor RM. 1996. The value of private safety versus the value of

1511	public safety. Journal of Risk and Uncertainty 13:263-175.

1512	Johannesson M, Johansson P-O, Lofgren K-G. 1997. On the value of changes in life expectancy: blips

1513	versus parametric changes. Journal of Risk and Uncertainty 15:221-239.

1514	Johansson P-O. 2002. On the definition and age-dependency of the value of a statistical life. The Journal of

1515	Risk and Uncertainty 25(3):251-263.

1516	Kaplow L. 2005. The value of a statistical life and the coefficient of relative risk aversion. The Journal of

1517	Risk and Uncertainty 31(l):23-34.

1518	Kidholm K. 1995. Assessing the value of traffic safety using the contingent valuation technique: the

1519	Danish survey. In Schwabe NG, Soguel NC, eds. Contingent Valuation, Transport Safety and the Value of

1520	Life: Studies in Risk and Uncertainty. Boston, MA: Kluwer.

1521	Kim S-W, Fishback PV. 1999. The impact of institutional change on compensating wage differentials for

1522	accident risk: South Korea, 1984-1990. Journal of Risk and Uncertainty 18:231-248.

1523	Kniesner TJ, Leeth JD. 1991. Compensating wage differentials for fatal injury risk in Australia, Japan, and

1524	the United States. Journal of Risk and Uncertainty 4(l):75-90.

1525	Kniesner, TJ, Viscusi WK, Ziliak JP. 2006. Life-cycle consumption and the age-adjusted value of life.

1526	Contributions to Economic Analysis and Policy 5(l):Article 4.

1527	http://www.bepress.com/bejeap/contributions/vol5/issl/art4.

1528	Kocherlakota NR. 1996. The equity premium: it's still a puzzle. Journal of Economic Literature 34:42-71.

1529	Kochi I, Hubbell B, Kramer R. 2006. An empirical Bayes approach to combining and comparing estimates

1530	of the value of a statistical life for environmental policy analysis. Environmental & Resource Economics

1531	34:385-406.

1532	Koop, G. 2003. Bayesian Econometrics. John Wiley and Sons Ltd. West Sussex England.

1533	Lanoie P, Pedero C, Latour R. 1995. The value of a statistical life: a comparison of two approaches. Journal

1534	of Risk and Uncertainty 10:235-257.

1535	Layard R, Nickell S, Mayraz G. 2008. The marginal utility of income. Journal of Public Economics 92:1846-

1536	1857.

1537	Leeth, J. D. and J. Ruser. 2003. Compensating wage differentials for fatal and nonfatal injury risk by

1538	gender and race. Journal of Risk and Uncertainty 27(3): 257-77.

68


-------
SAB Review Draft

1539	Leigh JP. 1995. Compensating wages, value of a statistical life, and inter-industry differentials. Journal of

1540	Environmental Economics and Management 28(l):83-97.

1541	Leigh JP. 1991. No evidence of compensating wages for occupational fatalities. Industrial Relations

1542	30(3):382-395.

1543	Leigh PJ, Folsom RN. 1984. Estimates of the value of accident avoidance at the job depend on concavity of

1544	the equalizing differences curve. The Quarterly Review of Economics and Business 24(l):55-56.

1545	Lott, John R., Jr., and Richard L. Manning. 2000. Have changing liability rules compensated workers twice

1546	for occupational hazards? The Journal of Legal Studies 29(1): 99-130.

1547	Magat WA, Viscusi WK, Huber J. 1996. A reference lottery metric for valuing health, Management Science

1548	42(8):1118-11130.

1549	Marin A, Psacharopoulos G. 1982. The reward for risk in the labor market: evidence from the United

1550	Kingdom and reconciliation with other studies. Journal of Political Economy 90(4):827-853.

1551	Martinello F, Meng R. 1992. Workplace risks and the value of hazard avoidance. Canadian Journal of

1552	Economics 25(2):333-345.

1553	McConnell KE. 1997. Does altruism undermine existence value? Journal of Environmental Economics and

1554	Management 32:22-37.

1555	Meng R. 1989. Compensating differences in the Canadian labour market. Canadian Journal of Economics

1556	12(2):413-424.

1557	Meng R, Smith D. 1990. The valuation of risk of death in public sector decision-making. Canadian Public

1558	Policy 16(2):137-144.

1559	Meng RA, Smith DA. 1999. The impact of workers' compensation on wage premiums for job hazards.

1560	Applied Economics 31:1101-1108.

1561	Miller P, Mulvey C, Norris K. 1997. Compensating differentials for risk of death in Australia. Economic

1562	Record 73(223):363-372.

1563	Miller T, Guria J. 1991. The value of statistical life in New Zealand. Report to the New Zealand Ministry of

1564	Transport, Land Transportation Division, #0-477-05255-X.

1565	Moeltner K, Woodward R. 2009. Meta-functional benefit transfer for wetland valuation: making the most

1566	of small samples. Environmental & Resource Economics 42:89-108.

1567	Moore CC, Phaneuf DJ, Thurman WN. 2009. A Bayesian bioeconometric model of invasive species

1568	control: the case of the Hemlock Wooly Adelgid. Working paper, http://www.

1569	ncsu.edu/cenrep/research/documents/Moore_Phaneuf_Thurman_invasivepaper-Aug09.pdf

1570	Moore MJ, Viscusi KW. 1988. Doubling the estimated value of life: results using new occupational fatality

1571	data. Journal of Policy Analysis and Management 7(3):476-490.

69


-------
SAB Review Draft

1572	Morris J, Hammitt JK. 2001. Using life expectancy to communicate benefits of health care programs in

1573	contingent valuation. Medical Decision Making 21:468-478.

1574	Morzek J, Taylor LO. 2002. What determines the value of life? A meta-analysis. Journal of Policy Analysis

1575	and Management 21(2):253-270.

1576	Murphy KM, Topel RH. 2006. The value of health and longevity. Journal of Political Economy 114(5):871-

1577	904.

1578	Nelson JP, Kennedy PE. 2008. The use (and abuse) of meta-analysis in environmental and natural

1579	resource economics: an assessment, http://papers.ssrn.com/sol3/papers. cfm?abstract_id=1117490

1580	Nordhaus W. 2008. A Question of Balance: Weighing the Options on Global Warming Policies. New Haven, CT:

1581	Yale University Press.

1582	Olson CA. 1981. An analysis of wage differentials received by workers on dangerous jobs. Journal of

1583	Human Resources 16(2):167-185.

1584	OMB (Office of Management and Budget). 2003. Circular A-4: Rgulatory Analysis. Executive Office of the

1585	President, Washington, DC. http://www.whitehouse.gov/omb/circulars

1586	Persson U, Norinder A, Hjalte K, Gralen K. 2001. The value of a statistical life in transport: findings from

1587	a new contingent valuation study in Sweden. Journal of Risk and Uncertainty 23(2):121-134.

