March
12,
2003
316(
b)
Memorandum
Date:
March
12,
2003
To:
The
§
316(
b)
Record
From:
Lynne
Tudor,
OW/
OST;
Ryan
Wardwell,
Elena
Besedin,
Robert
J.
Johnston,
Abt
Associates
Subject:
Comparison
of
Non­
use
and
Use
Values
from
Surface
Water
Valuation
Studies
Including
both
use
and
non­
use
values
when
estimating
the
benefits
of
an
environmental
resource
provides
the
most
comprehensive
estimate
of
the
value
of
the
resource,
so
that
total
value
can
be
compared
to
total
social
cost.
Use
values
alone
may
seriously
understate
total
social
value.
 
Nonuse
values,
like
use
values,
have
their
basis
in
theory
of
individual
preferences
and
the
measurement
of
welfare
changes.
According
to
theory,
use
values
and
non­
use
values
are
additive 
(
Freeman,
1993).
1
Recent
economic
literature
provides
substantial
support
for
the
hypothesis
that
non­
use
values
are
greater
than
zero.
When
small
per
capita
non­
use
values
are
held
by
a
substantial
fraction
of
the
population,
they
can
be
very
large
in
the
aggregate.
 ...
there
is
a
real
possibility
that
ignoring
non­
use
values
could
result
in
serious
misallocation
of
resources 
(
Freeman,
1993).
Stated
preference
methods,
or
benefits
transfers
based
on
stated
preference
studies,
are
the
only
generally
accepted
techniques
for
estimating
non­
use
values.
Stated
preference
methods
rely
on
surveys,
which
ask
people
to
state
their
willingness
to
pay
for
particular
ecological
improvements,
such
as
increased
protection
of
fishery
resources.
Benefits
transfer
involves
adapting
research
conducted
for
another
purpose
in
the
available
literature
to
address
the
policy
questions
in
hand.
Because
benefits
analysis
of
environmental
regulations
rarely
affords
enough
time
to
develop
original
stated
preference
surveys
that
are
specific
to
the
policy
effects,
benefits
transfer
is
often
the
only
option
for
providing
information
to
inform
a
policy
decision.

The
first
step
in
developing
a
benefits
transfer
approach
to
estimating
non­
use
values
of
environmental
regulations
is
systematic
analysis
of
the
available
economic
studies
of
non­
use
values.
In
response
to
public
comments
regarding
the
analysis
of
non­
use
values
in
the
proposed
rule,
the
U.
S.
Environmental
Protection
Agency
(
EPA)
continues
to
review
and
analyze
information
from
empirical
studies
of
non­
use
values,
including
studies
of
option,
existence,
and
preservation
values.
The
purpose
of
this
review
is
to
1
According
to
Freeman
(
1993),
this
additive
property
holds
under
traditional
conditions
related
to
resource
levels
and
prices
for
substitute
goods
in
the
household
production
model.

­
1­
March
12,
2003
report
on
the
range
of
non­
use
values
for
water
resources
in
the
economic
literature,
to
compare
estimates
of
use
and
non­
use
values
for
users
and
nonusers,
and
explore
the
feasibility
of
deriving
non­
use
values
based
on
these
comparisons.
The
Agency
intends
to
use
results
from
this
literature
review
to
develop
approaches
that
are
more
consistent
with
the
recent
economic
literature
for
estimating
the
non­
use
value
of
aquatic
resources
that
are
potentially
affected
by
impingement
and
entrainment
(
I&
E).

The
Agency s
review
of
the
relevant
economic
literature
showed
that
available
surface
water
valuation
studies
that
separate
use
and
non­
use
values
focus
primarily
on
water
quality
changes.
Although
all
reviewed
studies
focusing
on
water
quality
improvements
indicated
these
improvements
would
affect
recreational
fishery
resources,
the
non­
use
values
from
such
improvements
may
reflect
additional
services
provided
by
surface
water
bodies.
One
key
challenge
in
making
the
results
of
this
analysis
applicable
to
estimating
non­
use
benefits
of
the
316(
b)
regulation
is
to
develop
an
adjustment
factor
for
total
non­
use
values
from
water
quality
improvements
to
reflect
non­
use
values
for
fishery
resources
only.

The
rest
of
this
document
is
organized
as
follows:

Section
1:
Literature
Review
Procedure
and
Organization
­
Describes
the
literature
review
procedure,
the
organization
of
studies,
and
the
selection
of
studies
to
be
used
in
the
analysis;

Section
2:
Value
Concepts
and
Valuation
Techniques
­
Provides
a
brief
review
of
the
concept
of
value
and
valuation
techniques;

Section
3:
Description
of
Studies
­
Presents
characteristics
of
studies
that
could
be
used
in
a
future
meta­
analysis;

Section
4:
Methods
of
Measuring
Non­
Use
Value
­
Reviews
methods
of
isolating
use
and
nonuse
value
and
provides
examples
of
each;

Section
5:
Summary
of
Non­
Use
and
Use
Values
for
Water
Resources­
Describes
the
calculation
and
presentation
of
mean
and
median
use
values;

Section
6:
Regression­
Based
Meta­
Analysis
­
Discusses
further
steps
in
analyzing
non­
use
values
for
water
resources
using
a
regression­
based
meta­
analysis,
methodology,
and
potential
model
specifications;

Appendix
A:
Compendium
of
Non­
Use
Analysis
Studies
and
Characteristics
­
Contains
a
compendium
of
the
selected
studies
that
estimate
WTP
for
quality
changes
that
affect
aquatic
habitat
and
or
recreational
fishing.

Appendix
B:
Summary
Descriptions
of
Valuation
Studies
­
Presents
publication
information
and
provides
summary
descriptions
of
the
studies
that
could
be
used
in
a
future
meta­
analysis.

­
2­
March
12,
2003
1.0
Literature
Review
Procedure
and
Organization
EPA
performed
an
in­
depth
search
of
the
economic
literature
to
identify
valuation
studies
that
estimate
both
use
and
non­
use
portions
of
WTP
for
quality
changes
that
affect
aquatic
life
habitats
and/
or
recreational
fishing.
As
a
starting
point,
the
Agency
selected
several
surface
water
valuation
studies
from
a
meta­
analysis
conducted
by
Brown
(
1993).
In
addition,
the
Agency
conducted
extensive
additional
research
to
identify
potentially
relevant
studies
published
since
the
publication
of
Brown s
research
and
any
that
were
not
included
in
his
meta­
analysis.
EPA
used
a
variety
of
sources
and
search
methods
to
identify
relevant
literature:

<
Review
of
EPA s
research
and
bibliographies
dealing
with
non­
market
benefits
associated
with
water
quality
changes;
<
Systematic
review
of
the
tables
of
contents
of
recent
issues
of
resource
economics
journals
(
e.
g.
Land
Economics,
Marine
Resource
Economics,
Journal
of
Environmental
Economics
and
Management,
etc.);
<
Searches
of
online
reference
and
abstract
databases
(
e.
g.
Environmental
Valuation
Resource
Inventory
(
EVRI),
Benefits
Use
Valuation
Database
(
BUVD),
AgEcon
Search,
etc.);
<
Visits
to
homepages
of
authors
known
to
have
published
contingent
valuation
studies
and
or
water
quality
research;
<
Searches
of
Web
sites
of
agricultural
and
resource
economics
departments
at
several
colleges
and
universities;
<
Searches
of
Web
sites
of
organizations
and
agencies
known
to
publish
environmental
and
resource
economics
valuation
research
(
e.
g.,
Resources
for
the
Future
(
RFF),
National
Center
for
Environmental
Economics
(
NCEE),
Natural
Resource
Conservation
Service
(
NRCS),
National
Bureau
of
Economic
Research
(
NBER),
etc.).

Based
on
this
review,
EPA
identified
approximately
250
surface
water
valuation
studies
that
are
potentially
relevant
for
this
analysis,
and
compiled
a
bibliographic
database
to
organize
the
literature
review
process.
Eighteen
of
these
studies
met
the
three
specific
criteria
identified
for
inclusion
in
a
future
meta­
analysis,
detailed
below:

<
Specific
Amenity
Valued:
Environmental
quality
change
being
valued
affects
aquatic
life
and
or
habitat
in
a
water
body
that
provides
recreational
fishing
uses;
hypothetical
improvement
(
degradation)
results
from
a
change
in
water
quality
or
from
a
program
that
affects
recreational
fishing
(
e.
g.,
improvements
in
recreational
catch
rates
or
changes
in
aquatic
use
designation);
<
Non­
Use
Value:
Study
provides
isolated
estimates
of
both
use
and
non­
use
value
or
allows
estimation
of
non­
use
and
use
values
from
the
data
provided;
<
U.
S.
Studies:
Selected
studies
were
limited
to
those
that
surveyed
U.
S.
populations
in
order
to
value
domestic
resources;
<
Research
Methods:
Study
uses
research
methods
supported
by
journal
literature.

­
3­
March
12,
2003
The
Agency
obtained
the
full
text
articles
of
those
studies
that
seemed
most
relevant
for
benefits
transfer
and
compiled
key
information
from
the
selected
studies.
Appendix
A
contains
a
compendium
of
selected
studies
that
will
support
EPA s
estimation
of
the
non­
use
value
of
aquatic
habitat
improvements.
EPA
compiled
the
following
information
from
the
selected
studies:

<
full
study
citation
<
study
location
<
affected
water
body
type
<
environmental
quality
changes
(
i.
e.,
50%
increase
in
catch
rates
or
water
quality
change
from
boatable
to
fishable)
<
study
values
updated
to
2002
dollars
2.0
Value
Concepts
and
Valuation
Techniques
The
basic
approach
for
estimating
the
benefits
of
a
policy
event
is
to
evaluate
the
changes
in
consumer
and
producer
surplus,
as
an
approximation
of
the
change
in
social
welfare.
In
the
case
of
nonmarket
goods,
such
as
aquatic
habitat,
only
consumer
surplus
is
measured.
The
increase
in
consumer
surplus
due
to
an
improvement
in
environmental
quality
is
measured
as
the
change
in
the
area
under
the
shifted
aggregate
demand
curve,
given
an
increase
in
the
quality
of
a
commodity.
The
consumer
surplus
value
of
environmental
changes
is
estimated
using
non­
market
valuation
methods.

Values
associated
with
public
goods,
such
as
aquatic
habitat
quality,
are
generally
divided
among
three
categories:

1.
use
value
2.
non­
use
value
3.
option
value
Use
value
reflects
the
value
of
all
current
direct
and
indirect
physical
uses
of
a
good
or
service
(
Mitchell
and
Carson,
1989).
Use
value
of
surface
water
resources
is
largely
derived
from
recreational
uses,
which
include
activities
such
as
fishing,
boating,
swimming.
These
uses
may
also
include
such
 
terrestrial 
activities
as
hiking
and
picnicking,
if
the
value
of
these
activities
is
at
least
partially
influenced
by
the
quality
or
existence
of
nearby
surface
water.

Non­
use
value,
often
referred
to
as
passive
use
value,
is
based
on
the
concept
that
individuals
may
obtain
utility
without
actually
using
a
particular
resource.
Non­
use
value
consists
of
existence
and
bequest
values.
Existence
value
is
the
value
that
individuals
may
hold
for
simply
knowing
that
a
particular
good
exists
regardless
of
their
present
or
expected
use.
Bequest
value
exists
when
someone
gains
utility
­
4­
March
12,
2003
through
the
knowledge
that
an
amenity
will
be
available
for
others
(
family
or
future
generations)
in
the
future
(
Fisher
and
Raucher,
1984).
2
Option
value
is
often
considered
a
third
type
of
value,
apart
from
both
the
use
and
non­
use
components
of
total
value,
and
is
held
by
individuals
who
expect
to
use
a
resource
in
the
future.
Fisher
and
Raucher
(
1984)
define
option
price
for
such
an
individual
as
 
the
sum
of
the
expected
value
of
consumer
surplus
from
using
the
resource
plus
an
option
value
or
risk
premium
that
accounts
for
uncertainty
in
demand
or
in
supply. 
Mitchell
and
Carson
(
1989)
argue
that
on
theoretical
grounds
this
risk
premium
should
be
small
for
non­
unique
resources.
Option
value,
which
may
exist
regardless
of
actual
future
use,
has
been
classified
as
both
non­
use
value
and
use
value
in
older
published
studies.
Recent
years,
however,
have
seen
a
shift
toward
including
the
willingness
to
pay
(
WTP)
to
secure
an
individual s
option
or
access
to
use
a
resource
in
the
use
category
of
value.
Following
Brown s
(
1993)
methodology,
EPA
treats
option
value
as
a
component
of
total
use
value
in
the
following
analysis
where
specific
or
combined
estimates
are
available.

Quasi­
option
value
is
the
benefit
associated
with
delaying
a
decision
when
there
is
uncertainty
about
the
payoffs
of
various
policy
choices
and
when
at
least
one
the
choices
involves
the
irreversible
commitment
of
resources
(
Freeman,
1993).
This
value
does
not
relate
to
the
values
(
e.
g.,
option
value)
that
individuals
attach
to
a
specific
resource.
It
is
a
social
gain
associated
with
improved
decision­
making
strategies
under
uncertainty.

A
variety
of
non­
market
valuation
methods
exist
for
estimating
use
value,
including
both
 
revealed 
and
 
stated 
preference
methods
(
Freeman,
1993).
Where
appropriate
data
are
available
or
may
be
collected,
revealed
preference
methods
typically
represent
the
preferred
set
of
methods
for
estimating
use
values.
These
methods
use
actual
observed
behavior
to
infer
users 
value
for
environmental
goods
and
services.
Examples
of
revealed
preference
methods
include
travel
cost
and
hedonic
pricing.
Compared
to
non­
use
values,
use
values
are
often
considered
relatively
easy
to
estimate,
due
to
their
relationship
to
observable
behavior,
the
variety
of
revealed
preference
methods
available,
and
public
familiarity
with
the
recreational
services
provided
by
surface
water
bodies.

In
contrast,
non­
use
values
are
often
considered
more
difficult
to
estimate.
Stated
preference
methods,
or
benefits
transfers
based
on
stated
preference
studies,
are
the
only
generally
accepted
techniques
for
estimating
these
values.
Stated
preference
methods
rely
on
carefully
designed
surveys,
which
either
(
1)
ask
people
to
state
their
willingness
to
pay
for
particular
ecological
improvements,
such
as
increased
protection
of
aquatic
species
or
habitats
with
particular
attributes,
or
(
2)
ask
people
to
choose
between
2The
term
 
existence
value 
is
sometimes
used
interchangeably
with
or
in
place
of
 
non­
use
value. 
In
this
case,
where
the
whole
of
non­
use
benefits
is
represented,
existence
value
has
been
described
as
including
vicarious
consumption
and
stewardship
values.
Vicarious
consumption
reflects
the
value
individuals
may
place
on
the
availability
of
a
good
or
service
for
others
to
consume
in
the
current
time
period,
and
stewardship
includes
inherent
value
as
well
as
bequest
value.
In
this
case
inherent
value
may
be
considered
the
existence
value
individuals
hold
for
knowing
that
a
good
exists
(
described
above),
and
bequest
value
is
the
value
individuals
place
on
preserving
or
ensuring
the
availability
of
a
good
or
service
for
family
and
others
in
the
future.

