Interface Between Risk Assessment and Economics: Economic Valuation Methods

1
Interface Between Risk Assessment and Economics
Economic Valuation Methods

• Presentation to IRAC Conference
• Risk Assessment, Economic Analysis, and Foodborne
  Illness Regulations
• Al McGartland, USEPA
• November 16, 2007

The opinions in this presentation do not
necessarily represent those of EPA.
2
Evolution of Risk Assessment at EPA

• beginnings of the field of risk assessment
• emphasis on oral route per FDA precedent

• adopt RA/RM paradigm
• guidelines basic methodologies
• data bases (IRIS)

1980s

• new tools/data bases
• refinement of existing tools
• understanding mechanisms of
  action/interactions
• ecological assessment

1990s

• complex mixtures
• sensitive subpopulations
• non-linear dose-response
• input into economic analysis

2000s
3
Roles of Risk Assessment
Risk Assessment
Risk Management
Risk Assessment - cancer - non-cancer
Policy Decision - standards - information
Protective
4
Roles of Risk Assessment
Risk Assessment
Risk Management

Risk Assessment - cancer - non-cancer
Policy Decision - standards - information
Protective

• Economic Analysis
• benefit-cost analysis
• cost effectiveness
• economic impacts

Predictive
5
Damage Function Approach
Environmental Concentrations
Exposure
Emissions

• Benefits analysis draws upon results of other
  disciplines
• Simplified linear process of benefits analysis,
  but there may be feedback loops
• For example, changes in individual behaviors
  evaluated in the benefits box may have
  implications for emissions or exposure

Effects (dose-response)
Benefits ( for each effect)
6
From Risk Assessment to Economics

• Should focus on well-defined health endpoints
  that have an effect on individual sense of
  well-being.
• hormone level vs. hypothyroidism
• Should be probabilities to represent the expected
  changes in risk to an individual
• probabilities not thresholds
• upper bound risk estimates
• Should provide information on when risks change
  due to exposure changes
• substantial lags between exposure and cancer

7
Health Science Inputs and Economic Valuation

• Epidemiological data are preferred for valuation
  in policy analysis
• observed health effects in humans
• incorporates behavioral responses
• Results from animal studies may be used
• Cancer dose-response risk estimates are still
  usable although their applicability can be a
  concerns (e.g, upper-bound estimates)
• Non-cancer risk assessment Reference doses and
  other measures that dont estimate probabilities
  cannot be utilized in benefits analysis

8
Cancer vs. Non-Cancer Benefits
9
Economic Analysis
Predictive Risk Assessment Data
Cost- Effectiveness Analysis
Benefit-Cost Analysis
Analytic Approach
Health Value Measure
Cost of Illness
Willingness to Pay
Quality Adjusted Life Years
Disability Adjusted Life Years
10
Economic Analysis
Predictive Risk Assessment Data
Cost- Effectiveness Analysis
Benefit-Cost Analysis
Analytic Approach
Health Value Measure
Cost of Illness
Willingness to Pay
Quality Adjusted Life Years
Disability Adjusted Life Years
11
Benefit-Cost Analysis

• Simulates the private market test for public
  goods
• BCA quantifies all benefits and costs of
  producing environmental protection
• Consistent with private markets, the efficient
  outcome is the option that maximizes net benefits
• BCA is, therefore, an efficiency test for the
  production of environmental protection
• .

12
Benefits

• Rules for determining the benefit, or value, of
  environmental protection are the same as those in
  the private market quantify the willingness to
  pay for the environmental commodity
• Value is determined by what consumers are willing
  to pay for a commodity, not what the analyst
  believes the value to be
• Economists look to see what values society places
  on environmental goods
• Analysts do not assign their own values
• Because environmental protection it is a public
  good, the benefits of a policy are the sum total
  of each affected individuals willingness to pay
  for the policy

13
Values for use in Benefit-Cost Analysis

• Cost of illness ()
• Each avoided health effect converted to
  -equivalent based on lost production and health
  care costs
• lost production may include household production
• Clearly defined and measurable
• Limited measure
• not necessarily based on individual preferences
• incomplete, does not include pain and suffering
• price of health care may not equal costs
• generally, loosely, characterized as a lower
  bound on willingness to pay

14
Willingness to Pay

• Willingness to Pay for risk reduction ()
• Each avoided health effect converted to based
  on preferences of the individuals affected
• Well-being is expressed in a general utility
  function Utility(health, all other goods)
• Willingness to pay is...
• Ones own WTP for ones own risk reduction
• WTP for others may be included under strict
  conditions
• Affected and constrained by income
• willingness to pay includes ability to pay
• Willingness to pay is not
• A measure of on ex post compensation experiencing
  health effect
• WTP is measured ex ante for risk reductions

15
Health Benefits Methods

• Methods are based on where people appear to
  tradeoff risk and wealth
• Hedonics
• Wage-risk how much additional wages to workers
  require to take risky jobs?
• Housing prices extra housing price for safer
  area or higher env. quality?
• Averting Behaviors how much do individuals pay
  for safety devices and safer vehicles, etc.?
• Stated Preference (Surveys) how much do
  individuals state they will pay for a specified
  risk reduction

16
Value of a Statistical Life (VSL)

• The value of mortality risk reductions is often
  summarized by calculating the value of a
  statistical life or VSL
• What is a statistical life?
• Suppose a public policy were to reduce mortality
  risk by 1 in 10,000
• If 10,000 were affected by this policy then,
  statistically, 1 death would be avoided
• We have no idea before we implement the policy
  which life will be saved.
• What is the Value of a Statistical Life (VSL)?
• Derived from the value or WTP individuals place
  on a small change in risk