1588	Philips PR, Russell IT, Jones-Lee MW. 1989. The empirical estimation of individual valuation of safety:

1589	results of a national sample survey. In Jones-Lee MW, ed. The Economics of Safety and Physical Risk.

1590	Oxford, UK: Basil-Blackwell.

1591	Portney PR. 1981. Housing prices, health effects, and valuing reductions in risk of death. Journal of

1592	Environmental Economics and Management 8:72-78.

1593	Pratt JW, Zeckhauser RJ. 1996. Willingness to pay and the distribution of risk and wealth. Journal of

1594	Political Economy l(4):747-763.

1595	Revesz RL. 1999. Environmental regulation, cost-benefit analysis, and the discounting of human lives.

1596	Columbia Law Review 99(4):941-1017.

1597	Rice DP, Cooper BS. 1967. The economic value of human life. American Journal of Public Health 57(11):1954-

1598	1966.

1599	Rosen S. 1988. The value of changes in life expectancy. Journal of Risk and Uncertainty 1:285-304.

1600	Sandy R, Elliott RF. 1996. Unions and risks: their impact on the level of compensation for fatal risk.

1601	Economica 63:291-310.

1602	Savage 1.1993. An empirical investigation into the effect of psychological perceptions on the willingness-

1603	to-pay to reduce risk. Journal of Risk and Uncertainty 6(l):75-90.

70


-------
1604

1605

1606

1607

1608

1609

1610

1611

1612

1613

1614

1615

1616

1617

1618

1619

1620

1621

1622

1623

1624

1625

1626

1627

1628

1629

1630

1631

1632

1633

1634

1635

SAB Review Draft

Scotton CR, Taylor LO. 2009. Valuing risk reductions: incorporating risk heterogeneity into a revealed
preference framework. Working paper, http://www.ncsu.edu/cenrep/research/ documents/
riskhet_REE_2009.pdf

Shepard DS, Zeckhauser RJ. 1984. Survival versus consumption. Management Science 30(4):423-439.

Siebert SW, Wei X. 1994. Compensating wage differentials at workplace accidents: evidence for union and
nonunion workers in the UK. Journal of Risk and Uncertainty 9(l):61-76.

Slovic P. 1987. Perception of risk. Science 30(4):423-439.

Smith RS. 1974. The feasibility of an 'injury tax' approach to occupational safety. Law and Contemporary
Problems 38(4):730-744.

Smith et al. 2004. Do the Near Elderly Value Mortality Risks Differently? The Review of Economics and
Statistics 86(1): 423-429.

Smith VK, Van Houtven G, Pattanayak SK. 2002. Benefit transfer via preference calibration: "prudential
algebra" for policy. Land Economics 78(1):132-152.

Smith VK, Pattanayak SK, Van Houtven G. 2003. Using preference calibration for VSL estimation. RTI
Working Paper 03_03. http://www.rti.org/pubs/rtipaper_03_03.pdf

Smith VK, Pattanayak SK, Van Houtven GL. 2006. Structural benefit transfer: an example using VSL
estimates. Ecological Economics 60:361-371.

Smith TC, Spiegelhalter DJ, Thomas A. 1995. Bayesian approaches to random-effects meta-analysis: a
comparative study. Statistics in Medicine 14(24): 2685-2699.

Strand. 2002. Public- and private-good values of statistical life: results from a combined choice
experiment and contingent valuation survey. Working Paper 2002:2, University of Oslo, Health
Economics Research Program.

Subramanian U, Cropper M. 2000. Public choices between life saving programs: the tradeoff between
qualitative factors and lives saved. Journal of Risk and Uncertainty 21(1):117-149.

Sunstein CR. 2004. Valuing life: a plea for disaggregation. Duke Law Journal 54:385-445.

Szpiro GG. 1986. Measuring risk aversion: an alternative approach. The Review of Economics and Statistics
68(1):156-159.

Tsuge T, Kishimot A, Takeuchi K. 2005. A choice experiment approach to the valuation of mortality.

Journal of Risk and Uncertainty 31(l):73-95.

Um M, Kwak S, Kim T. 2002. Estimating willingness to pay for improved drinking water quality using
averting behavior method with perception measure. Environmental and Resource Economics 21: 287-302.

US EPA. 1999. The Benefits and Costs of the Clean Air Act, 1990 to 2010.

71


-------
SAB Review Draft

1636	USEPA. 2000a. Guidelines for Preparing Economic Analyses. EPA-240-R-00-003.

1637	http://yosemite.epa.gov/ee/epa/eed.nsf/pages/Guidelines.html

1638	USEPA. 2000b. Valuing fatal cancer risk reductions. Science Advisory Board review draft.

1639	USEPA. 2000c. An SAB Report on EPA's White Paper "Valuing the Benefits of Fatal Cancer Risk Reduction."

1640	EPA-SAB-EEAC-00-013. http://www.hks.harvard.edu/fs/rstavins/

1641	Monographs_&_Reports/SAB_Report_on_Fatal_Cancer.pdf

1642	USEPA. 2001. Arsenic Rule Benefits Analysis: An SAB Review. EPA-SAB-EC-01-008.

1643	http://www.epa.gov/ogwdw000/arsenic/pdfs/ec01008.pdf

1644	USEPA. 2003. Children's Health Valuation Handbook. EPA-100-R-03-003.

1645	http://yosemite.epa.gov/EE/epa/eed.nsf/pages/HandbookChildrensHealthValuation.html

1646	USEPA. 2004a. Value of statistical life analysis and environmental policy: a white paper.

1647	http://yosemite.epa.gov/ee/epa/eerm.nsf/vwAN/EE-0483-01.pdf/$file/EE-0483-01.pdf

1648	USEPA. 2004b. The Valuation of Mortality Risk Reduction: A Science Advisory Board Notification of a

1649	Consultation. EPA-SAB-CON-04-004. http://yosemite.epa.gOv/sab/SABPRODUCT.nsf/0/

1650	4CAC87DC95C91646852571AA00668C74/$File/sab_con_04004.pdf

1651	USEPA. 2006. Report of the EPA Work Group on VSL Meta-analyses.

1652	http://yosemite.epa.gov/ee/epa/eerm.nsf/vwAN/EE-0494-01.pdf/$file/EE-0494-01.pdf

1653	USEPA. 2007. SAB Advisory on EPA's Issues in Valuing Mortality Risk Reduction. (EPA-SAB-08-001)

1654	http://nepis.epa.gov/Adobe/PDF/P10007U3.PDF

1655	USEPA. 2008. Guidelines for Preparing Economic Analyses. External Review Draft.

1656	http://y osemitel.epa.gov/ee/epa/eerm.nsf/vwAN/EE-0516-01.pdf/$file/EE-0516-01.pdf

1657	Van Houtven G, Sullivan MB, Dockins C. 2008. Cancer premiums and latency effects: a risk tradeoff

1658	approach for valuing reductions in fatal cancer risks. Journal of Risk and Uncertainty 36(2):179-199.