­
5­
March
12,
2003
competing
hypothetical
 
packages 
of
ecological
improvements
and
household
cost.
In
either
case,
analysis
of
survey
responses
allows
estimation
of
values.

Non­
use
values
may
be
more
difficult
to
assess
than
use
values
for
several
reasons.
First,
non­
use
values
are
not
associated
with
easily
observable
behaviors.
Second,
non­
use
values
may
be
held
by
both
users
and
non­
users
of
a
resource,
and
non­
users
may
be
less
familiar
with
particular
services
provided
by
water
resources.
Third,
the
development
of
a
defensible
stated
preference
survey
that
meets
the
NOAA
blue
ribbon
panel
requirements
is
often
a
time
and
resource
intensive
process.
3
Fourth,
even
carefully
designed
surveys
may
be
subject
to
certain
biases
associated
with
the
hypothetical
nature
of
survey
responses
(
Mitchell
and
Carson,
1989).

Given
EPA s
regulatory
schedule,
developing
and
implementing
stated
preference
surveys
to
elicit
total
value
(
i.
e.,
non­
use
and
use
)
of
environmental
quality
changes
resulting
from
environmental
regulations
is
often
not
feasible.
However,
the
Agency
routinely
estimates
changes
in
use
values
of
the
affected
resources
as
part
of
regulatory
development.
The
following
analysis
develops
a
method
for
obtaining
an
approximate
estimate
of
non­
use
and
thus
total
value
of
aquatic
habitat
improvements
when
only
use
values
are
estimated.

3.0
Description
of
Studies
EPA
identified
18
surface
water
valuation
studies
using
stated
preference
surveys
that
allow
estimation
of
non­
use
values
associated
with
aquatic
habitat
improvements.
4,5
These
18
studies,
listed
in
Table
3.1,
include
10
journal
articles,
six
reports,
one
book,
and
one
unpublished
Ph.
D.
dissertation.
The
publication
year
of
these
studies
ranges
from
1978
to
2000.
Two
studies
(
Whitehead
et
al.,
1995
and
3In
1992,
the
National
Oceanic
and
Atmospheric
Administration
convened
a
panel
of
economic
and
survey
research
experts,
who
had
no
vested
interest
in
stated
preference
methods,
to
conduct
hearings
on
the
validity
of
the
contingent
valuation
(
CV)
method
(
form
of
stated
preference)
(
Arrow
et
al.
1993;
see
also
FR
58:
19,
4601­
14,
1993).
This
panel
issued
proposed
guidelines,
consisting
of
a
number
of
recommendations
about
survey
design
and
implementation,
"
compliance
with
which
would
define
an
ideal
CV
survey."
The
panel's
general
guidelines
address
the
following
issues:
sample
type
and
size;
minimizing
non­
responses;
use
of
personal
interviews;
pretesting
for
interviewer
effects;
reporting;
careful
pretesting
of
a
CV
questionnaire;
conservative
design;
elicitation
format;
referendum
format;
accurate
description
of
the
program
or
policy;
pretesting
of
photographs;
reminder
of
undamaged
substitute
commodities;
adequate
time
lapse
from
the
accident;
temporal
averaging;
"
no­
answer"
option;
yes/
no
follow­
ups;
cross­
tabulations;
checks
on
understanding
and
acceptance;
alternative
expenditure
possibilities;
deflection
of
transaction
value;
steady
state
or
interim
losses;
present
value
calculations
of
interim
losses;
advance
approval;
burden
of
proof;
and
reliable
reference
surveys.

4
Farber
and
Griner
(
2000)
conducted
a
conjoint
analysis
study
that
estimated
use
and
non­
use
values
for
water
quality
improvements,
but
the
Agency
did
not
include
this
study
in
the
analysis
of
use
and
non­
use
values
because
non­
use
WTP
estimates
were
not
statistically
significant.

5The
Agency
also
excluded
a
study
by
Bockstael
et
al.
(
1989)
that
estimated
total
use
and
non­
use
value
of
enhanced
swimming
opportunities
due
to
improved
water
quality
in
Chesapeake
Bay.
EPA
excluded
this
study
from
the
analysis
of
use
and
non­
use
values
presented
in
this
report
because
the
study
did
not
consider
water
quality
effects
on
fishery
resources.

­
6­
March
12,
2003
Whitehead
and
Groothuis,
1992)
had
the
same
primary
author
and
three
studies
(
Sanders,
Walsh,
and
Loomis,
1990;
Sutherland
and
Walsh,
1985;
and
Walsh
et
al.,
1978)
had
a
common
author.

Table
3.1
presents
the
characteristics
of
each
study,
including
survey
administration
information
and
the
type
of
water
body
associated
with
the
environmental
change
or
amenity
being
valued.
All
selected
studies
focus
on
environmental
quality
changes
that
affect
surface
water
resources.
All
surveys
occurred
in
the
contiguous
U.
S.,
with
surveyed
populations
representing
14
states
and
all
U.
S.
Census
Bureau
regions.
These
studies,
however,
vary
in
several
respects,
including
the
specific
environmental
change
valued,
the
types
of
values
estimated,
the
geographic
region
affected
by
environmental
changes,
the
scale
of
environmental
improvement,
and
survey
administration
methods.

Ten
studies
focused
on
rivers,
three
dealt
with
estuaries,
one
study
focused
on
wetlands,
and
one
study
valued
a
quality
change
involving
a
river­
lake
system.
In
addition,
three
studies
analyzed
WTP
for
environmental
quality
changes
concerning
all
freshwater
in
a
given
region.

­
7­
March
12,
2003
Table
3.1
ected
for
Non­
Use
Benefits
Meta­
Analysis
Authors
(
year)
Admin
Method
Response
Rate
State(
s)
/

Region
Water
body
Type
Elicitation
Method
Method
for
Estimating
Non­
Use
Value
Clonts
&
Malone
(
1990)
phone
unknown
AL
river
iterative
bidding
value
categories
­
user
/
non­
user
Croke
et
al.
(
1986­
87)
phone
85%
IL
river
iterative
bidding
user
/
non­
user
total
values
Cronin
(
1982)
personal
interview
82%
DC
river
open
ended
user
/
non­
user
total
values
Desvousges
et
al.
(
1983)
personal
interview
81%
PA
river
payment
card
assume
non­
use
­
value
categories
Huang
et
al.
(
1997)
phone
75%
NC
estuary
dichotomous
choice
revealed
&
stated
preference
data
Kaoru
(
1993)
mail
49%
MA
estuary
open
ended
allocation
­
entire
sample
Lant
&
Roberts
(
1990)
personal
interview
41%
IA,
IL
river
payment
card
value
categories
­
entire
sample
Magat
et
al.
(
2000)
personal
interview
unknown
CO,
NC
all
freshwater
iterative
bidding
assume
non­
use
Mitchell
&
Carson
(
1981)
personal
interview
73%
National
all
freshwater
payment
card
user
/
non­
user
total
values
Olsen
et
al.
(
1991)
phone
73%
ID,
MT,

OR,
WA
river
open
ended
user
/
non­
user
total
values
Roberts
&
Leitch
(
1997)
mail
62%
MN,
SD
wetland
payment
card
value
categories
­
entire
sample
Rowe
et
al.
(
1985)
mail
50%
CO
river
open
ended
revealed
&
stated
preference
data
and
allocation
­
entire
sample
Sanders
et
al.
(
1990)
mail
51%
CO
river
open
ended
allocation
­
entire
sample
Sutherland
&
Walsh
(
1985)
mail
61%
MT
river
&
lake
open
ended
allocation
­
entire
sample
Walsh
et
al.
(
1978)
personal
interview
unknown
CO
river
iterative
bidding
assume
non­
use
­
value
categories
Welle
(
1986)
mail
76%
MN
all
freshwater
open
ended
value
categories
­
user
/
non­
user
Whitehead
&
Groothuis
(
1992)
mail
61%
NC
river
open
ended
user
/
non­
user
total
values
Whitehead
et
al.
(
1995)
phone
71%
NC
estuary
iterative
bidding
user
/
non­
user
total
values
Characteristics
of
Studies
Sel
a.
Response
rates
are
unknown
for
Clonts
and
Malone
(
1990),
Magat
et
al.
(
2000),
and
Walsh
(
1978).

­
8­
March
12,
2003
Surveys
in
seven
studies
were
administered
by
mail;
six
studies
gathered
information
through
personal
interviews
in
the
home,
on­
site,
or
in
a
centralized
location;
and
five
surveys
were
conducted
by
telephone.
Survey
response
rates
ranged
from
27
to
85
percent,
with
a
median
rate
of
66
percent.

The
two
most
common
methods
for
eliciting
WTP
values
were
the
open­
ended
response,
used
in
eight
studies.
and
the
iterative
bidding
method,
used
in
five
studies.
Four
studies
used
the
payment
card
approach,
and
one
used
the
dichotomous
choice
method.
Most
surveys
used
higher
taxes
and/
or
prices
as
the
payment
vehicle
for
funding
the
specified
quality
change,
with
voluntary
contributions
and
earmarked
funds
the
next
most
common
methods.

The
18
valuation
studies
translate
into
27
observations
of
non­
use
and
use
values
associated
with
various
aquatic
habitat
improvements,
because
five
studies
generated
more
than
one
usable
non­
use
value
estimate.
Multiple
estimates
of
non­
use
and
use
value
from
a
single
study
are
available
due
to
the
following
variations
in
study
design:

<
The
quantity
of
the
amenity
provided
(
Sanders,
1990);
<
The
level
of
the
quality
change
(
Croke
et
al.,
1986;
Lant
and
Roberts,
1990);
<
The
technique
employed
to
estimate
non­
use
value
(
Huang
et
al.
1997;
Rowe
et
al.
1985).

Appendix
A
presents
key
information
from
each
study,
including
characteristics
of
the
resource(
s)
valued,
the
geographic
scale
of
resource
improvements,
elicitation
method,
estimated
use
and
non­
use
values,
and
other
specifics
of
each
study.

4.0
Methods
of
Measuring
Non­
Use
Value
EPA s
procedure
for
looking
at
non­
use
and
use
values
associated
with
water
resources
is
based
on
the
approach
developed
by
Brown
(
1993).
Based
on
this
approach,
the
Agency
first
isolated
non­
use
and
use
values
from
each
study
as
follows:

 
Non­
use
values
were
isolated
in
the
studies
based
on
the
following
five
methods:
(
1)
total
WTP
for
nonuser
sub­
samples
of
respondents;
(
2)
a
separate
non­
use
value
question
in
the
survey,
(
3)
apportionment
of
total
WTP
among
categories
of
value
(
e.
g.,
use,
option,
existence,
bequest)
by
survey
respondents,
(
4)
total
WTP
of
a
survey
sub­
sample
who
were
asked
to
assume
they
would
not
use
the
resource
being
valued;
and
(
5)
total
WTP
of
the
sample
of
users
minus
WTP
for
the
direct
use
of
the
resource
estimated
based
on
revealed
preference
data.

 
Corresponding
use
values
were
obtained
using
similar
methods:
(
1)
total
WTP
of
the
sample
of
users
minus
total
WTP
of
the
sample
of
non­
users,
(
2)
a
separate
use
value
question,
(
3)
apportionment
by
the
respondent,
(
4)
total
WTP
of
the
sample
of
actual
or
potential
users
minus
total
WTP
of
the
sample
asked
to
assume
zero
use,
(
5)
recreational
use
value
estimated
based
on
revealed
preference
data.

Table
4.1
presents
these
methods
and
their
frequency
across
the
meta­
analysis
studies.
A
detailed
description
of
each
method
is
provided
below.
Several
studies
included
multiple
valuation
methods/
questions
and/
or
divided
samples
into
various
groups
based
on
recreational
use
or
other
­
9­
March
12,
2003
characteristics.
EPA
selected
the
most
appropriate
method
for
estimating
non­
use
value
for
this
analysis
from
each
study.

Table
4.1
Method
Number
Non­
Use
Method
Number
of
Studiesa
Number
of
Studies
with
User
Non­
Use
Value
1
Total
WTP
Values
for
User
and
6
­
2
WTP
for
Value
Components
­
Separate
Questionsb
4
3
WTP
for
Value
Components
­
Allocation
4
­
4
Assume
Non­
Use
of
Resource
3
2
5
Revealed
&
Stated
Preference
Techniques
2
­
Total
19
4
Methods
of
Isolating
Non­
Use
Value
Non­
User
Groups
2
a.
Rowe
et
al.
used
several
valuation
methods,
resulting
in
two
methods
of
isolating
non­
use
value
that
are
used
in
this
analysis:
combination
of
revealed
and
state
preference
techniques,
and
allocation
of
WTP
among
value
categories.
Therefore,
EPA
lists
a
total
of
19
instances
of
isolating
non­
use
value.
b.
In
addition
to
asking
separate
questions
to
elicit
use
and
non­
use
values,
two
studies
using
this
method
also
distinguished
between
users
and
non­
users
of
the
resources.
Both
studies
estimated
non­
use
value
for
users.
This
approach
is
a
combination
of
methods
1
and
2.

Method
1:
Total
WTP
Values
for
User
and
Non­
User
Groups
Method
1,
used
in
six
surveys,
is
the
most
common
method
of
isolating
non­
use
value
among
stated
preference
studies.
This
method
estimates
total
WTP
for
user
and
non­
user
sub­
samples
of
respondents,
with
definitions
of
 
user 
and
 
non­
user 
varying
across
studies.
As
Freeman
(
1993)
notes:
 
By
definition
non­
users
of
a
resource
can
hold
only
non­
use
values,
but
users
may
hold
both
use
and
non­
use
value
for
a
resource. 
Respondents
were
generally
characterized
as
non­
users
of
the
resource
if
the
respondent
(
1)
did
not
use
the
resource
during
some
defined
historical
period,
or
(
2)
does
not
expect
to
use
the
resource
during
some
defined
future
period.
Studies
included
in
this
analysis
used
a
range
of
definitions
to
distinguish
between
users
and
non­
users
with
various
levels
of
stringency
for
determining
that
a
respondent
does
not
have
a
history
of
resource
use,
and/
or
will
not
become
a
user
in
the
future.