17
VSL Key Points

• VSL estimates do NOT capture the value of an
  identifiable person
• VSL estimates do NOT capture how much an
  individual would be willing to pay to save his
  own life with certainty or the life of a loved
  one
• Instead, VSL estimates represent (or are derived
  from) individual willingness to pay for a SMALL
  change in the probability of dying

18
Example VSL Calculations

• Suppose that each of 10,000 people were willing
  to pay 500 each for the 1 in 10,000 risk
  reduction
• The value of that statistical life is the sum of
  the total of the individual amounts the 10,000
  people would pay for the risk reduction
• Put more generally, the value of a statistical
  life is calculated
• VSL WTP (1/risk reduction)
• In our case this would amount to
• VSL 500(1/(1/10,000)) 5,000,000

19
Issue VSL and Life Expectancy
20
Valuing changes in morbidity risk

• Morbidity (illness) value of statistical
  illness based on WTP to reduce risks or avoid
  illness
• We lack WTP estimates for many non-fatal
  illnesses
• each illness is unique to some extent (severity,
  frequency, duration)
• requires unique research or systematic
  extrapolation from others
• EPA policy and practice
• No set values for particular illness
• Often use WTP values for similar illnesses
• Non-fatal bladder cancer use range based on WTP
  per statistical chronic bronchitis (600,000) to
  non-fatal lymphoma (3 million)
• Cost-of-illness methods can be difficult
• Reporting is censored. Likely that many minor
  illnesses are never seen by physician.

21
Additional Considerations (WTP)

• Equity Considerations
• WTP is related to income
• What if studies demonstrate the elderly have
  lower VSL?
• recent results suggest health status has little
  effect on valuing mortality risks
• Economic criticisms
• sensitive to risk perceptions
• does not include peoples WTP for risk reductions
  to other people
• Dollar terms can be compared directly to costs
• Very flexible theoretically...
• incorporates broad range of health effects
• can incorporate context (source of risk)

22
Issue Timing of Exposure changes and risk
reductions

• The timing of risk reductions matters for the
  value of benefits
• Generally, risk reduction like any economic good
  is preferred sooner rather than later (time
  discounting)
• Also, the shorter the cessation lag
• The more cases are avoided (or lives saved)
  over the life of the rule or the analysis
• The fewer deaths from other causes before the
  benefit is realized
• These data are not often part of the risk
  assessment

23
Impact of Cessation Lags on Benefits
A
B
C
D
Alternative Cessation Lag cases avoided Benefits (m) 3 Benefits (m) 7
Immediate Benefits (ABCD) 120 429 257
Exponential Decay (ABC) 100 330 176
Lagged Step Function (AB) 60 162 64
30-year fixed lag (A) 30 65 18
24
Cost Effectiveness Analysis

• Benefits are not measured in non-monetary units
• Ranks alternatives by per unit,
• may determine the best alternative, but cannot
  determine if any of the alternatives should be
  done.
• For mortality risks typical units
• Lives saved gt rank by per life saved
• Life years gt rank by per life-year saved
• Neither of these measures accomodates non-fatal
  illnesses
• Cost-effectiveness cannot accommodate mortality
  and morbidity unless they are combined into some
  common measure
• Quality Adjusted Life Years (QALYs) combine fatal
  and non-fatal illnesses into one measure

25
QALYs for Cost-Effectiveness

• Quality-Adjusted Life Years (QALYs)
• health conceptualized as having two dimensions
• longevity (years)
• health-related quality of life
• outcome is total years weighted by quality in
  each year
• well-being is defined by health profile over time
• Utility U(Quality, longevity)
• wealth does not enter into this framework

26
Quality-Adjusted Life Years (QALYs)

• Assume a policy extends the baseline health
  profile (white)
• The gain in QALYs is shown in gray
• Quality years at that quality
• Note the risk assessment requirements a profile
  of expected future health outcomes and time in
  each health state

27
Additional Considerations (QALYs)

• Equity considerations
• interventions for young preferred to
  interventions for old
• Young have more life years remaining
• life extensions for healthy preferred to life
  extensions to less-healthy
• Healthy have a higher quality than chronically
  ill
• Economic criticisms
• assumptions are restrictive and may not reflect
  actual preferences
• concern about quality of surveys (data)
• How to deal with non-health effects (e.g.,
  environmental) from the policy in a
  cost-effectiveness framework

28
WTP QALY Some Conceptual Comparisons
WTP QALY
Type of Analysis Benefit-Cost Analysis Cost-Effectiveness Analysis
Tradeoff basis money and risks to health health and longevity
Endowment / Comparison individual wealth health profile
Social Goal Max. self-assessed welfare Sum (individual WTP () values) Max. health - weighted longevity Sum (quality years)
29
Concluding Thoughts

• Economics offers tools for risk management
• Risk assessment must be predictive to support
  economic analysis
• This kind of risk assessment is being actively
  pursued
• There are two widespread economic approaches to
  valuing changes to health and risk benefit-cost
  analysis and cost-effectiveness analysis
• Benefit-cost analysis and cost-effectiveness
  analysis approaches use very different metrics to
  assess the value of risk changes
• BCA and CEA will not necessarily point to the
  same policy or standard.