1659	Viscusi WK. 1978. Labor market valuations of life and limb: empirical evidence and policy implications.

1660	Public Policy 26(3):359-386.

1661	Viscusi, WK. 1981. Occupational Safety and Health Regulation: Its Impact and Policy Alternatives.

1662	Research in Public Policy Analysis and Management 2: 281-299.

1663	Viscusi WK. 1992. Fatal Tradeoffs: Public and Private Responsibilities or Risk. New York, NY: Oxford

1664	University Press.

1665	Viscusi WK. 2003. Racial differences in labor market values of statistical life. Journal of Risk and Uncertainty

1666	27(3):239-256.

1667	Viscusi WK. 2004. The value of life: estimates with risks by occupation and industry. Economic Inquiry

1668	42(l):29-48.

72


-------
1669

1670

1671

1672

1673

1674

1675

1676

1677

1678

1679

1680

1681

1682

SAB Review Draft

Viscusi WK, Aldy JE. 2003. The value of a statistical life: a critical review of market estimates throughout
the world. Journal of Risk and Uncertainty 27(l):5-76.

Viscusi WK, Aldy JE. 2007. Labor market estimates of the senior discount for the value of statistical life.

Journal of Environmental Economics and Management 53(3):377-392.

Weiss P, Maier G, Gerking S. 1986. The economic evaluation of job safety: a methodological survey and
some estimates for Austria. Empirica 13(l):53-67.

Weitzman ML. 2010a. What is the "damages function" for global warming —and what difference might it
make? Climate Change Economics l(l):57-69.

Weitzman ML. 2010b. GHG targets as insurance against catastrophic climate damage. Working paper.

Wolfe J. 2007. What is the value of preventing a fatality? In Lewens, Tim, ed. Risk: Philosophical
Perspectives. New York, NY: Routledge.

Yaari ME. 1964. On the consumer's lifetime allocation process. International Economic Review 5(3):304-317.

Zhang J, Adamowicz W, Krupnick A, Dupont D. 2009. The role of altruism in the valuation of community
drinking water risks. Working Paper.

73


-------
SAB Review Draft

Tables and figures

Table 1: Cancer Valuation Literature Summary

Study

Timing

Treatment of
Morbidity

Dread

Risk context and
characteristics

Affected
Pop.

Other health
effects

Findings / Notes

Hammitt &

Latency periods

Yi sample: no

Not separately

Pesticide risks from

Adult selves

Organ: brain,

No statistical difference

Haninger (2010)

of 1,10, 20 years

symptom

treated

food. Safer food from



liver, bladder,

by cancer/non-cancer;

Choice experiment



descriptions;



Pesticide Safety

Adult others

lymphocytes

target organ; auto/other

Implied discount





System (not organic)





risks (for protecting



rates not stat diff

Vi sample: 150-200





Children

Mortality only (no

whole family



from zero ( -1.2

word descriptions



Auto accident for



non-fatal

simultaneously).



to 3.9)

Self-assessed
severity based on
EQ-5D and visual
analog scales



whole family
simultaneously from
"product" on next car
purchase.



outcomes)

Child VSL=1.8 Adult
Other VSL=1.15 Self

Insensitivity to number
of people for the "whole
household" question.

Alberini & Scasny

Latency periods

Description of

Rated

"cancer" designation

Adults

Respiratory

EC Cancer differential of

(2010a) "Labels &

of 0, 2, 5,10 years

morbidity or illness

subjectively

varied independently



fatality

50% relative to "general"

Perceptions . .





by

from dread

Children



VSL is consistent with

Discount rate =

Focus of study is

respondents



(Italy)

Cancer fatality

findings.

Choice experiment

zero (may reflect

mortality risk

for each type

Private good v.







changes in future
baseline risk)



of risk

nationwide public
program

Other independent
variables:

salience("familiarity")
exposure, sensitivity
to illness, beliefs in
prevalence



Auto fatality

Auto accident risks
valued less than
respiratory or cancer

VSL higher for public v.
private (if public
programs are effective)

"cancer" designation
effect persists after
controlling for other risk
characteristics.

74


-------
SAB Review Draft















VSL increases with
dread;

Alberini & Scasny
(2010b) "Context
and the VSL ..

Choice experiment

* The Italy sample
appears to overlap
with Alberini et al.
(2010a)

Yes (0, 2, 5,10
years)

Implicit discount
rates from 0.3 to
7.4%

Description of
morbidity or illness

Focus of study is
mortality risk

No

Public & Private
programs

Perceived

"effectiveness" of the
program

Finds that "risk
characteristics and
mode of delivery
primarily drive
heterogeneity in VSL"

Adults

Children

(CzechRep.
& Italy)

Respiratory illness

Cancer

Road-Traffic
accidents

"Evidence of cancer
premium":

~ 1.25x (Italy-children)
~ 1.90x (Italy-adults)

~ 1.75x (Czech-children)
~ 2.5 x (Czech-adults)

Premium for public
programs

Any premium for
reduced children's risk is
modest (small for cancer
risks, larger for other
causes).

Adamowicz et al.
(2008)

CVM and CE

Risks described
as community
deaths over a 35-
year time period
for microbials
and carcinogens

Symptoms
described for
microbial illness
and for bladder
cancer

Not addressed

Risk reductions are
strictly public

Describes tradeoffs
between reduced
microbials in DW and
reduced carcinogens

Households
(Canada)

Microbial illness
Microbial fatality

Bladder cancer

-	fatal

-	nonfatal

Modest cancer
"discount" (for
mortality)

Cancer VSL =
.85*Microbial VSL

Cancer illness = 20-50%
of cancer mortality

Cameron &
DeShazo (2008)

Choice experiment

Illness profile
over specific,
varying times

Results support
lower values for
longer latency.

No implicit

Health states
defined as:

-	Current health

-	Sickness

-	Remission years

-	Lost life-years

Illness characterized

Not addressed

Intervention is
generally a screening
and treatment
program to prevent
the given risk profile.

For auto accidents it is
a safety program.

Adults

12 major common
risks, including:

Heart disease,
heart attack,
stroke, respiratory
disease, diabetes,
Alzheimer's

Difficult to draw general
conclusions.