Olsen
et
al.
(
1991)
defined
resource
users
as
individuals
who
have
been
involved
in
the
commercial
fishing
industry
or
who
have
participated
in
sport
fishing
during
the
past
two
years,
and
non­
users
as
those
who
had
not
or
were
uncertain
of
their
participation.
They
also
defined
a
third
category
of
respondents
as
those
who
stated
they
did
not
expect
to
fish
in
the
next
five
years.
These
individuals
were
considered
to
have
some
probability
of
future
use.
6
Whitehead
et
al.
(
1992)
and
Whitehead
and
Groothuis
(
1992)
used
a
more
explicit
definition
of
future
use
by
defining
a
user
as
a
respondent
who
stated
that
he
or
she
would
use
the
resource
for
fishing
after
the
specified
environmental
improvement,
and
non­
users
as
those
who
would
not.
In
theory,
this
method
accounts
for
ex
ante
non­
users
who
become
users
following
the
improvement.

6This
analysis
does
not
use
the
WTP
for
these
 
non­
users 
who
may
have
a
potential
future
use
in
order
to
be
consistent
with
the
two­
group
method
of
developing
non­
use
to
use
WTP
ratios.

­
10­
March
12,
2003
Cronin
(
1982)
applied
a
rather
strict
definition
of
a
non­
user
by
considering
a
respondent
a
user
if
he
or
she
or
any
member
of
his
or
her
household
uses
the
river
in
question,
the
seashore,
or
any
other
river,
lake,
or
local
pool,
for
swimming,
boating,
fishing,
hiking,
camping,
or
picnicking.

Mitchell
and
Carson
(
1981)
also
employed
a
narrow
definition
of
non­
users
by
considering
any
respondent
who
had
boated,
fished,
or
gone
swimming
within
the
past
two
years
to
be
a
resource
user.

Croke
et
al.
(
1986)
questioned
respondents
on
their
frequency
of
river
use
and
the
types
of
activities
in
which
they
had
engaged.
Respondents
who
currently
use
the
rivers
for
outing,
boating,
or
fishing
were
identified
as
users.
7
The
Agency
used
best
professional
judgement
to
select
the
most
consistent
definitions
of
user
and
nonuser
groups
across
the
relevant
studies
to
approximate
use
and
non­
use
value.
EPA
notes
that
this
method
for
estimating
non­
use
may
underestimate
non­
use
values
for
users
of
aquatic
resources
(
Whitehead
and
Blomquist,
1991).
As
shown
in
Tables
5.1
and
5.2,
the
non­
use
values
for
users
tend
to
be
significantly
higher
compared
to
the
non­
use
values
held
by
non­
users
of
these
resources.

Method
2:
WTP
for
Value
Components
­
Separate
Questions
Four
studies
used
method
2,
in
which
respondents
reported
separate
WTP
bids
for
different
components
of
total
value.
Each
respondent
usually
answered
a
set
of
valuation
questions,
designed
to
estimate
value
for
separate
components
of
total
value
(
i.
e.,
use,
option,
existence,
bequest).
Lant
and
Roberts
(
1990)
obtained
bids
from
each
respondent
for
both
recreational
value
and
 
intrinsic 
value.
8
For
the
purposes
of
this
analysis,
EPA
interpreted
these
values
as
those
for
total
use
and
total
non­
use,
respectively
(
see
section
5
for
further
detail).
The
fundamental
difference
between
this
method
of
isolating
non­
use
value
and
the
user/
non­
user
method
(
method
1)
is
that,
in
this
case,
users 
non­
use
value
is
captured
in
intrinsic
value,
and
non­
users 
option
value
is
captured
in
recreational
(
use)
value,
although
the
latter
concept
is
arguable
and
dependent
upon
the
framing
of
the
valuation
question.

Roberts
and
Leitch
(
1997)
used
a
similar
method
to
that
employed
by
Lant
and
Roberts
(
1990),
but
elicited
separate
estimates
of
recreation,
option,
and
existence
values.
To
determine
use
value,
respondents
answered
the
question,
 
If
Mud
Lake
was
managed
primarily
for
water­
related
recreation
and
fish/
wildlife
habitat,
what
would
you
be
willing
to
pay
through
an
annual
use
permit
to
participate
in
recreational
activities
at
Mud
Lake? 
The
authors
intended
the
following
valuation
question
to
measure
existence
value:
 
What
is
the
maximum
amount
you
would
be
willing
to
pay
through
an
annual
voluntary
donation
to
ensure
that
recreational
activities
and
fish/
wildlife
habitat
at
Mud
Lake
are
available
for
other
people,
even
if
you
do
not
intend
to
visit
the
Mud
Lake
area? 

7Only
about
10
percent
were
users.

8Lant
and
Roberts
define
'
intrinsic
values'
as
the
nonrecreational
values
of
local
water
resources
(
i.
e.,
based
on
their
definition
intrinsic
values
equal
to
non­
use
values).

­
11­
March
12,
2003
Stated
preference
surveys
that
use
method
2
to
isolate
non­
use
value
may
also
distinguish
between
users
and
non­
users
to
obtain
estimates
of
the
different
value
components
for
both
groups.
This
approach
combines
the
approaches
taken
in
methods
1
and
2.
Welle
(
1986)
represents
one
of
two
such
studies
in
this
analysis
that
use
this
approach.
Welle
conducted
a
survey
that
characterized
respondents
by
their
certainty
of
future
use,
to
distinguish
between
non­
users
and
users.
Welle
then
estimated
existence
values
for
both
population
groups,
enabling
a
comparison
of
non­
users 
and
users 
non­
use
value.

Clonts
and
Malone
(
1990)
also
combined
methods
1
and
2,
using
separate
bidding
procedures
to
estimate
values
for
recreational
use,
option,
existence,
and
bequest
values.
Respondents
were
asked
three
times
to
verify
the
individual
and
summed
values.
Respondents
were
also
defined
as
members
of
user
or
non­
user
households,
where
user
households
had
at
least
one
member
who
had
visited
a
relevant
river
for
the
purpose
of
outdoor
recreation
within
the
past
three
years.

The
main
limitation
of
method
2
is
that
users
may
include
their
personal
use
values
in
non­
use
values,
which
could
potentially
result
in
double
counting
of
use
values
(
Silberman
et
al.
1992).
The
Agency,
however,
notes
that
non­
use
values
for
users
are
not
included
in
the
regression­
based
meta­
analysis
presented
in
Section
6.0
to
prevent
an
upward
bias
in
coefficient
estimates.

Method
3:
WTP
for
Value
Components
­
Allocation
With
method
3,
used
in
four
studies,
respondents
reported
their
total
WTP
and
were
then
asked
to
allocate
this
value
among
categories
of
value
(
e.
g.,
use,
option,
existence,
bequest).
Sanders
et
al.
(
1990)
asked
respondents
to
state
their
total
WTP
for
river
protection,
and
then
asked
them
to
allocate
that
value
across
four
categories.
The
researchers
used
the
following
four­
part
question
to
capture
recreational,
option,
existence,
and
bequest
values,
respectively:

People
value
the
protection
of
rivers
for
several
purposes.
What
proportion
(
percent
of
100)
of
the
highest
dollar
value
you
reported
above
would
you
assign
to
each
of
the
following
purposes?
Read
the
entire
question
first,
then
answer
each
of
the
four
parts
together,
they
should
total
100
percent.

a.
Payment
to
actually
visit
these
rivers
for
recreation
use.
________%

b.
In
addition
to
your
actual
recreation
use
value,
how
much
of
an
 
insurance
premium 
would
you
be
willing
to
pay
each
year
to
guarantee
your
choice
of
recreation
use
of
these
rivers
in
the
future?
________%

c.
Payment
to
protect
these
rivers
for
reason
other
than
your
own
personal
use.

1)
The
value
to
you
from
just
knowing
these
rivers
exist
as
natural
habitats
for
plants,
fish,
wildlife,
etc.
________%

2)
The
value
to
you
from
knowing
that
these
rivers
will
be
protected
for
future
generations.
________%

­
12­
March
12,
2003
Sutherland
and
Walsh
(
1978)
employed
a
similar
method
of
isolating
non­
use
value
where
respondents
stated
total
WTP
and
then
allocated
this
figure
among
the
recreation,
option,
existence,
and
bequest
components
of
total
value.

Kaoru
(
1993)
and
Rowe
et
al.
(
1985)
both
used
comparable
methods
but
did
not
elicit
estimates
for
each
of
the
four
categories
above.
For
example,
Kaoru
(
1993)
estimated
recreational
use,
option,
and
existence
values,
and
Rowe
et
al.
(
1985)
estimated
use,
existence,
and
bequest
values
without
an
explicit
inclusion
of
respondents'
option
value.

Method
4:
Assume
Non­
Use
of
Resource
Method
4,
used
in
three
studies,
asked
respondents
to
assume
they
would
not
use
the
resource
being
valued.
For
example,
Walsh
et
al.
(
1978)
asked
the
following
question
of
each
respondent
to
measure
existence
value:
 
If
it
were
certain
you
would
not
use
the
South
Platte
River
Basin
for
water­
based
recreation,
would
you
be
willing
to
add
____
cents
on
the
dollar
to
present
sales
taxes
every
year,
just
to
know
clean
water
exists
at
level
A
as
a
natural
habitat
for
plants,
fish,
wildlife,
etc.? 
Respondents
answered
similar
questions
to
determine
their
option
and
bequest
values.
This
example
incorporates
the
central
concept
of
method
2
of
asking
separate
questions
to
elicit
different
types
of
value.
The
key
concept
distinguishing
method
4
is
that
respondents
must
assume
non­
use
of
the
resource
regardless
of
their
true
user
status.
Magat
et
al.
(
2000)
also
asked
respondents
to
assume
they
would
not
use
the
resource,
but
WTP
was
only
elicited
for
total
non­
use
value.
Desvousges
et
al.
(
1983)
estimated
WTP
values
using
a
combination
of
methods.
In
order
to
estimate
use
and
option
values,
respondents
apportioned
their
option
price
bid
between
present
and
expected
use,
respectively.
The
authors
used
a
separate
question
to
obtain
existence
values,
where
respondents
were
asked
to
assume
they
would
never
use
the
river.

Method
5:
Revealed
&
Stated
Preference
Techniques
Method
5,
used
in
two
studies,
is
distinct
in
that
it
combines
revealed
and
stated
preference
data
to
estimate
the
non­
use
value
of
a
resource.
The
main
advantage
of
this
method
is
that
revealed
preference
data
can
provide
reliable
estimates
of
recreational
use
value(
s).
Huang
et
al.
(
1997)
combined
revealed
and
stated
preference
data
from
a
single
survey
to
estimate
non­
use
value
without
explicitly
posing
a
question
concerning
potential
non­
use
values
associated
with
the
resource.
Survey
questions
determined
recreation
participation;
the
authors
compared
the
recreational
value
from
the
resulting
travel
cost
model
to
an
estimate
of
total
consumer
surplus
from
the
Stated
Preference
portion
of
the
survey
to
estimate
non­
use
value.
Only
one
other
study
(
Rowe
et
al.,
1985)
included
in
this
meta­
analysis
combined
revealed
and
stated
preference
data,
but
this
approach
to
estimating
non­
market
values
is
becoming
more
widely
used.

5.0
Summary
of
Non­
Use
and
Use
Values
from
Surface
Water
Valuation
Studies
As
noted
above,
the
primary
purpose
of
this
analysis
is
to
identify
a
set
of
new
studies
that
may
contain
information
about
the
relative
magnitude
of
use
and
non­
use
values
for
environmental
changes
that
affect
aquatic
habitat
and/
or
water
quality.

­
13­
March
12,
2003
Table
5.1
presents
the
27
observations
of
non­
use
and
use
value
estimates
from
the
18
studies
used
in
EPA s
analysis
to
calculate
the
mean
and
median
non­
use
and
use
values.
To
calculate
the
mean
and
median
of
ratios
of
non­
use
to
use
values,
EPA
first
assigned
a
weight
to
each
observation
in
Table
5.1.
Weights
are
defined
such
that
each
unique
study
is
given
identical
weight
(
i.
e.,
weights
on
multiple
observations
within
each
study
sum
to
one;
for
studies
with
only
one
observation,
the
weight
is
equal
to
one).
The
mean
and
median
water
resource
use
values
based
on
the
18
surface
water
valuation
studies
are
$
68.61
and
$
33.33
(
2002$)
per
household
per
year.
The
mean
and
median
non­
use
values
of
the
water
resources
are
$
79.51
and
$
74.41
(
2002$)
per
household
per
year,
respectively.
EPA s
analysis
of
non­
use
and
use
value
presented
in
Table
5.1
shows
that
non­
use
values
for
water
resources
are
substantial.

­
14­
March
12,
2003
­
15­
Table
5.1
Total
Non­
Use
and
Use
Values
(
per
Household,
2002$)
a
Author
Survey
Year
Non­
Use
Valueb
Mean
Use
Valuec
Clonts
&
Malone
(
1990)
1987
$
55.43
$
30.88
Croke
et
al.
(
1986)
1986
$
53.31
$
17.61
Croke
et
al.
(
1986)
1986
$
60.91
$
11.11
Croke
et
al.
(
1986)
1986
$
75.11
$
6.35
Cronin
(
1982)
1973
$
59.37
$
23.75
Desvousges
et
al.
(
1983)
1981
$
111.41
$
89.65
Huang
et
al.
(
1997)
1995
$
59.02
$
158.25
Huang
et
al.
(
1997)
1995
$
119.52
$
97.75
Kaoru
(
1993)
1989
$
112.57
$
77.04
Lant
&
Roberts
(
1990)
1987
$
68.56
$
57.30
Lant
&
Roberts
(
1990)
1987
$
88.40
$
77.04
Lant
&
Roberts
(
1990)
1987
$
85.29
$
78.34
Magat
et
al.
(
2000)
1997
$
11.49
$
22.98
Mitchell
&
Carson
(
1981)
1980
$
242.34
$
275.09
Olsen
et
al.
(
1991)
1989
$
38.48
$
69.12
Roberts
&
Leitch
(
1997)
1996
$
2.09
$
5.17
Rowe
et
al.
(
1985)
1985
$
69.15
$
47.88
Rowe
et
al.
(
1985)
1985
$
12.46
$
6.54
Sanders
et
al.
(
1990)
1983
$
46.91
$
24.96
Sanders
et
al.
(
1990)
1983
$
87.60
$
46.62
Sanders
et
al.
(
1990)
1983
$
111.99
$
59.61
Sanders
et
al.
(
1990)
1983
$
115.35
$
63.45
Sutherland
&
Walsh
(
1985)
1981
$
91.53
$
35.78
Walsh
et
al.
(
1978)
1976
$
132.63
$
250.66
Welle
(
1986)
1985
$
113.69
$
20.06
Whitehead
&
Groothuis
(
1992)
1991
$
27.74
$
18.49
Whitehead
et
al.
(
1995)
1990
$
68.08
$
29.83
Median
Valued,
e
$
74.41
$
33.33
Mean
Valued,
e
$
79.51
$
68.61
a.
All
values
are
adjusted
to
2002$
using
the
Consumer
Price
Index
(
U.
S.
Bureau
of
Labor
Statistics
2002).
b.
Represents
the
mean
of
non­
users
total
value,
including
option
value
in
some
cases.
c.
Represents
users'
mean
total
value
minus
non­
users'
mean
non­
use
value
for
8
observations.
d.
EPA
calculated
the
mean
and
median
non­
use
and
use
values
by
assigning
a
weight
to
each
observation
based
on
the
number
of
values
taken
from
each
study.
For
example,
use
values
from
a
study
that
provided
three
total
use
WTP
were
each
assigned
a
weight
of
1/
3.
The
number
of
values
used
to
calculate
the
mean
non­
use
and
use
values
for
the
sample
was
the
sum
of
the
weights.