Heart attacks & heart
disease risks valued
similarly to some cancers
(and more than others).

75


-------
SAB Review Draft



discount rate
estimated.

by length and
severity (pain,
disability)







Cancer (5 types)
Auto accidents



Tsuge et al. (2005)

Choice experiment

Latency periods
of 0, 5,10 years

Implied discount
rate = 20%

Unclear, but does
not appear to be
detailed.

Focus is on
mortality.

Subjective

perceptions of

voluntariness,

controllability,

dread(pain),

dread(fear),

severity,

exposure



Adults
(Japan)

Accidents

Generalized
cancer

Heart disease
Non-specific

Unique formulation of
"quantity-based" VSL
distinguishing WTP for
opportunities for risk
reduction

Depends on model
specifications, but
perhaps 20% differential
over "general" risks;

reduced cancer risks
preferred to reduced
heart attack risk;

Hammitt & Liu
(2004)

CVM

Latent: 20 years
to onset of
symptoms
Acute: symptoms
"within a few
months"

Implied discount
rates of 1.5%
(with up to 3%
plausible)

brief description of
symptoms

progressive severity
over time from mild
to bedridden and
unable to care for
themselves

lasting 2-3 years
before mortality

Not addressed
directly

All symptoms held the
same except for
"cancer" designation:

Lung cancer v.
bronchitis (from
pollution from
factories)

Liver cancer v. liver
failure (from drinking
water contaminants)

Adults
(Taiwan)

Liver (failure v.
cancer)

Lung (bronchitis
v. cancer)

~30% differential for
cancer relative to
identical non-cancer
degenerative disease
(marginal significance)

Environmental context.

No "trauma" or
"accident" alternative for
comparison.

Philips et al. (1989)

No

No

No

No

Adults
(U.K.)

Motor vehicles
Heart disease
Fatal & Nonfatal

Mean estimates higher
for cancer; median
estimates are not

The following two studies are risk-risk studies

Van Houtven et al.
(2008)

Latency periods
of 5,15, 25-year
periods specified

Symptoms
described for three
types of cancer;

Not treated
separately

Organ-specific cancer
risks vs. auto-accident
risks.

Adults

Fatal cancer
(stomach, liver,
brain);

Significant cancer
differential (3x over auto
accidents at 5-year

76


-------
SAB Review Draft

Risk-Risk Survey

Morbidity varied
from 2 or 5 years

morbidity duration
varied separately
from latency







Fatal auto accident

latency; 1.5x at 25 years)

Differential declines with
length of latency;

Latency would need to
be 30+ years for
indifference

Magat et al. (1996)

risk-risk survey

Not addressed
explicitly

Symptoms
described for
lymphoma and
nerve disease

No

Included separate
treatments for non-
fatal lymphoma and
nerve disease

Respondents told not
to consider out of
pocket medical costs

Adults

Fatal lymphoma,

non-fatal
lymphoma;

fatal auto
accidents

nerve disease

No evidence of
differential for cancer
fatality (ratio of fatal
cancer: fatal auto is 1:1.)

Ratio of non-fatal cancer
to auto is ~.58.

The following studies examine cancer only (without comparison to other risks)

Carson & Mitchell
(2006)

Open-ended CVM

Not in survey;
VSL estimates
assume 25 years

No

No

Public/social decision

Cancer risks from
THM in drinking
water

Adults
(Household?
)

No

Cancer VSL depends
upon assumptions about
latency and discount
rate. Also sensitive to
risk reduction.

Assuming 0.4/100,000
reduction, 25-yr latency,
results range from

$3.4m at 3% to
$8.8m at 7%

Alberini et al.
(2010)

0, 2, 5,10 years

Employed a zero
discount rate for
estimation based
on prior work

Extent unclear

Unclear

Cancer risks from
hazardous waste in
Italy

Adults

Fatal Cancer (type
unspecified)

New estimates of the
cancer VSL using data
from 2008 survey in
Milan

Cancer VSL of ~$5.6m

77


-------
SAB Review Draft















About 20% higher if
delivered via public
program (if public
programs are considered
"effective").

Buzby et al. (1995)

No

No

No

Exposure to pesticides
in grapefruit

Grapefruit
customers
(Adults)

No

Makes assumptions
about lifetime exposure
to estimate VSL=$6.99m



Revealed Preference Cancer Valuation Studies

Study

Timing

Treatment of
Morbidity

Dread

Risk context and
characteristics

Affected

PoP-

Other health
effects

Findings / Notes

Gayer et al. (2000)

Hedonic Property

No

No

No

No; just "cancer" w/o
distinction between
fatal and non-fatal
cancers

Adults /

Household
near

Superfund
sites

No

cancer risk reductions
valued similarly to
workplace fatal risks

Gayer et al. (2002)

Hedonic Property

No

No

No

No; just "cancer" w/o
distinction between
fatal and non-fatal
cancers

Adults /

Household
near

Superfund
sites

No

$5.2m to $10.0m cancer
VSL with no latency (and
100% fatality.)

With 10-year latency:
-$6.2 to $11.7 at 3%
- $10.2 to $19.8 at 7%

Davis (2004)

Hedonic Property

Unclear

No

No

Pediatric leukemia
from cancer cluster; no
distinction between
fatal and non-fatal

Children

No

value of prevented
pediatric leukemia
ranges from $4.1m to
$11.5m depending on
model used

Ho and Hite (2008)

No

No

No

cancer mortality only
(didn't include non-

Adults

No

Hedonic property with
$6.0m Value of statistical

78


-------
SAB Review Draft

Hedonic Property







fatal)





cancer fatality (without
latency treatment or
assumptions).

Lott & Manning
(2000)

Hedonic wage

No

No

No

Cancer

Workers

No

Hedonic Wage
Cancer VSL = $12.4
million

79


-------
SAB Review Draft

Table 2. Select variables included in the stated preference meta-analysis dataset

Variable Name

Description

STUDY

Study identifier

PUBYEAR

Year study was published or released

PUBLISH

0=unpublished or working paper; l=published in a peer-review outlet
(includes book chapters)

JRU

0=does not appear in Journal of Risk and Uncertainty (JRU); l=published
in JRU

ALBSERIES

0=not part of the Alberini, Krupnick, Cropper and Simon series of
studies; l=part of this series

AUTO

O=non-auto/traffic risk; l=auto/traffic risk

ENVIRONMENTAL

O=non-environmental risk source; l=environmental-related risk (i.e., air
pollution, drinking water, hazardous waste site, or unspecified general
death risk)

PUBLIC

0=risk affects individual only; l=risk affects public

CANCER

0=non-cancer death; l=cancer death

ESTIMATES

Number of estimates reported or calculated from study

WTP

Willingness to pay for risk reduction (2009 US dollars)

WTP_SE

Standard error for WTP

VSL

VSL in millions, adjusted for inflation and income growth (2009 dollars)

SE

Standard error in millions of VSL estimate

MEAN

0=WTP/VSL is based on median WTP; 1=WTP/VSL is based on mean
WTP

YEARCONDUCT

Year study was conducted

US

0=non-US study; 1=US study

cv

0=choice experiment; l=contingent valuation

BASE

Baseline risk presented to survey respondents

REDUCE

Size of risk reduction presented to respondents

PCTREDUCE

Percent reduction in risk

TIMING

0=immediate risk reduction, l=latent risk reduction

LENGTH

Length of latency period in years (0=immediate risk reduction)

SIZE

Sample size used to calculate WTP/VS: estimate

MALE

0=female, l=male

AGE

Average age

RACE

Percent white

INCOME

Annual mean household income (thousands, 2007 US dollars)

HEALTH

Percent reporting exceptional or very good health, no reported disease
or illness, or non-smoker

NSCENARIO

Number of scenarios each respondent was asked to value

MODE

0= self administered survey mode, l=survey administered with an
interviewer (e.g., in-person, telephone

DOTS

0=ladder, bar chart used for visual aid; l=grid used for visual aid

SCOPE

0=no scope test performed or calculated, l=scope test performed or
calculated

WEAK

0=does not pass a weak scope test, l=passes a weak test, but WTP is less
than proportional to the size of the risk reduction

80


-------
SAB Review Draft

Variable Name

Description

STRONG

0=does not pass a strong scope test, l=passes a strong test; WTP is
proportional to the size of the risk reduction

81


-------
SAB Review Draft

Table 3. Stated preference dataset







Risk Characteristics







Study

Country

Sample Size

Cancer

Public

Latency (yrs)

Auto risk

Env. risk

Unspec. Source

Other risk type

Risk
reduction

WTP

(2009$)*

SE

Adamowicz et al. (2008)

USA

366

0

1

0

0

1

0

0

0.0000029

6.65 (1)

0.91

Adamowicz et al. (2008)

USA

366

1

1

10

0

1

0

0

0.0000029

6.03 (1)

0.75

Alberini and Chiabai (2007)

Italy

756

0

0

0

0

1

0

0

0.0001

6.03 (2)



Alberini et al. (2007)

Italy

782

0

1

0

0

1

0

0

0.000001

6.96 (1)



Alberini et al. (2004)

USA

548

0

0

0

0

0

1

0

0.0001

6.59 (3)

1.00

Alberini et al. (2004)

Canada

292

0

0

0

0

0

1

0

0.0001

5.05 (3)

0.66

Alberini et al. (2006a)

USA

403

0

0

10

0

0

1

0

0.0005

0.95 (4)

0.44

Alberini et al. (2006a)

Canada

589

0

0

10

0

0

1

0

0.0005

1.42 (4)

0.26

Alberini et al. (2006b)

France, Italy, UK



0

0

0

0

0

1

0

0.0005

3.22 (4)

0.57

Alberini and Scasny (2010)

Italy

1906

1

1

4.25

1

1

0

0

0.000425

4.68 (16)

0.30

Alberini and Scasny (2010)

Czech Republic

1506

1

1

4.25

1

1

0

0

0.000425

1.27 (16)

0.14

Alberini et al. (2006c)

Czech Republic

954

0

0

0

0

1

0

0

0.0003

3.11 (4)

0.21

Andersson and Lindberg (2009)

Sweden

216

0

0

0

1

0

0

0

0.0002

13.02 (5)



Andersson and Lindberg (2009)

Sweden

222

0

1

0

1

0

0

0

0.0002

7.45 (5)



Buzby et al. (1995)

USA

512

1

0

75

0

0

0

1

0.00000066

6.99 (6)



Cameron et al. (2008)

USA

1619

1

0

10

0

0

1

0

0.000001

0.86 (7)



Carson and Mitchell (2006)

USA

121

1

1

25

0

1

0

0

0.0000004

8.64 (8)



Corso et al. (2001)

USA

275

0

0

0

1

0

0

0

0.00005

4.29 (9)



Desaigues and Rabl (1995)

France

1000

0

1

0

1

0

0

0

0.000046

1.64 (1)



Gerking et al. (1988)

USA

861

0

0

0

0

0

0

1

0.00025

6.86 (6)



Gyrd-Hansen et al. (2007)

Norway

1168

0

0

0

0

0

0

1

0.0028

0.04 (1)



Hakes and Viscusi (2007)

USA

465

0

0

0

1

0

0

0

0.0001

7.22 (10)



Hammitt and Graham (1999)

USA

992

0

0

0

1

0

0

0

0.00005

2.96 (11)

0.32

Hammitt and Graham (1999)

USA

978

0

0

0

0

0

0

1

0.000073

2.72 (11)

0.56

Hammitt and Haninger (2010)

USA

1997

0.5

0

1

0

0

0

1

0.00015

6.77 (12)

1.24

Hammitt and Liu (2004)

Taiwan

1248

1

0

20

0

1

0

0

0.00005

1.94 (13)



Hultkrantz et al. (2006)

Sweden

225

0

0

0

1

0

0

0

0.000165

6.40 (14)



Itaoka et al. (2007)

Japan

248

0

0

0

0

0

1

0

0.001

2.92 (17)

0.76

Johannesson et al. (1997)

Sweden

2029

0

0

22.5

0





0

0.0002

5.13 (10)



Johannesson et al. (1996)

Sweden

389

0

0

0

1

0

0

0

0.000162

4.49 (18)

0.48

82


-------
SAB Review Draft







Risk Characteristics







Study

Country

Sample Size

Cancer

Public

Latency (yrs)

Auto risk

Env. risk

Unspec. Source

Other risk type

Risk
reduction

WTP

(2009$)*

SE

Johannesson et al. (1996)

Sweden

410

0

1

0

1

0

0

0

0.000162

3.73 (18)

0.48

Kidholm (1995)

Denmark

908

0

0

0

1

0

0

0

0.000022

2.38 (19)



Lanoie et al. (1995)

Canada

162

0

0

0

1

0

0

0

0.0002

2.92 (10)



Miller and Guria ((1991)

New Zealand

629

0

0

0

1

0

0

0



1.59 (21)



Morris and Hammitt (2001)

USA

167

0

0

20

0

0

0

1

0.046

0.19 (20)



Persson et al. (2001)

Sweden

675

0

0

0

1

0

0

0

0.00003

3.59 (1)



Philips et al. (1989)

U.K.