Similarly,
EPA
calculated
mean
and
median
non­
use
and
use
values
for
users
using
information
from
four
studies
that
provide
separate
non­
use
values
for
users.
The
mean
and
median
water
resource
use
values
based
on
these
four
studies
are
$
97.81
and
$
60.27
(
2002$)
per
household
per
year.
The
mean
and
median
non­
use
values
for
users
are
$
155.83
and
$
164.78
(
2002$)
per
household
per
year,
respectively.
As
shown
in
Table
5.2,
the
non­
use
values
for
users
of
aquatic
resources
are
significantly
higher
than
the
non­
use
values
for
non­
users.
This
may
result
from
additional
information
about
water
resources
associated
with
past
or
expected
future
use,
which
is
likely
to
enhance
non­
use
value
(
Whitehead
and
Blomquist,
1991).
March
12,
2003
­
16­
Other
studies
(
e.
g.,
Silberman
et
al.
1992),
however,
suggest
that
users
may
include
their
personal
use
values
in
non­
use
values,
which
could
potentially
result
in
double
counting
of
use
values.

Table
5.2
Total
Use
and
Non­
Use
Values
for
Resource
Users
(
per
Household,
2002$)
a,
b
Author
Survey
Year
Non­
Use
Value
for
Users
Mean
Use
Value
Clonts
&
Malone
(
1990)
1987
$
82.35
$
30.88
Desvousges
et
al.
(
1983)
1981
$
130.59
$
89.65
Walsh
et
al.
(
1978)
1976
$
211.42
$
250.66
Welle
(
1986)
1985
$
198.96
$
20.06
Median
Valuec
$
164.78
$
60.27
Mean
Valuec
$
155.83
$
97.81
a.
All
values
are
adjusted
to
2002$
using
the
Consumer
Price
Index
(
U.
S.
Bureau
of
Labor
Statistics
2002).
b.
All
four
studies
that
provided
estimates
of
users 
non­
use
value
were
published
before
the
publication
of
the
National
Oceanic
and
Atmospheric
Administration
Panel
on
Contingent
Valuation
report.
c.
Calculated
using
empirical
estimates
of
user
non­
use
value
only.

EPA
is
considering
applying
the
results
of
this
review
and
analysis
to
estimate
non­
use
value
for
aquatic
resources
potentially
affected
by
impingement
and
entrainment
for
the
final
rule
analysis.
Specifically,
the
Agency
is
considering
estimating
non­
use
values
that
are
based
on
(
1)
a
percent
or
fraction
of
use
values
per
households
and
(
2)
specific
user
and
nonuser
populations
affected
by
the
316(
b)
regulation.

6.0
Regression­
Based
Meta­
Analysis
EPA
recognizes
that
approach
benefits
transfer
of
non­
use
values
from
existing
studies
requires
careful
accounting
of
factors
that
are
likely
to
affect
non­
use
values
of
aquatic
resources
such
as
the
geographic
scale
of
environmental
improvements,
regional
or
national
importance
of
the
affected
resources,
and
the
magnitude
of
environmental
quality
changes.
In
addition
to
simply
reviewing
available
information
about
the
relative
magnitudes
of
non­
use
and
use
values
EPA
is
also
considering
regression­
based
meta­
analysis
of
non­
use
WTP
for
water
resources.
Glass
(
1976)
characterizes
meta­
analysis
as
 
the
statistical
analysis
of
a
large
collection
of
results
for
individual
studies
for
the
purposes
of
integrating
the
findings.
It
provides
a
rigorous
alternative
to
the
casual,
narrative
discussion
of
research
studies
which
is
commonly
used
to
make
some
sense
of
the
rapidly
expanding
research
literature
(
p.
3;
cited
in
Poe
et
al.
(
2001),
p.
138).

Depending
on
the
suitability
of
available
data,
a
meta­
analysis
can
provide
a
superior
alternative
to
the
calculation
and
use
of
a
simple
arithmetic
mean
over
the
27
observations,
as
it
allows
estimation
of
the
relative
influence
of
various
study,
economic,
and
natural
resource
characteristics
on
non­
use
willingness
to
pay.
The
primary
advantage
of
a
regression
based
approach
is
that
it
may
account
for
differences
among
study
sites
that
may
contribute
to
changes
in
non­
use
values,
to
the
extent
permitted
by
available
data.

Meta­
analysis
is
typically
a
data­
driven
process.
Given
a
reliance
on
information
available
from
the
underlying
studies
that
comprise
the
meta­
data,
meta­
analysis
models
most
often
represent
a
middle
ground
between
model
specifications
that
would
be
most
theoretically
appropriate
and
those
specifications
that
are
possible
given
available
data.
Poe
et
al.
(
2001)
provide
excellent
insight
into
the
March
12,
2003
mechanics
of
specifying
and
estimating
such
meta­
equations.
EPA
intends
to
match
its
general
approach
regression­
based
meta­
analysis,
to
as
great
an
extent
as
possible,
to
that
of
Poe
et
al.
(
2001)
and
others
(
e.
g.,
Bergstrom
et
al.
2001;
Desvousges
et
al.
1998).

The
dependent
variable
in
the
following
meta­
analysis
can
be
either
the
estimated
non­
use
value
or
total
value
(
including
use
and
non­
use
value)
for
water
quality
improvement.
The
total
or
non­
use
values
can
be
modeled
as
a
function
of
independent
variables
characterizing
specifics
of
the
resource(
s)
valued
such
as
whether
they
are
estuarine
or
freshwater,
the
geographic
scale
of
resource
improvements
(
e.
g.,
single
water
body
versus
multiple
water
bodies),
elicitation
method,
estimated
use
and
non­
use
values,
and
other
specifics
of
each
study.
Following
Poe
et
al.
(
2001),
these
variables
may
be
categorized
into
core
economic
variables
and
study
design
effects.
Core
economic
variables
describe
the
resource
policy
and
resulting
resource
change,
resulting
use
values,
the
geographic
scale
over
which
changes
occur,
and
respondent
income.
Study
design
variables
characterize
features
of
the
survey
and
question
format
(
Poe
et
al.
2001).

Core
economic
variables
available
from
the
18
studies
considered
for
this
analysis
include:

<
Estimated
use
values
for
water
quality
improvements,
reported
by
the
same
studies
from
which
non­
use
values
were
estimated.
<
The
geographical
region
of
the
nation
in
which
the
original
study
was
conducted.
The
rationale
behind
this
variable
is
that
non­
use
values
may
differ
systematically
from
region
to
region.
<
The
type
of
water
body
in
which
water
quality
would
be
improved.
These
include
estuaries,
wetlands,
and
lakes/
rivers.
<
The
geographical
scale
over
which
water
quality
would
improve
(
e.
g.,
over
a
single
water
body,
statewide,
etc.)
<
The
estimated
mean
income
of
survey
respondents
(
or
of
the
geographical
region
in
which
the
survey
was
conducted,
if
the
original
survey
did
not
report
income).

Study
design
effects
characterize:

<
The
year
in
which
a
study
was
conducted.
<
The
elicitation
format
of
the
survey
(
e.
g.,
telephone,
mail,
etc.)
<
The
elicitation
method
(
e.
g.,
open
ended
WTP
method).

Based
on
the
regression­
based
meta­
analysis
results,
EPA
could
predict
non­
use
WTP
for
water
resource
changes
as
a
function
of
both
core
economic
and
study
design
attributes.
This,
in
general,
would
provide
a
superior
alternative
to
the
calculation
of
non­
use
benefits,
as
it
allows
WTP
to
be
adjusted
to
account
for
the
characteristics
of
the
transfer
site.

The
Agency
considers
using
the
estimated
regression
model
in
estimating
non­
use
values
of
aquatic
habitat
improvements
from
the
316(
b)
rule.
The
use
(
and
interpretation)
of
the
value
estimates
to
predict
WTP
in
specific
cases
will
follow
the
methodologies
from
the
benefits
transfer
literature
(
e.
g.,
Vandenberg
et
al.
2001;
Desvousges
et
al.,
1998).

EPA
also
notes
potential
limitations
of
this
approach.
Limitations
of
the
regression
analysis
approach
specifically
stem
from
the
number
of
studies
that
meet
criteria
for
inclusion,
the
number
of
variables
that
could
be
included
in
the
regression
analysis
(
which
depends
on
the
number
of
and
information
available
­
17­
March
12,
2003
from
the
original
studies),
as
well
as
degrees
of
freedom
and
statistical
significance.
For
example,
study
differences
often
prevent
the
use
of
a
single
measure
of
the
degree
of
environmental
quality
improvements.
Prior
meta­
analyses
of
this
type,
including
Woodward
and
Wui
(
2001)
and
Poe
et
al.
(
2001),
lack
a
continuous
and
quantified
measure
of
environmental
quality
improvement.
The
use
of
other
economic
variables
that
might
be
desirable
from
a
theoretical
perspective
(
e.
g.,
information
on
substitute
goods)
may
complicate
extraction
of
suitable
data
from
the
underlying
studies.
EPA
also
recognizes
that
clear
and
objective
criteria
are
needed
to
determine
which
studies
are
suitable
for
inclusion
in
meta
analysis;
criteria
should
acknowledge
issues
related
to
potential
bias
associated
with
stated
preference
studies,
and
steps
that
the
researchers
should
take
to
minimize
bias.
One
key
challenge
in
this
analysis
is
to
determine
the
applicability
of
study
results
to
the
policy
case
of
interest
(
i.
e.,
fish
impacts
due
to
impingement
and
entrainment
in
this
rule)
because
of
significant
variations
in
study
objectives
and
methodologies.

­
18­
March
12,
2003
References
Arrow,
Kenneth,
R.
Solow,
P.
R
Portney,
E.
E.
Leamer,
R.
Radner,
and
H.
Schuman.
1993.
Report
of
the
NOAA
Panel
on
Contingent
Valuation.
National
Oceanic
and
Atmospheric
Association,
Damage
Assessment
and
Restoration
Program.

Bergstrom,
J.
C.,
K.
J.
Boyle
and
G.
L.
Poe.
2001.
The
Economic
Value
of
Water
Quality.
Northampton,
MA:
Edward
Elgar.

Bockstael,
N.
E.,
McConnell,
K.
E.,
Strand,
I.
E.
1989.
Measuring
the
Benefits
of
Improvements
in
Water
Quality:
The
Chesapeake
Bay.
Marine
Resource
Economics.
6,
1­
18.

Brown,
Thomas
C.
1993.
Measuring
Nonuse
Value:
A
Comparison
of
Recent
Contingent
Valuation
Studies.
W­
133
Sixth
Interim
Report
University
of
Georgia,
Department
of
Agriculture
and
Applied
Economics.
Athens,
GA.

Clonts,
Howard
A.,
and
Malone,
Joy
W.
1990.
Preservation
attitudes
and
consumer
surplus
in
free
flowing
rivers.
In:
Social
Science
and
Natural
Resource
Recreation
Management,
Joanne
Vining,
editor.
Westview
Press,
Boulder,
CO.
Vol.
pp.
301­
317.

Croke,
Kevin,
Fabian,
Robert
G.,
and
Brenniman,
Gary.
1986­
1987.
Estimating
the
Value
of
Improved
Water
Quality
in
an
Urban
River
System.
Journal
of
Environmental
Systems.
Vol.
16,
No.
1.
pp.
13­
24.

Cronin,
Francis
J.
1982.
Valuing
nonmarket
goods
through
contingent
markets.
Pacific
Northwest
Laboratory,
PNL­
4255,
Richland,
WA.

Desvousges,
William
H.,
Smith,
V.
Kerry,
and
McGivney,
Matthew
P.
1983.
A
Comparison
of
Alternative
Approaches
for
Estimating
Recreation
and
Related
Benefits
of
Water
Quality
Improvements.
U.
S.
Environmental
Protection
Agency,
Economic
Analysis
Division,
Washington,
D.
C.
pp.
345.

Desvousges,
W.
H.,
F.
R.
Johnson,
and
H.
S.
Banzhaf.
1998.
Environmental
Policy
Analysis
with
Limited
Information:
Principles
and
Applications
of
the
Transfer
Method.
Northampton,
MA:
Edward
Elgar
Publishers.

Farber,
Stephen,
and
Griner,
Brian.
2000.
Valuing
Watershed
Quality
Improvements
Using
Conjoint
Analysis.
Ecological
Economics.
Vol.
34,
No.
1.
pp.
63­
76.

Fisher,
A.
and
R.
S.
Raucher.
1984.
 
Intrinsic
Benefits
of
Improved
Water
Quality:
Conceptual
and
Empirical
Perspectives. 
Advances
in
Applied
Microeconomics
Vol.
3,
V.
K.
Smith
(
ed.),
JAI
Press.

Freeman,
A.
Myrick
III.
1993.
 
Nonuse
Values
in
Natural
Resource
Damage
Assessment 
in
Valuing
Natural
Assets,
Raymond
J.
Kopp
and
V.
Kerry
Smith
(
eds).
Resources
for
the
Future,
Washington,
D.
C.

Glass,
G.
V.
1976.
Primary,
Secondary,
and
Meta­
Analysis
of
Research.
Educational
Researcher
5(
10):
3­
8.

­
19­
March
12,
2003
Huang,
Ju­
Chin,
Haab,
Timothy
C.,
and
Whitehead,
John
C.
1997.
Willingness
to
Pay
for
Quality
Improvements:
Should
Revealed
and
Stated
Preference
Data
Be
Combined?
Journal
of
Environmental
Economics
and
Management.
Vol.
34,
No.
3.
pp.
240­
255.