1563

1

0

0

1

1

0

0



6.90 (1)



Strand (2002)

Norway



0

0

1

0

1

0

0



0.57 (22)



Tsuge et al. (2005)

Japan

400

1

0

5

0

1

1



0.0001

3.62 (15)



Zhang, et al. (2009)

Canada

366

1

0



0

0

0

0



12.69 (23)



* The WTP and SE estimates reported in this table are adjusted for inflation (using the CPI) and income growth (using an elasticity of 0.5).

(1)	author's preferred

(2)	healthy 30-49 year old, based on mean and smaller risk reduction (from Table 7 in paper)

(3)	based on mean and smaller risk reduction (Table 6 in paper)

(4)	based on mean

(5)	based on parametric estimation (Table 7 in paper)

(6)	only estimate reported in paper

(7)	45 year old who is diagnosed with lung cancer 10 years after exposure, is sick for 5 years and then dies; estimate is chosen because it most
closely matches many EPA policy scenarios (Table 3 in paper)

(8)	based on corrected mean for the smallest risk reduction (Table 19.2 in paper) (note: We could also obtain other independent estimates for
different risk reductions)

(9)	from a model with co-variates for the smaller risk reduction using dots for a visual aid (Table 3 in paper)

(10)	based on the full sample

(11)	based on median (mean not reported) for the smallest risk reduction (Table 5 and 7 in paper)

(12)	based on model of WTP for reductions in risk to self, which is based on median WTP, one year latency and cancer set to 0.5 and affected
organs set to 0.25 (options are brain, bladder, liver and lymphocytes) (Table 2 in paper)

(13)	based on latent lung cancer from model with full set of co-variates (Table 3 in paper)

(14)	based on private risk reduction (there is also an estimate for a public risk reduction, but they are not independent)

83


-------
SAB Review Draft

(15)	only estimate in paper; reflects the idea that wtp is independent of the source of risk; CE asks about cancer, accidents, heart disease, over
different latency periods

(16)	based on pooled model (Table 5 in paper)

(17)	based on smaller risk reduction with no latency from wave 2 (where smaller risk reduction was presented first (Table 7 in paper)

(18)	based on standard estimates (Table 2 in paper)

(19)	based on mean estimate for risk reduction provided through an air bag (assumed to be a private risk reduction) using the maximum WTP
results (Table 2 in paper)

(20)	based on WTP for vaccine at age 60 (Table 3 in paper)

(21)	based on WTP for a safer car (Table 3 in paper)

(22)	based on WTP for private reductions in risk from environmental causes (Table 10 in paper)

(23)	based on WTP for private cancer risk reductions assuming no treatment or purchase of bottled water (Table 9 in paper)

84


-------
SAB Review Draft

Table 4: Hedonic Wage dataset









Sample
Characteristics













Study

Country

Sample

Sample
Size

Union

White

Male

Manual/Mfg

Blue Collar

Risk Variable

Mean
Risk

Nonfatal

Risk
Included
(l=Yes)

Workers'

Comp
Included?
(l=Yes)

WTP
(2009$)

SE

Smith (1974)

USA

CPS 1967; Census
of Manufactures

1960;
Employment and
Earnings 1963

3183

0

1

1

0

0

BLS 1966,1967

0.000125

1

0

14.06

5.87

Viscusi (1978)

USA

SWC 1969-1970

496

0

0

0

0

1

BLS, subjective
risk of job (SWC)

0.000118

1

0

3.72

2.15

Olson (1981)

USA

CPS 1978

5993

0

0

0

0

0

BLS 1973

0.0001

1

0

18.15

7.30

Viscusi (1981)

USA

PSID 1976

3977

0

0

0

0

0

BLS 1973-1976

0.000104

1

0

12.33

2.13

Marin and

Psacharopoulos

(1982)

UK

General
Household
Survey 1975

5509

0

0

0

0

0

OPCS
Occupational
Mortality
Decenniel Survey
1970-1972

0.00009

0

0

9.09

2.01

Dorsey and
Walzer (1983)

USA

CPS May 1978

1697

1

0

0

0

1

BLS 1976

0.000058

1

1

17.27

7.29

Dillingham and
Smith (1984)

USA

CPS May 1979

879

0

1

0

0

0

BLS industry data
1976,1979

0.00012

1

0

4.81

2.30

Leigh and Folsom
(1984)

USA

PSID 1974, QES
1977

1529

0

1

0

0

0

BLS

0.00014

1

0

15.12

6.40

Dillingham (1985)

USA

QES 1977

514

0

0

0

0

0

BLS 1976; NY

workers'
compensation
data 1970

0.00014

0

0

6.21

3.47

85


-------
SAB Review Draft









Sample
Characteristics













Study

Country

Sample

Sample
Size

Union

White

Male

Manual/Mfg

Blue Collar

Risk Variable

Mean
Risk

Nonfatal

Risk
Included
(l=Yes)

Workers'

Comp
Included?
(l=Yes)

WTP
(2009$)

SE

Weiss et al. (1986)

Austria

Austrian
Microcensus File
of Central Bureau
of Statistics 1981

4225

0

0

0

0

0

Austrian Social
Insurance Data
on Job-related
Accidents 1977-
1984

0.00013

1

0

12.23

5.03

Moore and
Viscusi (1988)

USA

PSID 1982

1349

0

1

0

0

0

BLS1972-1982,
NIOSH National
NTOF Survey
1980-85

0.00008

0

1

13.15

5.21

Garen (1988)

USA

PSID 1981-1982

2863

0

0

0

0

1

BLS 1980,1981

0.000108

1

0

24.08

5.17

Meng (1989)

Canada

National Survey
of Class Structure
and Labour
Process 1981

718

0

0

0

0

0

Labour Canada
and Quebec
Occupational
Health and Safety
Board 1981

0.00019

0

0

6.85

3.99

Meng and Smith
(1990)

Canada

National Election
Survey

777

0

0

0

1

0

Labour Canada
and Quebec
Occupational
Health and Safety
Board 1981-83

0.00012

0

0

1.78

3.28

Berger and
Gabriel (1991)

USA

1980 Census

22837

0

0

1

0

0

BLS 1979

0.000097

0

0

11.17

1.95

Leigh (1991)

USA

PSID 1974,1981

1502

0

0

1

0

1

BLS 1979

0.000134

0

0

10.74

3.23

Kniesner and
Leeth (1991)

USA

CPS 1978

8868

0

0

0

1

0

NIOSH NTOF
Survey 1980-1985

0.000436

1

1

0.67

0.46

Gegax (1991)

USA

Authors' mail
survey 1984

228

1

0

0

0

0

Workers' assessed
fatality risk at
work 1984

0.00086

0

0

3.92

1.99

Martinello and
Meng (1992)