Kaoru,
Yoshiaki.
1993.
Differentiating
Use
and
Nonuse
Values
for
Coastal
Pond
Water
Quality
Improvements.
Environmental
and
Resource
Economics.
Vol.
3.
pp.
487­
494.

Lant,
Christopher
L.,
and
Roberts,
R.
S.
1990.
Greenbelts
in
the
Cornbelt:
Riparian
Wetlands,
Intrinsic
Values,
and
Market
Failure.
Environment
and
Planning.
Vol.
22.
pp.
1375­
1388.

Magat,
Wesley
A.,
Huber,
Joel,
Viscusi,
Kip
W.,
and
Bell,
Jason.
2000.
An
Iterative
Choice
Approach
to
Valuing
Clean
Lakes,
Rivers,
and
Streams.
Journal
of
Risk
and
Uncertainty.
Vol.
21,
No.
1.
pp.
7­
43.

Mitchell,
Robert
Cameron,
and
Carson,
Richard
T.
1981.
An
experiment
in
determining
willingness
to
pay
for
national
water
quality
improvements.
Preliminary
Draft
of
a
report
to
the
U.
S.
Environmental
Protection
Agency.
Resources
for
the
Future,
Inc.,
Washington,
D.
C.

Mitchell,
Robert
Cameron,
and
Carson,
Richard
T.
1989.
Using
Surveys
to
Value
Public
Goods:
The
Contingent
Valuation
Method.
Resources
For
the
Future,
Washington
D.
C.

Olsen,
D.,
Richards,
Jack,
and
Scott,
R.
D.
1991.
Existence
and
Sport
Values
for
Doubling
the
Size
of
Columbia
River
Basin
Salmon
and
Steelhead
Runs.
Rivers.
Vol.
2,
No.
1.
pp.
44­
56.

Poe,
G.
L.,
K.
J.
Boyle,
and
J.
C.
Bergstrom.
2001.
A
Preliminary
Meta
Analysis
of
Contingent
Values
for
Ground
Water
Quality
Revisited,
in
J.
C.
Bergstrom,
K.
J.
Boyle
and
G.
L.
Poe,
eds.
The
Economic
Value
of
Water
Quality.
Northampton,
MA:
Edward
Elgar
Publishers.

Roberts,
Lisa
A.,
and
Leitch,
Jay
A.
1997.
Economic
Valuation
of
Some
Wetland
Outputs
of
Mud
Lake.
Agricultural
Economics
Report
No.
381,
Department
of
Agricultural
Economics,
North
Dakota
Agricultural
Experiment
Station,
North
Dakota
State
University.

Rowe,
Robert
D.,
Schulze,
William
D.,
Hurd,
Brian,
and
Orr,
Douglas.
1985.
Economic
Assessment
of
Damage
Related
to
the
Eagle
Mine
Facility.
Energy
and
Resource
Consultants,
Inc.,
Boulder,
CO.

Sanders,
L.
B.,
Walsh,
Richard
G.,
and
Loomis,
John
B.
1990.
Toward
Empirical
Estimation
of
the
Total
Value
of
Protecting
Rivers.
Water
Resources
Research.
Vol.
26,
No.
7.
pp.
1345­
1357.

Silberman,
J.,
D.
A.
Gerlowski,
and
N.
A.
Williams.
1992.
Estimating
existence
value
for
users
and
nonusers
of
New
Jersey
beaches.
Land
Economics
68
(
2):
225­
236.

Sutherland,
R.
J.,
and
Walsh,
Richard
G.
1985.
Effect
of
Distance
on
the
Preservation
Value
of
Water
Quality.
Land
Economics.
Vol.
61,
No.
3.
pp.
282­
291.

U.
S.
Bureau
of
Labor
Statistics.
2002.
Consumer
Price
Index,
All
Urban
Consumers.
URL:
http://
data.
bls.
gov/
servlet/
SurveyOutputServlet?
data_
tool=
latest_
numbers&
series_
id=
CUUR0000SA0&
o
utput_
view=
pct_
12mths.
Accessed
December
2002.

­
20­
March
12,
2003
U.
S.
Census
Bureau.
2002.
Table
H­
8,
Median
Household
Income
by
State:
1984
to
2001.
Current
Population
Survey,
Annual
Demographic
Supplements.
http://
www.
census.
gov/
hhes/
income/
histinc/
h08.
html.
Last
revised:
September
30th,
2002.

U.
S.
Census
Bureau.
2003a.
Table
C1,
Median
Household
Income
by
County:
1969,
1979,
1989.
1970,
1980,
and
1990
Censuses
of
Population
and
Housing.
http://
www.
census.
gov/
income/
ftp/
decennial/
historical/
county/
c1.
prn.
Accessed
January
2003.

U.
S.
Census
Bureau.
2003b.
Table
MSA1,
Median
Household
Income
by
Metropolitan
Statistical
Areas
(
MSA):
1969,
1979,
and
1989.
1970,
1980,
and
1990
Censuses
of
Population
and
Housing.
http://
www.
census.
gov/
income/
ftp/
decennial/
historical/
msa/
msa1.
prn.
Accessed
January
2003.

VandenBerg,
T.
P.,
G.
L.
Poe,
and
J.
R.
Powell.
2001.
Assessing
the
Accuracy
of
Benefits
Transfers:
Evidence
from
a
Multi­
Site
Contingent
Valuation
Study
of
Ground
Water
Quality,
in
J.
C.
Bergstrom,
K.
J.
Boyle
and
G.
L.
Poe,
eds.
The
Economic
Value
of
Water
Quality.
Northampton,
MA:
Edward
Elgar
Publishers.

Walsh,
Richard
G.,
Greenley,
Douglas
A.,
Young,
Robert
A.,
McKean,
John
R.,
and
Prato,
Anthony
A.
1978.
Option
values,
preservation
values
and
recreational
benefits
of
improved
water
quality:
a
case
study
of
the
South
Platte
River
Basin,
Colorado.
EPA­
600/
5­
78­
001,
Socioeconomic
Environmental
Studies
Series,
Office
of
Research
and
Development,
U.
S.
Environmental
Protection
Agency,
Research
Triangle
Park,
NC.

Welle,
Patrick
G.
1986.
Potential
Economic
Impacts
of
Acid
Deposition:
A
Contingent
Valuation
Study
of
Minnesota.
Dissertation,
University
of
Wisconsin­
Madison.

Whitehead,
John
C.,
and
Groothuis,
Peter
A.
1992.
Economic
Benefits
of
Improved
Water
Quality:
a
case
study
of
North
Carolina's
Tar­
Pamlico
River.
Rivers.
Vol.
3.
pp.
170­
178.

Whitehead,
John
C.,
Blomquist,
Glenn
C.
1991.
A
Link
Between
Behavior,
Information,
and
Existence
Value.
Leisure
Sciences
13:
97
­
109.

Whitehead,
John
C.,
Blomquist,
Glenn
C.,
Hoban,
Thomas
J.,
and
Clifford,
William
B.
1995.
Assessing
the
Validity
and
Reliability
of
Contingent
Values:
A
Comparison
of
On­
Site
Users,
Off­
Site
Users,
and
Non­
users.
Journal
of
Environmental
Economics
and
Management.
Vol.
29.
pp.
238­
251.

Woodward,
Richard
T.,
and
Yong­
Suhk
Wui.
2001.
"
The
Economic
Value
of
Wetland
Services:
A
Meta­
Analysis,"
Ecological
Economics
37:
257­
270.

­
21­
March
12,
2003
A­
1
Appendix
A
­
Non­
Use
Analysis
Studies
and
Characteristics
Table
A
­
Non­
Use
Analysis
Studies
and
Characteristicsa
Author
&

Year
Amenity
Valued
Sampled
Population
Admin
Method
&

Response
Rate
Elicitation
Method,
b
Non­
Use
Method;
Mean
Annual
Non­

Use
WTP/
hh
(
2002$);

Sample
Sizec
Mean
Annual
Use
WTP/
hh
(
2002$)
d
Mean
Annual
User
Non­

Use
WTP/
hh
(
2002$)
Mean
Annual
Household
Income
(
2002$)
e
Water
Body
Type
Geographic
Scale
of
Improvement
Year
Indexf
Clonts
&

Malone
(
1990)
preservation
of
15
AL
rivers
in
a
free­
flowing
state,
free
of
alterations
by
dams,

infrastructure,
diversion,
or
extraction
random
sample
of
AL
households
phone
(
na)
IB
m=
2
$
55.43
N=
733g
$
30.88
$
82.35
$
39,591
rivers
statewide
20
Croke
et
al.

(
1986)
water
quality
improvement
from
unusable
to
a
level
suitable
for
outings
in
urban
Chicago
rivers
households
in
Chicago
and
suburbs
phone
(
85%)
IB
m=
1
$
53.31
N=
296h
$
17.61
$
47,402
urban
rivers
local/
regional
16
Croke
et
al.

(
1986)
water
quality
improvement
from
unusable
to
a
level
suitable
for
outings
&
boating
in
urban
Chicago
rivers
households
in
Chicago
and
suburbs
phone
(
85%)
IB
m=
1
$
60.91
N=
279h
$
11.11
$
47,402
urban
rivers
local/
regional
16
Croke
et
al.

(
1986)
water
quality
improvement
from
unusable
to
a
level
suitable
for
outings,
boating,
&
fishing
in
urban
Chicago
rivers
households
in
Chicago
and
suburbs
phone
(
85%)
IB
m=
1
$
75.11
N=
280h
$
6.35
$
47,402
urban
rivers
local/
regional
16
Cronin
(
1982)
prevent
the
degradation
of
water
quality
in
the
Potomac
River
from
boatable
to
unusable
for
recreation
representative
sample
of
Washington
D.
C.

residences
personal
interview
(
82%)
OE
m=
1
$
58.45
N=
1432
$
48.62
$
56,896
river
single
water
body
12
Desvousges
et
al.
(
1983)
prevent
the
degradation
of
water
quality
in
the
Monongahela
River
from
boatable
to
unusable
for
recreation
households
in
five
counties
in
PA
personal
interview
(
81%)
PC
m=
4
$
111.41
N=
205­
280i
$
89.65
$
130.59
$
38,664
river
single
water
body
13
Huang
et
al.

(
1997)
program
to
restore
quality
of
Pamlico
and
Albemarle
Sounds
to
1981
level
in
terms
of
increased
fish
catches
and
shellfish
beds
eastern
NC
households
phone
(
75%)
DC
m=
5
$
59.02
N=
511
$
158.25
$
37,498
estuary
single
water
body
27
Huang
et
al.

(
1997)
program
to
restore
quality
of
Pamlico
and
Albemarle
Sounds
to
1981
level
in
terms
of
increased
fish
catches
and
shellfish
beds
eastern
NC
households
phone
(
75%)
DC
m=
5
$
119.52
N=
511
$
97.75
$
37,498
estuary
single
water
body
27
Kaoru
(
1993)
improvement
in
water
quality
of
3
coastal
ponds
on
Martha's
random
sample
of
Martha's
Vineyard
mail
(
49%)
OE
m=
3
$
112.57
$
77.04
$
137,693
estuaries
multiple
water
bodies
23
March
12,
2003
Table
A
­
Non­
Use
Analysis
Studies
and
Characteristicsa
Author
&

Year
Amenity
Valued
Sampled
Population
Admin
Method
&

Response
Rate
Elicitation
Method,
b
Non­
Use
Method;
Mean
Annual
Non­

Use
WTP/
hh
(
2002$);

Sample
Sizec
Mean
Annual
Use
WTP/
hh
(
2002$)
d
Mean
Annual
User
Non­

Use
WTP/
hh
(
2002$)
Mean
Annual
Household
Income
(
2002$)
e
Water
Body
Type
Geographic
Scale
of
Improvement
Year
Indexf
A­
2
Vineyard,
MA
to
allow
them
to
be
open
year­
round
for
shell
fishing
(
poor
to
fair)
property
owners
N=
145
Lant
&
Roberts
(
1990)
improvement
in
local
river
water
quality
from
poor
to
fair
in
selected
river
basins
in
IA
&
IL
residents
in
14
towns
in
IA
and
IL,

drawn
from
telephone
listings
personal
interview
(
41%)
PC
m=
2
$
68.56
N=
195
$
57.30
$
016
rivers
local/
regional
20
Lant
&
Roberts
(
1990)
improvement
in
local
river
water
quality
from
fair
to
good
in
selected
river
basins
in
IA
&
IL
residents
in
14
towns
in
IA
and
IL,

drawn
from
telephone
listings
personal
interview
(
41%)
PC
m=
2
$
88.40
N=
195
$
77.04
$
016
rivers
local/
regional
20
Lant
&
Roberts
(
1990)
improvement
in
local
river
water
quality
from
good
to
excellent
in
selected
river
basins
in
IA
&
IL
residents
in
14
towns
in
IA
and
IL,

drawn
from
telephone
listings
personal
interview
(
41%)
PC
m=
2
$
85.29
N=
195
$
78.34
$
016
rivers
local/
regional
20
Magat
et
al.

(
2000)
15%
increase
in
"
good"
water
quality
in
lakes
and
rivers
in
the
region
where
a
respondent
may
hypothetically
live
various
populations
in
CO
and
NC
personal
interview
(
na)
IB
m=
4
$
11.49
N=
348
$
22.98
$
283
all
freshwater
(
lakes
&

rivers)
local/
regional
30
Mitchell
&

Carson
(
1981)
improvement
in
nation's
water
quality
from
boatable
to
a
level
that
"
supports
game
fish
such
as
bass"
national
probability
sample
of
persons
18
or
older
personal
interview
(
73%)
PC
m=
1
$
242.34
N=
695
$
275.09
$
549
all
freshwater
national
11
Olsen
et
al.

(
1991)
doubling
the
Columbia
River
Basin
salmon
&
steelhead
runs
with
the
assurance
of
ecological
Pacific
Northwest
residents
with
phones
phone
(
73%)
OE
m=
1
$
38.48
N=
1177j
$
69.12
$
475
river
single
water
body
21
39,
39,
39,
45,

43,

39,
March
12,
2003
Table
A
­
Non­
Use
Analysis
Studies
and
Characteristicsa
Author
&

Year
Amenity
Valued
Sampled
Population
Admin
Method
&

Response
Rate
Elicitation
Method,
b
Non­
Use
Method;
Mean
Annual
Non­

Use
WTP/
hh
(
2002$);

Sample
Sizec
Mean
Annual
Use
WTP/
hh
(
2002$)
d
Mean
Annual
User
Non­

Use
WTP/
hh
(
2002$)
Mean
Annual
Household
Income
(
2002$)
e
Water
Body
Type
Geographic
Scale
of
Improvement
Year
Indexf
A­
3
stability
and
diversity
Roberts
&

Leitch
(
1997)
water­
related
recreation
and
fish
and
wildlife
habitat
at
Mud
Lake,

MN­
SD
random
sample
of
households
within
30
mile
radius
of
Mud
Lake
mail
(
62%)
PC
m=
2
$
2.09
N=
575k
$
5.17
$
396
wetland
single
water
body
27
Rowe
et
al.

(
1985)
clean
up
and
protection
of
the
Eagle
Mine
area
of
the
Eagle
River,
CO
from
very
poor
quality
(
watering
lawn,
industrial
use)
to
excellent
quality
(
swimming,

washing)
residents
of
Eagle
County,
CO
mail
(
50%)
OE
m=
5
$
69.15
N=
507l
$
47.88
$
606
river
single
water
body
15
Rowe
et
al.

(
1985)
clean
up
Eagle
Mine
hazardous
waste
site,
which
may
currently
preclude
safe
recreation
use
CO
households
mail
(
27%)
PC
m=
3
$
12.46
N=
570m
$
6.54
$
978
river
single
water
body
15
Sanders
et
al.

(
1990)
program
to
protect
3
of
11
most
valuable
wild
and
scenic
CO
rivers
CO
residents;
representative
of
state
mail
(
51%)
OE
m=
3
$
46.91
N=
214
$
24.96
$
rivers
multiple
water
bodies
20
Sanders
et
al.

(
1990)
program
to
protect
7
of
11
most
valuable
wild
and
scenic
CO
rivers
CO
residents;
representative
of
state
mail
(
51%)
OE
m=
3
$
87.60
N=
214
$
46.62
$
rivers
multiple
water
bodies
20
Sanders
et
al.

(
1990)
program
to
protect
11
wild
and
scenic
CO
rivers
CO
residents;
representative
of
state
mail
(
51%)
OE
m=
3
$
111.99
N=
214
$
59.61
$
674
rivers
statewide
20
Sanders
et
al.

(
1990)
program
to
protect
11
wild
and
scenic
CO
rivers
plus
an
additional
4
other
rivers
CO
residents;
representative
of
state
mail
(
51%)
OE
m=
3
$
115.35
N=
214
$
63.45
$
674
rivers
statewide
20
Sutherland
&

Walsh
(
1985)
protection
of
water
quality
in
Flathead
Lake
and
River,
MT,
to
avoid
potential
degradation
sample
drawn
from
phone
directories
of
four
major
cities
adjacent
to
rural
areas
in
MT
mail
(
61%)
OE
m=
3
$
91.53
N=
171
$
35.78
$
907
river
&
lake
single
water
body
15
Walsh
et
al.

(
1978)
postpone
mining
to
avoid
degradation
in
water
quality
throughout
the
South
Platte
River
Basin,
CO,
to
avoid
permanent
preclusion
of
riparian
recreation
(
to
a
level
absent
of
heavy
metals
random
sample
of
households
in
Denver
and
Fort
Collins,
CO
personal
interview
(
na)
IB
m=
4
$
132.63
N=
174­
205n
$
250.66
$
211.42
$
941
river
basin
local/
regional
8
30,
62,
45,
44,674
44,674
44,
44,

39,

43,
March
12,
2003
Table
A
­
Non­
Use
Analysis
Studies
and
Characteristicsa
Author
&

Year
Amenity
Valued
Sampled
Population
Admin
Method
&

Response
Rate
Elicitation
Method,
b
Non­
Use
Method;
Mean
Annual
Non­

Use
WTP/
hh
(
2002$);

Sample
Sizec
Mean
Annual
Use
WTP/
hh
(
2002$)
d
Mean
Annual
User
Non­

Use
WTP/
hh
(
2002$)
Mean
Annual
Household
Income
(
2002$)
e
Water
Body
Type
Geographic
Scale
of
Improvement
Year
Indexf
A­
4
/
contamination)

Welle
(
1986)
avoid
degradation
in
MN
water
quality
from
"
unpolluted"
to
"
severely
polluted"
due
to
reduction
in
acid
deposition
representative
sample
of
MN
adults
drawn
from
list
of
driver's
license
holders
mail
(
76%)
OE
m=
2
$
113.69
N=
415o
$
20.06
$
198.96
$
470
all
freshwater
statewide
16
Whitehead
&

Groothuis
(
1992)
water
quality
improvement
in
the
Tar­
Pamlico
River
due
to
program
to
reduce
non­
point
source
pollution,
resulting
in
double
catch
rate
for
anglers
&

general
improvements
in
fishing,

swimming,
boating,
and
drinking
Eastern
NC
households
the
Tar­

Pamlico
River
basin
phone
(
71%)
OE
m=
1
$
27.74
N=
91
$
18.49
$
164
river
single
water
body
22
Whitehead
et
al.
(
1995)
avoid
degradation
in
water
quality
and
fish
and
wildlife
habitat
in
the
Albemarle­
Pamlico
estuary
system,
NC
100
counties
in
NC,

16
counties
in
VA
mail
(
61%)
IB
m=
1
$
68.08
N=
1033p
$
29.83
$
49,541
estuary
single
water
body
25
a.
All
values
are
adjusted
to
2002
dollars
using
the
Consumer
Price
Index
(
U.
S.
Bureau
of
Labor
Statistics
2002).

b.
Abbreviations
refer
to
elicitation
methods
as
follows:
CC
=
contingent
choice;
DC
=
dichotomous
choice;
IB
=
iterative
bidding;
OE
=
open
ended;
PC
=

payment
card.

c.
m=
method
of
isolating
non­
use
value
and
takes
on
a
value
of
1,
2,
3,
4,
or
5
(
see
section
2.3
for
a
discussion
of
methods
of
isolating
non­
use
value);

N=
sample
size
used
in
regression
analysis
if
available
from
the
original
study,
unless
otherwise
noted.
ple
size
used
for
regression
analysis
of
the
entire
usable
sample
was
not
reported,
or
if
regression
analysis
was
not
conducted,
EPA
selected
the
sample
size
from
each
study
ost
closely
reflects
the
sample
of
responses
used
to
calculate
mean
WTP
values,
and/
or
the
total
 
usable 
survey
responses.
ple
sizes
reported
in
studies
vary
in
terms
of
relevance
to
WTP
estimates
and
other
analyses
(
i.
e.
regression
analyses),
and
some
studies
report
several
sample
sizes
that
correspond
to
variouslydefined
sub­
samples
(
e.
g.
user
versus
non­
user
groups,
or
non­
protest
responses).
some
cases,
sample
sizes
reflect
the
sample
of
responses
used
to
calculate
non­
parametric
WTP.

d.
Use
value
adjusted
based
on
assumption
that
users 
non­
use
value
is
equal
to
non­
users 
WTP.

e.
Mean
or
median
annual
income
estimates
were
not
available
for
the
survey/
regression
samples
in
seven
studies
(
14
observations),
and
were
therefore
estimated
using
Census
income
data.
me
in
place
of
sample
estimates
for
Lant
and
Roberts
(
1990),
Olsen
et
al.
(
1991),
and
Sanders
et
al.
(
1990)
(
U.
S.
Census
2002);
county­
level
income
data
for
Croke
et
al.
(
1986),
Roberts
and
Leitch
(
1997),
and
Sutherland
et
al.
(
1985)
(
U.
S.
Census
43,

46,

If
the
sam
that
m
The
sam
In
EPA
used
state­
level
inco
March
12,
2003
2003a);
and
Metropolitan
Statistical
Area
median
income
for
Cronin
(
1982)
(
U.
S.
Census
2003b).
See
Appendix
B
for
details
on
the
estimation
of
median
income
for
each
study.

f.
Year
in
which
the
study
was
conducted,
converted
to
an
index
by
subtracting
1970.

g.
Clonts
and
Malone
(
1990)
conducted
a
random
telephone
survey
of
733
households.
The
number
of
responses
used
in
calculating
WTP
is
unknown;

regression
results
(
if
applicable)
are
not
reported
in
the
publication.

h.
Represents
sample
size
used
to
calculate
non­
parametric
WTP
for
the
specific
quality
improvement;
sample
sizes
are
different
for
the
WTP
estimates
Croke
et
al.
(
1986)
estimated
for
the
three
water
quality
levels.

i.
Desvousges
et
al.
(
1983)
estimated
components
of
total
value
separately,
and
using
slightly
different
samples.
Therefore,
the
following
sample
sizes
are
associated
with
each
value
type:
existence
value:
205;
option
value:
211;
recreational
use
value:
280.

j.
EPA
interpreted
this
samples
size,
reported
in
the
Olsen
et
al.
(
1991),
to
represent
those
usable
responses
used
in
the
non­
parametric
calculation
of
mean
WTP
estimates.

k.
A
total
of
968
questionnaires
were
completed.

l.
Represents
usable
surveys,
not
the
responses
used
to
calculate
WTP.

m.
This
represents
total
surveys
with
usable
data.
There
were
439
responses
to
the
WTP
question,
and
356
were
used
in
calculating
WTP.

n.
Walsh
et
al.
(
1978)
estimated
components
of
total
value
using
slightly
different
samples,
and
apparently,
non­
parametric
methods.
Therefore,
the
following
sample
sizes
are
associated
with
each
value
type:
existence
value:
203;
bequest
value:
205;
option
value:
177;
recreational
use
value:
174.

o.
Sample
used
in
non­
parametric
estimation
of
mean
WTP.

p.
Represents
the
total
of
off­
site
users,
on­
site
users,
and
non­
users,
used
in
three
separate
regression
models.

A­
5
March
12,
2003
Appendix
B
­
Summary
Descriptions
of
Valuation
Studies
This
appendix
provides
a
short
description
of
each
of
the
18
studies
used
in
the
non­
use
benefits
meta­
analysis.

Clonts
&
Malone
(
1990)
Clonts
&
Malone
used
a
stated
preference
study
(
SP)
to
estimate
the
values
associated
with
preserving
15
Alabama
rivers
in
a
free­
flowing
state.
The
survey
instrument
used
in
this
study
describes
free­
flowing
rivers
as
those
that
have
not
been
significantly
altered
by
dams,
waterways,
diversions,
or
other
measures,
including
water
impoundment
or
extraction.
Some
of
the
objectives
of
this
research
were
to
 
determine
the
recreational
use
of
selected
free­
flowing
Alabama
rivers; 
and
to
 
estimate
preservation
values
Alabamians
place
on
selected
free­
flowing
rivers
in
the
State, 
which
include
recreational
use,
option,
existence,
and
bequest
value.

The
authors
conducted
a
telephone
survey
of
733
Alabama
households
in
1986­
87
to
meet
these
objectives
and
to
obtain
attitudinal
information.
The
article
does
not
provide
a
detailed
description
of
the
survey
instrument
or
the
specific
definition
of
the
amenity
being
valued;
however,
respondents
cited
the
top
two
reasons
for
preserving
rivers
as
 
protecting
fish
and
wildlife
habitat 
and
 
protection
of
water
scenery
and
air
quality. 
For
this
reason,
and
for
the
purposes
of
this
analysis,
EPA
assumed
that
the
WTP
estimates
largely
reflect
the
value
of
aquatic
habitats
and
recreational
uses
in
general.

In
this
study,
households
were
considered
user
households
if
at
least
one
member
had
visited
a
free­
flowing
river
for
the
purpose
of
outdoor
recreation
in
the
past
three
years.
The
survey
asked
both
user
and
non­
user
households
the
same
set
of
valuation
questions
to
elicit
WTP
for
the
value
categories
listed
above
(
method
2
of
isolating
non­
use
value).
Both
users
and
non­
users
placed
the
highest
value
on
the
existence
of
free­
flowing
rivers,
followed
by
bequest,
and
option
values.
The
authors
reported
recreational
value
for
both
groups
and
valued
the
least
of
all
categories,
but
the
context
and
interpretation
of
non­
users 
WTP
for
recreational
use
is
unclear.
Respondents
might
have
misinterpreted
the
question,
or
might
have
included
stewardship
value
in
their
responses,
therefore,
due
to
these
uncertainties,
EPA s
analysis
does
not
consider
non­
users 
recreational
WTP.

This
study
did
not
report
mean
sample
income.
EPA
used
the
average
of
income
ranges
for
river
users
and
river
non­
users
presented
in
the
study
to
approximate
mean
sample
income
to
the
regression­
based
meta­
analysis.

Croke
et
al.
(
1986­
87)
Croke
et
al.
used
a
stated
preference
study
designed
to
estimate
the
value
of
cleaner
rivers
in
the
Chicago
urban
river
system.
This
study
is
the
only
one
among
18
studies
selected
for
the
meta­
analysis
that
deals
exclusively
with
urban
water
resources.
The
purpose
of
the
study
was
to
evaluate
the
benefits
of
an
urban
storm
water­
wastewater
sewer
system
project
that
would
reduce
effluent
discharge
to
rivers
and
waterways.

The
authors
conducted
a
phone
survey
of
350
households
from
a
sample
that
included
the
entire
city
of
Chicago
and
the
surrounding
suburban
communities.
Respondents
defined
baseline
water
quality.
The
authors
obtained
values
for
three
incremental
levels
of
water
quality,
defined
by
the
suitability
of
each
level
for
outings,
boating,
and
fishing.
In
order
to
distinguish
between
users
and
non­
users,
the
survey
asked
respondents
to
determine
the
frequency
with
which
they
used
Chicago s
urban
rivers
for
B­
1
March
12,
2003
recreational
purposes.
9
The
survey
elicited
total
WTP
for
each
respondent
for
each
quality
change,
with
a
separate
estimate
intended
to
represent
the
sum
of
use,
option,
and
existence
value.
The
EPA
analysis
used
WTP
estimates
for
each
of
these
three
quality
changes.
This
study
found
that
non­
use
values
dominate
the
values
of
the
water
quality
improvements.
The
study
also
suggests
that
users
and
non­
users
constitute
distinct
populations
and
that
the
Mitchell
and
Carson
approach
to
estimating
non­
use
values
by
separating
the
sample
into
users
 
who
have
both
use
and
non­
use
values,
and
non­
users
 
who
have
only
non­
use
values
may
introduce
errors
in
estimates
of
non­
use
value
for
users.

Mean
sample
income
was
not
available
in
this
study.
EPA
used
1989
Cook
County,
IL
median
household
income
from
the
decennial
Census
(
U.
S.
Census
Bureau
2003a)
to
assign
a
value
to
the
income
variable
in
meta­
data.