Canada

Labor Market
Activity Survey
1986

4352

0

0

0

1

0

Labor Canada
and Statistics
Canada 1986

0.00025

1

0

4.45

1.34

86


-------
SAB Review Draft









Sample
Characteristics













Study

Country

Sample

Sample
Size

Union

White

Male

Manual/Mfg

Blue Collar

Risk Variable

Mean
Risk

Nonfatal

Risk
Included
(l=Yes)

Workers'

Comp
Included?
(l=Yes)

WTP
(2009$)

SE

Cousineau et al.
(1992)

Canada

Labor Canada
Survey 1979

32713

0

0

0

1

0

Quebec
Compensation
Board

7.64E-05

1

0

7.01

0.67

Siebert and Wei
(1994)

UK

General
Household
Survey 1983

1353

1

0

1

1

0

Health and Safety
Executive 1986-88

3.32E-05

1

0

20.70

9.85

Leigh (1995)

USA

PSID 1981

1528

0

0

1

0

1

NIOSH 1980-85

0.00011

0

0

16.23

3.04

Sandy and Elliot
(1996)

UK

Social Change
and Economic
Life Initiative
Survey 1986

440

0

0

1

1

0

OPCS
Occupational
Mortality
Decenniel Survey
1979/80-1982/83

4.52E-05

0

0

76.00

32.55

Milleret al. (1997)

Australia

Australian
Census of
Population and
Housing 1991

18,850

0

0

1

0

0

Worksafe
Australia,
National
Occupational
Health and Safety
Commission
1992-93

0.000068

0

0

23.86

1.82

Meng and Smith
(1999)

Canada

Labor Market
Activity Survey
1986

1503

0

0

0

0

0

Ontario Workers'
Compensation
Board

0.00018

1

1

3.33

0.86

Kim and Fishback
(1999)

South
Korea

Ministry of
Labor's Report on
Monthly Labor

Survey and
Survey on Basic
Statistics for the
Wage Structures

321

0

0

1

0

0

Ministry of
Labor's Analysis
for Industrial
Accidents

0.000485

1

1

2.20

0.45

Arabsheibani and
Marin (2000)

UK

General
Household
Survey (1980s)

3608

0

0

1

0

0

OPCS
Occupational
Mortality
Decennial Survey
1979-80

0.00005

1

0

43.88

8.82

87


-------
SAB Review Draft









Sample
Characteristics













Study

Country

Sample

Sample
Size

Union

White

Male

Manual/Mfg

Blue Collar

Risk Variable

Mean
Risk

Nonfatal

Risk
Included
(l=Yes)

Workers'

Comp
Included?
(l=Yes)

WTP
(2009$)

SE

Gunderson and
Hyatt (2001)

Canada

Survey of Ontario
Workers with
Perma- nent
Impairment

2014

0

0

0

0

1

Ontario Workers'
Compensation
Board

0.000167

1



34.03

4.83

Viscusi (2003)

USA

CPS MORG 1997

83625

0

1

0

0

0

CFOI1992-1997

3.62E-05

1

1

21.45

2.01

Leeth and Ruser
(2003)

USA

CPS ORG 1996-98

45001

0

0

1

0

1

CFOI1996-1998

9.76E-05

1

1

3.61

0.80

Smith et al. (2004)

USA

Health &
Retirement
Survey (Wave 1)

3632

0

0

0

0

0

BLS1993

5.8E-05

0

0

7.97



Viscusi (2004)

USA

CPS MORG 1997

99033

0

0

0

0

0

CFOI 1992-1997

4.02E-05

1

1

6.79

0.80

Kniesner et al.
(2006)

USA

PSID 1997

1875

0

0

1

0

0

CFOI 1992-1997

0.00004

0

0

29.59



Viscusi and Aldy
(2007)

USA

CPS MORG 1992-
1997

120,008

0

0

0

0

0

CFOI 1992-1997

0.00004

1

1

12.23



Aldy and Viscusi
(2008)

USA

CPS MORG 1993-
1997

123,439

0

0

0

0

0

CFOI 1992-2000



1

1

13.09



Evans and Smith
(2008)

USA

Health &
Retirement
Survey

2,708

0

0

0

0

0

CFOI

0.000064

0

0

13.06



Scotton and
Taylor (2009)

USA

CPS MORG 1996-
1998

43,261

0

0

0

0

0

CFOI 1992-1997

4.895E-05

1

0

6.16

1.89

88


-------
SAB Review Draft:

Appendix A

This appendix gives some illustrative numerical examples using the simple static (single-period)
structural benefit transfer function from Section 5.2.1, and a more formal exposition of the life-cycle
modeling framework discussed in Section 5.2.2. Table B1 shows willingness to pay values for a range of
mortality risk reductions using the static model in Section 5.2.1. The first three columns in the table show
the difference between the marginal approximation and the exact WTP [$] for a range of changes in
baseline risks Ap [yr_1 ]. The final six columns in the table show WTP [$] and m" [yr-1] (explained
below) for a range of Ap's and three possible values of ('>, accounting for the behavioral response
described in Section 5.2.1. To determine the maximum willingness to pay for an exogenous change in
background mortality risks, we must solve the two-equation system comprised of the equality between
expected utility with and without the policy,

p0-m In a y + W0 + amp - p0+Ap- m* In a y + W0 + am- WTP ,

and the first-order condition for maximized expected utility with respect to job-risk with the policy and a
reduction in income equal to WTP, i.e.,

flam"13'1 p0+Ap- ra" / y + W0 + am- WTP - In a y + W0 + am"p - WTP = 0,

where m" is the job-risk level that the individual would choose if her baseline survival probability were
increased by Ap and if she were charged the amount WTP for this change. The level of m that she would

actually choose after the policy is implemented would depend on the actual cost of the policy to her.

The main lesson from these examples is that—when preferences for consumption and risk are not
separable, as in this example—if individuals are able to freely adjust their job risk level, then WTP
generally will be higher and the total number of "statistical lives saved" will be lower than otherwise
predicted under the assumption of no behavioral response. In fact, if (3 = 1 and if each individual were

89


-------
SAB Review Draft-

charged their maximum WTP for the change, then the individuals' behavioral responses would fully

offset the changes in their baseline mortality risk. In this extreme case, WTP would exactly equal

wtp ¦ Ap and, if each individual had to pay this full amount to fund the policy, then the number of "lives

saved" would be zero. If the full costs of the policy were less than the aggregate WTP, then both the net
social benefits and the number of statistical lives saved would be positive, though the latter still would be
less than ApxN . If the full costs of the policy were greater than the aggregate WTP, then of course the

net social benefits would be negative, but also note that the number of statistical lives "saved" would be
negative as well—that is, even though environmental risks were reduced, the policy would increase
overall mortality rates since people's behavioral responses to the increased costs would involve shifting
to jobs with higher mortality risks. The numerical results in Table B1 are not necessarily intended to be
realistic, especially considering that they involve mortality risk reductions that are much larger than
those we would typically expect from most environmental regulations, but they nevertheless highlight
the importance of calculating benefits and costs simultaneously for non-marginal policies when
behavioral adjustments are expected.