Cronin
(
1982)
Cronin
conducted
a
stated
preference
study
in
the
Washington,
D.
C.
area
to
estimate
WTP
for
water
quality
improvements
in
the
Potomac
River.
The
study
obtained
data
from
1,579
respondents
in
the
Washington,
D.
C.
area
through
personal
interviews.
The
survey
instrument
defined
water
quality
based
on
five
physical
and
bio­
chemical
characteristics
and
six
types
of
outputs
(
e.
g.,
swimming,
boating,
type
and
extent
of
fish
species,
odor,
river
appearance,
and
ecological
and
miscellaneous
outputs).
Cronin
used
method
1
of
isolating
non­
use
value.
This
method
assumes
that
user
and
non­
user
populations
have
identical
non­
use
values.
The
author
applied
a
narrow
definition
of
non­
use
by
considering
a
respondent
to
be
a
user
if
at
least
one
member
of
a
household
uses
 
the
river
in
question,
the
seashore,
or
any
other
river,
lake,
or
local
pool,
for
swimming,
boating,
fishing,
hiking,
camping,
or
picnicking. 

The
study
elicited
option
price
for
both
respondent
groups
 
users
and
non­
users
 
for
preventing
the
deterioration
of
water
quality
from
usable
for
boating
to
unusable
for
recreation.
For
non­
users
who
do
not
plan
to
use
the
water
body
in
the
future
it
can
be
interpreted
as
non­
use
value
for
non­
users
The
authors
also
valued
three
other
quality
changes,
which
roughly
map
to
the
water
quality
ladder
as
follows:
improvement
from
(
1)
usable
for
boating
to
suitable
for
rough
fishing,
(
2)
rough
fishing
to
game
fishing,
and
(
3)
game
fishing
to
suitable
for
swimming.
The
EPA
analysis
used
mean
WTP
for
prevention
of
the
loss
of
the
Potomac
River
for
recreation
for
users
and
non­
users,
which
represents
mean
total
value
for
each
group.
10
Mean
sample
income
was
not
reported
in
this
study.
For
the
regression
analysis,
EPA
used
median
household
income
for
the
Washington
D.
C.
Metropolitan
Statistical
Area
from
the
1970
decennial
Census
to
approximate
the
year
(
1973)
and
geographic
area
(
Washington
D.
C.)
of
the
original
survey
(
U.
S.
Census
Bureau
2003b).

Desvousges
et
al.
(
1983)
Desvousges
et
al.
report
the
results
of
a
U.
S.
EPA
study
that
compared
alternative
approaches
for
estimating
water
quality
benefits,
using
the
Pennsylvania
portion
of
the
Monongahela
River
basin
as
a
case
study.
The
study
used
a
stated
preference
method
to
estimate
WTP
for
water
quality
improvements.

9The
user
population
is
less
than
10
percent
of
the
entire
sample,
which
is
likely
due
to
the
very
poor
quality
of
Chicago s
urban
rivers
at
the
time
of
the
study.

10Fisher
and
Raucher
(
1984)
calculated
WTP
values
from
Table
6.1
of
Cronin
1982,
which
are
presented
in
Tables
3
and
5
of
their
review.
EPA
used
the
values
from
Table
5
in
this
analysis.

B­
2
March
12,
2003
The
authors
conducted
personal
interviews
of
301
households
in
1981
based
on
a
stratified
sample
covering
five
Pennsylvania
counties.
Four
groups
of
respondents
received
slightly
different
versions
of
the
stated
preference
survey,
all
of
which
valued
the
same
changes
in
water
quality.
Only
one
version
of
the
survey,
which
used
the
direct
question
payment
card
elicitation
format,
measured
respondents 
non­
use
value.
The
survey
asked
these
respondents
to
assume
zero
use
of
the
river
and
to
state
their
WTP
to
avoid
a
decrease
in
water
quality
from
boatable
to
not
suitable
for
recreation.
Option
and
use
values
were
obtained
through
the
same
elicitation
format
(
i.
e.,
separate
questions).
This
study
used
method
4
of
measuring
non­
use
value,
with
separate
non­
use
value
estimates
for
user
and
non­
user
groups.

The
authors
found
some
respondents
reported
the
same
figure
for
the
existence
value
question
as
they
did
for
option
price.
This
indicates
that
some
respondents
may
confuse
existence
(
i.
e.,
non­
use)
with
option
values.
EPA
found,
however,
that
the
ratio
of
non­
user
to
user
value
from
this
study
closely
approximates
the
median
of
the
ratios
used
in
this
analysis,
representing
a
conservative
estimate
of
non­
use
benefits
relative
to
the
sample.
Moreover,
Mitchell
and
Carson
(
1981)
argue
that
option
value
is
small
for
non­
unique
resources.

Huang
et
al.
(
1997)
Huang
et
al.
analyzed
revealed
and
stated
preference
data
from
a
telephone
survey
of
1077
eastern
North
Carolina
households
conducted
in
1995.
The
survey
consisted
of
both
revealed
and
stated
preference
questions.
Respondents
stated
their
WTP
for
a
program
that
would
restore
the
quality
of
the
Pamlico
and
Albemarle
Sounds
to
1981
levels,
which
would
result
in
a
60
percent
increase
in
fish
catches
and
a
25
percent
increase
in
the
quantity
of
open
shellfish
beds.
The
survey
informed
respondents
of
the
causes
of
the
degradation
and
asked
for
their
WTP
to
restore
the
water
quality
of
the
sounds
through
more
stringent
laws.

The
authors
estimated
travel
cost
and
contingent
behavior
models
to
estimate
respondents 
respective
recreational
and
total
values
for
restoring
water
quality
in
the
sounds.
The
study
measured
consumer
surplus
using
both
ex­
ante
and
ex­
post
recreation
trips
taken
at
the
current
quality
level,
yielding
separate
estimates
of
recreational
value.
The
authors
compared
these
values
with
total
WTP
based
on
stated
preference
data
to
calculate
two
estimates
of
the
non­
use
value
of
the
quality
change.
This
study
represents
one
of
two
instances
in
this
meta­
analysis
where
revealed
and
stated
preference
data
were
combined
to
estimate
the
non­
use
value
of
an
environmental
quality
change
(
method
5).

Kaoru
(
1993)
This
study
estimated
the
benefits
of
water
quality
improvements
in
three
coastal
ponds
on
the
island
of
Martha s
Vineyard,
Massachusetts.
The
current
level
of
water
quality
in
these
coastal
ponds
corresponds
to
intermittent
closures
of
the
ponds
for
shell
fishing.
Kaoru
surveyed
559
local
property
owners
in
1989
and
asked
their
WTP
for
water
quality
improvements
in
the
three
ponds
from
the
current
level
to
a
level
that
would
allow
them
to
be
open
for
shell
fishing
year­
round.
Kaoru
employed
method
3
of
isolating
non­
use
value
by
asking
respondents
for
their
total
WTP
and
then
asking
them
to
allocate
this
amount
into
use,
option,
and
existence
values.
The
mean
value
respondents
placed
on
existence
value
accounts
for
over
half
of
their
total
mean
WTP.

Lant
and
Roberts
(
1990)
Lant
and
Roberts
used
the
contingent
valuation
method
to
estimate
recreational
and
intrinsic
benefits
of
river
water
quality
in
the
corn
belt,
in
the
context
of
agricultural
non­
point
source
pollution.
The
authors
conducted
200
personal
interviews
with
residents
in
14
selected
towns
in
Iowa
and
Illinois
in
1987,

B­
3
March
12,
2003
concerning
selected
river
basins
in
the
two
states.
Questions
to
the
entire
sample
of
respondents
elicited
both
recreational
and
intrinsic
values
for
the
following
three
water
quality
improvements:
poor
to
fair,
fair
to
good,
and
good
to
excellent.
The
authors
used
photographs
and
further
defined
improvements
as
incrementally
allowing
boating,
rough
fishing,
game
fishing,
and
swimming,
respectively.
The
water
quality
levels
considered
in
this
study
maps
onto
the
Resources
For
the
Future
(
RFF)
water
quality
ladder
as
follows:

Water
Quality
Level
Supported
Uses
poor
6
boating
fair
6
boating,
rough
fishing
good
6
boating,
rough
fishing,
game
fishing
excellent
6
boating,
rough
fishing,
game
fishing,
swimming
Therefore,
water
quality
changes
from
 
poor 
to
 
fair 
and
from
 
fair 
to
 
good 
are
expected
to
affect
the
quality
and
quantity
of
recreational
fishery
resources.
Water
quality
changes
from
 
good 
to
excellent
may
further
improve
recreational
game
fishing.

This
study
did
not
report
mean
sample
income.
EPA
used
the
average
of
Illinois
and
Iowa
median
income
provided
by
the
Census
in
the
annual
Historical
Income
Tables
(
U.
S.
Census
Bureau
2002)
to
assign
a
value
to
the
income
variable
used
in
the
regression
analysis.

Magat
et
al.
(
2000)
Magat
et
al.
conducted
a
unique
iterative
choice
method
survey
for
valuing
freshwater
quality.
The
authors
used
personal,
centralized,
and
intercept
interview
methods
to
survey
410
respondents
in
North
Carolina
and
Colorado.
Respondents
addressed
several
iterations
of
hypothetical
scenarios,
in
which
they
were
asked
to
consider
moving
to
a
new
region
where
15
percent
more
of
the
freshwater
could
be
characterized
as
 
good, 
compared
with
the
region
in
which
they
currently
lived.

The
authors
defined
"
good"
water
quality
as
safe
for
swimming
and
supportive
of
a
variety
of
plants
and
aquatic
life,
including
a
variety
of
fish
safe
for
consumption.
The
authors
defined
water
quality
that
was
 
not
good 
as
unsafe
for
swimming,
supportive
of
only
a
small
number
of
plants
and
aquatic
life,
and
not
supportive
of
edible
fish.
The
study
obtained
WTP
values
via
increased
costs
of
living
associated
with
respondents 
decisions
of
whether
or
not
to
move
to
a
region
with
a
greater
percentage
 
good 
water
quality.

The
authors
obtained
non­
use
values
in
a
separate
section
of
the
survey,
through
a
similar
iterative
choice
approach
where
the
respondents
assumed
that
they
would
have
a
zero,
one­
third,
or
one­
tenth
probability
of
visiting
lakes
and
rivers
in
the
region.
This
approach
can
be
classified
as
method
4
of
isolating
non­
use
value.
EPA s
meta­
analysis
uses
the
mean
value
of
instances
where
respondents
would
have
no
chance
of
visiting
lakes
and
streams
in
the
new
region,
which
most
closely
approximates
pure
non­
use
value.

Mitchell
and
Carson
(
1981)
In
this
seminal
study,
Mitchell
and
Carson
estimated
values
for
national
freshwater
quality
improvements
using
CVM.
The
authors
conducted
personal
interviews
of
1,576
households
in
1980.
Based
on
the
RFF
water
quality
ladder,
respondents
stated
their
perceptions
regarding
the
current
quality
of
the
nation s
B­
4
March
12,
2003
lakes,
rivers,
and
streams,
and
subsequently
reported
their
WTP
to
improve
the
water
quality
to
three
improved
levels:
a
level
safe
for
boating,
a
level
safe
for
fishing,
and
a
level
safe
for
swimming.
11
Mitchell
and
Carson
used
method
1
for
isolating
non­
use
value,
thereby
obtaining
total
WTP
values
for
respondents
in
both
the
user
and
non­
user
groups.
Respondents
were
classified
as
non­
users
if
 
they
had
not
boated,
fished,
or
swum
in
freshwater
in
the
past
two
years. 
The
authors
estimated
non­
use
WTP
only
for
improving
the
nationwide
water
quality
to
a
fishable
level.
In
addition,
Mitchell
and
Carson
estimated
non­
use
value
for
users
by
assuming
that
users
and
non­
users
hold
the
same
intrinsic
values
for
water
quality.

Olsen
et
al.
1991
Olsen
et
al.
conducted
a
CVM
study
to
estimate
non­
market
benefits
associated
with
a
major
fishery
management
program
in
the
Pacific
Northwest,
which
was
designed
to
double
the
size
of
salmon
and
steelhead
fish
runs
in
the
Columbia
River
Basin
by
the
year
2000.
In
an
effort
to
estimate
both
existence
and
sport
values
of
the
resource,
the
authors
conducted
a
telephone
survey
of
households
in
Idaho,
Montana,
Oregon,
and
Washington
in
1989,
resulting
in
1,395
usable
responses.

The
survey
used
method
4
of
isolating
non­
use
value,
and
was
designed
to
distinguish
between
three
respondent
types:
non­
users
with
no
probability
of
future
use,
non­
users
with
some
probability
of
future
use,
and
users.
Non­
users
with
no
probability
of
future
use
had
not
participated
in
the
commercial
or
sport
fishery
and
did
not
plan
to
do
so;
those
with
some
probability
of
future
use
had
not
participated
in
the
previous
five
years
but
were
uncertain
about
fishing
during
the
next
five
years;
and
resource
users
reported
having
participated
in
the
salmon
or
steelhead
commercial
or
sport
fisheries
within
the
previous
two
years.
The
authors
estimated
total
WTP
for
each
individual,
capturing
existence
value,
option
value,
and
expected
consumer
surplus
(
use
value).
For
the
purposes
of
this
analysis,
EPA
considered
only
the
users
and
the
non­
users
with
no
probability
of
future
use.

This
study
did
not
report
mean
sample
income.
Therefore,
EPA
used
the
average
of
1989
state
median
income
for
Idaho,
Montana,
Oregon,
and
Washington,
available
from
Census 
annual
Historical
Income
Tables,
to
estimate
sample
income
for
the
regression
analysis
(
U.
S.
Census
Bureau
2002).

Roberts
and
Leitch
(
1997)
Roberts
and
Leitch
conducted
an
economic
evaluation
of
wetland
outputs
of
Mud
Lake,
MN­
SD,
including
a
CVM
study
designed
to
estimate
recreational
and
aesthetic
values.
The
authors
conducted
a
mail
survey
of
1,034
randomly
selected
households
within
a
30­
mile
radius
of
Mud
Lake
in
1996.

The
survey
asked
separate
questions
of
each
respondent
to
determine
annual
per
acre
WTP
for
use,
option,
and
existence
value.
Discussion
of
method
2
of
isolating
non­
use
value
(
see
Section
2.3
of
this
memorandum)
presents
the
text
of
the
survey
questions
that
elicited
use
and
existence
values.
The
authors
asked
a
third
question
to
estimate
option
value:
 
What
is
the
maximum
amount
you
would
be
willing
to
pay
through
an
annual
voluntary
donation
to
ensure
that
recreational
activities
and
fish/
wildlife
habitat
at
Mud
Lake
are
available
in
the
future
to
you
or
your
descendants? 
This
question
seems
to
elicit
both
option
and
bequest
values.
For
this
meta­
analysis,
EPA
included
the
values
obtained
from
this
question
in
the
estimate
of
use
value
due
to
the
inability
to
discern
between
option
and
bequest
values.

11
See
description
of
Lant
and
Roberts
(
1990)
above
for
a
description
of
the
RFF
water
quality
ladder.

B­
5
March
12,
2003
This
study
did
not
report
mean
sample
income.
Therefore,
EPA
used
the
average
of
Census
estimates
of
1989
median
household
income
for
counties
located
within
a
30­
mile
radius
of
Mud
Lake
in
order
to
estimate
sample
income
for
the
regression
analysis
(
U.
S.
Census
Bureau
2003a).

Rowe
et
al.
(
1985)
In
1985,
Rowe
et
al.
conducted
an
economic
assessment
of
natural
resource
damages
associated
with
the
Eagle
Mine
facility
near
Gilman,
CO.
12
Mine
tailings
from
the
Eagle
Mine
had
polluted
the
Eagle
river,
reducing
the
trout
stock
and
adversely
affecting
the
aesthetic
quality
of
the
area.
The
authors
conducted
two
contingent
valuation
surveys
to
estimate
damages
to
both
Eagle
County
residents
and
the
statewide
population,
as
well
as
to
gather
recreational
participation
information.

In
the
first
survey,
Eagle
County
residents
received
a
questionnaire
by
mail
that
elicited
WTP
for
various
potential
actions
to
ameliorate
degraded
water
quality
and
other
aesthetic
damages
in
the
area.
For
this
analysis
EPA
was
interested
in
the
mean
WTP
of
respondents
to
 
clean
up
and
protect 
the
area.
Under
this
option
water
quality
would
be
raised
from
a
level
 
okay
for
watering
your
lawn
and
some
industrial
uses 
to
a
level
acceptable
for
activities
such
as
swimming
and
rafting.
This
value,
representing
the
total
of
use,
option,
and
bequest
values,
can
be
compared
with
the
adjusted
unit
day
value
estimate
of
wildlife
and
fish
recreation
from
the
recreational
demand
analysis
used
in
the
study,
to
estimate
non­
use
value.
This
approach
uses
both
revealed
and
stated
preference
data
to
estimate
non­
use
value
(
method
5).
Although
the
study
did
not
compare
unit
estimates,
the
report
presented
the
results
from
these
separate
analyses
to
compare
aggregate
values.
13
The
authors
sent
the
second
survey
to
2,800
Colorado
households
with
the
primary
objective
of
estimating
use,
bequest,
and
existence
values
for
cleaning
up
problems
at
Colorado
hazardous
waste
sites.
The
Eagle
Mine
facility
was
one
of
seven
primary
sites
of
interest
for
which
values
were
obtained.
The
quality
improvement
was
similar
to
that
described
in
the
County
survey,
but
respondents
were
to
state
their
total
WTP
and
then
allocate
that
figure
among
hazardous
waste
sites
and
value
categories.
Thus,
this
survey
employed
method
3
of
isolating
non­
use
value.

Sanders
et
al.
(
1990)
Sanders
et
al.
report
the
results
of
a
stated
preference
survey
conducted
in
1983
to
estimate
Colorado
residents 
WTP
to
protect
wild
and
scenic
rivers
in
the
state.
A
mail
survey
resulted
in
214
responses
in
which
respondents
ranked
the
four
most
important
of
11
rivers
described
in
the
survey.
The
survey
described
river
protection
as
resulting
in
an
array
of
potential
benefits,
which
respondents
were
asked
to
12The
authors
represented
Energy
and
Resource
Consultants,
Inc.,
and
prepared
the
report
for
the
State
of
Colorado
in
connection
with
State
of
Colorado
v.
Gulf
and
Western
Industries,
Inc.,
et
al.

13Rowe
et
al.
used
the
unit
day
value
of
outdoor
wildlife
and
fish
recreation,
as
recommended
by
the
U.
S.
Forest
Service,
Rocky
Mountain
Region,
as
a
basis
for
estimating
recreational
value.
The
authors
first
multiplied
this
value
by
an
estimate
of
the
average
annual
recreation
days
per
person
obtained
from
the
survey,
the
adult
population
in
the
region
of
interest,
and
the
unit
day
price
for
wildlife
and
fish
recreation.
The
authors
then
adjusted
for
discounting
and
population
growth
to
estimate
aggregate
recreation
value.
They
also
made
an
adjustment
to
deflate
the
value
to
represent
two
of
the
12
hours
the
unit
day
value
from
the
Forest
Service
represents,
based
on
the
average
length
of
Eagle
County
residents 
recreation
trips.
For
this
meta­
analysis,
EPA
simply
multiplied
unit
values
by
recreation
days
and
adjusted
for
the
average
length
of
a
recreation
trip
for
comparison
with
mean
total
WTP
from
the
CVM
survey.

B­
6
March
12,
2003
indicate
each
of
a
list
of
benefits
were
to
them.
The
top
two
reasons
for
protecting
rivers
were
to
 
protect
the
quality
of
water,
air,
and
scenery, 
and
 
protecting
fish
and
wildlife
habitat. 
Respondents
stated
their
WTP
to
protect
their
top
one,
two,
three,
and
four
rivers,
as
well
as
all
11
rivers,
and
then
allocated
their
value
among
four
value
categories:
use,
option,
existence,
and
bequest
(
method
3
of
isolating
non­
use
value).
The
study
generated
mean
household
WTP
values
for
the
top
three
and
seven
most
valuable
rivers,
11
study
rivers,
and
the
11
study
rivers
plus
the
four
most
important
additional
Colorado
rivers
indicated
by
respondents.
These
data
represent
four
observations
in
the
meta­
analysis.

This
study
did
not
report
mean
sample
income.
For
the
regression
analysis,
EPA
used
the
Census
estimate
of
1984
median
annual
household
income
for
the
state
of
Colorado
(
U.
S.
Census
Bureau
2002)
to
assign
a
value
for
the
income
variable.

Sutherland
and
Walsh
(
1985)
Sutherland
and
Walsh
report
the
results
of
a
CVM
study
designed
to
estimate
use
and
non­
use
values
associated
with
the
water
quality
of
the
Flathead
River
and
Lake
system
in
Montana.
Specifically,
the
study
focused
on
testing
the
hypothesis
that
WTP
for
non­
use
values
of
a
resource
declines
with
distance.
A
mail
survey
resulted
in
171
responses
from
residents
in
four
areas
of
the
state
of
Montana.
All
respondents
first
stated
their
total
WTP
for
protection
of
water
quality
at
Flathead
River
and
Lake
to
prevent
potential
degradation.
Users
of
the
resource
then
allocated
this
amount
among
recreational
use,
option,
existence,
and
bequest
values,
while
non­
users
allocated
their
WTP
to
existence
and
bequest
value
categories
only.

The
following
allocation
questions
were
asked
to
elicit
the
WTP
for
each
indicated
value
category:

Recreational
use
value:
 
Payment
to
visit
Flathead
Lake
or
River
this
year
(
in
addition
to
traveling
or
lodging
expenses).
_____% 
Option
value:
 
Payment
for
the
opportunity
to
visit
the
Lake
or
River
in
the
future
at
the
same
level
of
water
quality
and
fishing
conditions.
_____% 
Existence
value:
 
Payment
to
preserve
water
quality
in
Flathead
River
and
Lake.
The
value
to
you
from
knowing
that
good
water
quality
exists
here.
_____% 
Bequest
value:
"
Payment
to
preserve
water
quality
in
Flathead
River
and
Lake.
The
value
to
you
from
knowing
that
future
generations
will
have
good
water
quality.
_____% 

This
study
represents
the
allocation
method
(
method
3)
of
isolating
non­
use
value.
Mean
WTP
estimates
are
available
for
the
entire
sample
only
despite
the
distinction
between
users
and
non­
users
of
the
resource.

This
study
did
not
report
mean
sample
income.
EPA
used
Census
estimates
of
1979
annual
median
household
income
to
estimate
sample
income
for
the
regression
analysis.
The
study
stated
that
four
major
Montana
cities
were
surveyed,
and
reported
their
respective
distances
from
the
study
site.
We
inferred
the
four
most
likely
cities
included
in
the
sample
frame
based
on
this
information
and
used
the
average
of
median
household
income
for
the
counties
inclusive
of
these
cities
(
U.
S.
Census
Bureau
2003a).

Walsh
et
al.
(
1978)
Walsh
et
al.
conducted
a
study
for
the
U.
S.
EPA
with
the
primary
objective
of
estimating
the
option,
existence,
and
bequest
portions
of
benefits
resulting
from
pollution
abatement,
relative
to
user
value.
The
B­
7
March
12,
2003
authors
conducted
a
contingent
valuation
survey
through
personal
interviews
of
202
Colorado
residents.
The
valuation
questions
estimated
household
WTP
to
postpone
mining
that
would
degrade
water
quality
throughout
the
South
Platte
Basin.
The
survey
used
three
photos
to
describe
the
current
level
of
pollution
in
the
River
Basin,
characterized
as
including
both
pristine
waters
at
high
elevations
and
water
bodies
severely
polluted
with
heavy
metals,
algae,
chemicals,
and
bacteria.
The
survey
informed
respondents
of
the
potential
for
further
degradation,
which
would
permanently
preclude
riparian
recreation.
Their
payments
would
be
used
to
improve
water
quality
to
a
level
absent
of
heavy
metals
and
contamination
in
order
to
enhance
recreational
enjoyment.

To
isolate
non­
use
values,
the
survey
asked
respondents
to
assume
with
certainty
that
they
would
not
use
the
River
Basin
for
water­
based
recreation
activities
(
method
4).
Recreational
use
questions
enabled
the
authors
to
distinguish
users
from
non­
users,
who
were
presented
with
separate
questions
to
elicit
option,
bequest,
and
existence
values.
As
a
result,
the
authors
estimated
mean
WTP
for
existence
and
bequest
values
for
non­
users
and
estimated
use
and
option
values
for
users.

Welle
(
1986)
Welle s
dissertation
concerns
the
potential
economic
impacts
of
acid
deposition.
The
analysis
makes
use
of
data
from
a
contingent
valuation
study
conducted
in
1985,
in
which
a
mail
survey
resulted
in
669
usable
responses.
Welle
estimated
benefits
for
reducing
acid
deposition
pollution
and
thus
avoiding
the
degradation
of
water
quality
in
the
state
of
Minnesota.
The
study
used
an
environmental
quality
ladder
to
define
the
amenity
change
being
valued,
which
can
be
mapped
to
the
water
quality
ladder
to
represent
a
change
from
 
unpolluted 
to
 
severely
polluted. 
Respondents
valued
the
avoidance
of
such
a
decrease
in
the
quality
of
freshwater
across
the
state.

The
study
grouped
respondents
into
the
following
four
categories
according
to
their
likelihood
of
participating
in
water­
based
recreation
in
the
future:
extremely
unlikely
(
non­
user),
somewhat
likely,
likely,
and
extremely
likely
(
user).
For
the
purposes
of
this
meta­
analysis,
EPA
is
concerned
only
with
the
extremely
unlikely
and
extremely
likely
groups,
which
may
be
considered
non­
users
and
users,
respectively.
The
survey
included
several
valuation
questions
and
value
estimates.
Significantly,
the
study
estimated
demand­
side
option
price
(
option,
existence,
and
recreational
value)
and
existence
values
for
all
respondent
groups.
The
difference
between
these
estimates
for
users
is
the
sum
of
recreation
and
option
value.
As
expected,
the
demand­
side
option
price
for
non­
users
was
roughly
equal
to
their
use
value,
indicating
their
certainty
of
not
using
the
resource.

Whitehead
and
Groothuis
(
1992)
Whitehead
and
Groothuis
examined
the
economic
benefits
of
improving
water
quality
in
North
Carolina s
Tar­
Pamlico
River,
which
had
experienced
decreased
fish
catch
rates,
disease,
and
algae
blooms,
largely
as
a
result
of
agricultural
non­
point
source
pollution.
The
authors
conducted
a
contingent
valuation
survey
of
Eastern
North
Carolina
households
via
mail
in
1991,
yielding
109
responses.
Respondents
valued
water
quality
improvement
in
the
river
due
to
the
implementation
of
Best
Management
Practices
by
farmers
to
reduce
non­
point
source
pollution.
The
result
would
be
a
doubled
catch
rate
for
anglers
and
general
improvements
in
swimming,
boating,
and
drinking.

The
authors
divided
the
sample
into
expected
users
and
expected
non­
users.
Users
reported
that
they
would
fish
in
the
river
after
the
specified
quality
improvement,
and
non­
users
indicated
that
they
would
8not
fish
in
the
river.
This
definition
of
users
has
the
advantage
of
capturing
those
respondents
who
would
become
users
of
the
resource
due
to
the
amenity
change,
yet
lacks
the
comprehensiveness
of
other
B­
8
March
12,
2003
approaches
by
only
considering
the
recreational
use
of
fishing.
The
authors
elicited
total
WTP
for
both
groups.
Non­
user
WTP
may
be
interpreted
as
the
sum
of
existence
and
intrinsic
values.

Whitehead
et
al.
(
1995)
Whitehead
et
al.
examined
the
WTP
bids
that
various
user
groups
reported
in
a
contingent
valuation
survey.
The
survey,
conducted
by
telephone
in
1990,
sampled
100
counties
in
North
Carolina
and
16
counties
in
Virginia,
yielding
1,133
responses.
Respondents
valued
water
quality
and
fish
and
wildlife
habitat
in
the
Albemarle­
Pamlico
(
A­
P)
estuary
system.
The
survey
explained
the
use
and
non­
use
benefits
of
estuaries
to
respondents
before
they
stated
their
WTP
to
avoid
further
degradation
and
to
ensure
the
current
level
of
wildlife
habitat
in
the
A­
P
system.

The
authors
divided
the
sample
into
on­
site
users,
off­
site
users,
and
non­
users,
based
on
their
responses
to
questions
in
the
survey.
On­
site
users
stated
that
they
had
 
engaged
in
fishing,
swimming,
boating,
or
some
other
activity
during
the
past
year.... 
Off­
site
users
had
not
been
to
the
site,
and
answered
 
a
little,"
"
some,"
or
"
a
lot,"
to
the
question,
"
How
much
have
you
heard
or
read
about
the
resources,
uses,
and
problems
of
the
A­
P
system? 
Non­
users
answered
 
nothing 
to
the
question
regarding
off­
site
use,
and
were
not
users
of
the
Albemarle­
Pamlico
system.
This
study
used
method
1
of
isolating
non­
use
value
by
estimating
total
WTP
values
for
the
members
of
each
group.
To
avoid
confusion
in
interpreting
non­
use
value,
EPA
was
concerned
only
with
on­
site
users
and
non­
users
in
this
analysis.
The
WTP
of
non­
users,
which
was
the
lowest
of
the
three
groups,
may
be
interpreted
as
the
sum
of
existence
and
bequest
values,
plus
some
potential
option
value.

B­
9