Next, a brief exposition of a generalized life-cycle (multi-period) model may help to describe the
potential usefulness of this framework as a basis for structural benefit transfers of mortality risk
reductions. Suppose that the value function for a representative individual is given by

T

Va = ct,ht,t sa te p ° , where u ct,ht,t is utility in period t (assumed here to depend on

t=a

consumption ct, health status ht, and possibly age t), sr is the probability of surviving to the beginning

of age t +1 given that the individual is alive at the beginning of age t , sat — ]~[^ Sr, and T is the
individual's maximum possible lifespan. Marginal willingness to pay at age a for mortality risk

reductions (or, equivalently, an increase in survival probability) at age b (> a) is wtp , h =

dca dVJ8sb
ds, dVJdc

90


-------
SAB Review Draft-

To help interpret this willingness to pay measure, we can break the value function into two parts

h-l	T

at some future age t = b, Va = ZM ct,ht,t sate~pt~a + ct,ht,t sate pt " , then re-write second term

t=b

on the right hand side of this equation in terms of the value function at age b,

h-1	dV

Va = ^u ct,ht,t sate pt " +Vhsahe ph " , which means —- = Vbsa ble pb " -33 Thus, the marginal

t-a	SS b

willingness to pay at age a for a reduction in mortality risk at some future age b is

V-p b-a

bSab-ie

uotp h =			,34 This is the expected remaining lifetime utility at the beginning of age b,

du c ,h ,a /dc

a' a' ' a

discounted by the survival probability and the pure rate of time preference between ages a and b, and
then monetized by the marginal utility of consumption at age a.

Developing a usable structural benefit-transfer function based on a lifecycle framework would be
challenging. Estimating or calibrating such a model would require specifying or solving for the life-cycle
pattern of consumption, calibrating or estimating the pure rate of time preference, and specifying a

33	Throughout this section we treat the path of consumption over the life cycle as exogenous; that is, we ignore any
behavioral responses to changes in mortality risks that would adjust the levels of consumption in future periods.

This simplification will be strictly valid only under some special conditions—namely, that that the individual can
never be a net borrower (Cropper and Sussman 1990, USEPA 2007 p D-15)— but it should provide a close
approximation for small changes in exogenous mortality risks. More specifically, we would expect it to provide a
close lower bound on willingness to pay in most cases of interest—a lower bound because it assumes that the
individual is constrained to maintain the same consumption path after the change, and a close approximation
because we would expected any adjustments in future consumption levels to be very small for reasonably small
changes in mortality risks.

34	Direct inspection of this equation suggests some simple comparative static results: (1) wtpa b decreases with the
latency period b-a because all elements of the numerator — Vb+1, sa hl, and e p h —decrease and the denominator
does not change. (2) wtpa a could increase or decrease with a because, while Va+1 and sa a l decrease with a, the

denominator could decrease or increase with a depending on the pattern of consumption and health status over the
life cycle (USEPA 2007 p D-16). If the pattern of consumption were perfectly flat over the life cycle, and if utility
depended only on consumption and not health status or age per se, then wtpa a would unambiguously decrease with

age. However, observed consumption patterns generally are not flat; consumption typically is low in the early
(adult) years, high in middle age, and lower again in later years, which, all else equal, would tend to increase then
decrease wtp .

r a,a

91


-------
SAB Review Draft-

functional form for the period utility function u c,, h,, t and calibrating or estimating its parameters.

The simplest reasonable implementation of such an approach might proceed as follows:

1.)	Specify the lifetime pattern of consumption for a "representative" individual as the pattern of
average consumption levels for a random sample of individuals of various ages from the population
of interest. Alternatively, multiple representative life-cycle consumption patterns could be generated
based on average consumption levels for sub-samples of the population, e.g., by gender, race,
geographic region, etc., as appropriate for the exposed sub-population relevant for the policy to be
examined.

2.)	Set p equal to a suitable central value from a relevant set of revealed or stated preference studies
(presumably somewhere between, say, 0% and 5% per year).

3.)	Assume the utility function is of the standard CRRA form with a lower bound on utility:

ut = n - d1^ / 1 - Tj . Then either

a.	set 71 equal to a suitable central value from a relevant set of revealed or stated preference studies
(presumably somewhere between, say, 0.5 and 3), and use at least one valid estimate of
willingness to pay for well-specified mortality risk changes from the revealed or stated
preference literature to calibrate d, or

b.	use at least two valid estimates of marginal willingness to pay from the RP or SP literature to
calibrate i] and d simultaneously.

Such a calibrated life-cycle model then could be used to calculate wtpa b for all combinations of a and b

for each representative individual identified in step 1. These estimates then could be transferred to any
pattern of mortality risk changes that are projected for one or more policies under consideration. More
sophisticated versions of this approach could specify ut as a function of age and/or health status, which
might facilitate a link to the QALY literature.

92


-------
SAB Review Draft

93


-------
SAB Review Draft

Table Al. Maximum willingness to pay for a range of changes in survival probabilities, Ap, based on a
marginal approximation (wtp ¦ Ap) and direct calculation (WTP ), with and without a behavioral
response. Baseline job risk is m = 0.006. Estimates of the adjusted job risk with a behavioral response (
m" ) assume that the individual's income is simultaneously reduced by WTP (that is, expected utility
without the policy is equal to that with the policy combined with the charge WTP).



No behavioral
response

With behavioral response







P =

0.33

P =

0.67

P

= 1

Ap [yr

wtp ¦ Ap

WTP

WTP

m" [yr

WTP

m" [yr

WTP

m" [yr



[$]

[$]

[$]



[$]



[$]



0.000005

40.0

25.0

40.0

0.0060034

400

0.0060040

40.0

0.0060050

0.00005

400.0

397.7

399.2

0.0060337

399.6

0.0060404

400.0

0.0060500

0.0005

4,000.0

3,778.1

3,926.6

0.0063399

3,956.7

0.0064062

4,000.0

0.0065000

0.005

40,000.0

23,773.6

33,977.6

0.0096250

36,431.3

0.0102351

40,000.0

0.0110000

94


-------
SAB Review Draft

95


-------