MEMORANDUM
Date:
April
22,
2002
To:
Docket
A­
2000­
01
From:
Bryan
Hubbell
Senior
Economist,
OAQPS
Re:
Estimated
NO
x,
SO
2,
and
PM
Emissions
Health
Damages
for
Heavy
Duty
Vehicle
Emissions
This
note
provides
a
simple
analysis
of
the
health
costs
of
NO
x,
SO
2,
and
PM
emissions
from
diesel
vehicles
(
or
similarly
health
benefits
of
emission
reductions
from
these
vehicles).
It
is
important
to
recognize
that
these
are
screening
level
ballpark
estimates.
These
numbers
are
uncertain
and
may
significantly
under
or
overstate
actual
benefits
of
any
emission
reductions.
These
transfer
coefficients
may
be
appropriate
for
screening
level
analyses
or
in
informal
discussions
to
gain
a
general
understanding
of
the
value
of
reducing
diesel
emissions,
as
long
as
the
limitations
and
major
underlying
assumptions
are
clearly
explained.
Please
contact
me
before
using
these
results
for
any
other
purpose
(
919)
541­
0621.

Our
approach
is
to
develop
estimates
of
health
costs
expressed
in
per
ton
terms.
These
transfer
estimates
will
then
allow
us
to
examine
scenarios
involving
reductions
in
emissions
from
diesel
vehicles
without
having
to
prepare
model­
ready
emissions
inventories
or
having
to
perform
new
air
quality
modeling
runs.
We
employ
the
REMSAD
modeling
used
for
the
HD
Engine/
Diesel
Fuel
Rule
RIA
benefits
analysis
to
determine
estimates
of
damages/
ton.
Aggregate
damage
estimates
at
the
national
level
can
be
scaled
by
national
population
to
account
for
population
changes
between
years
of
analysis.
Because
of
the
assumptions
inherent
in
these
national
estimates,
it
is
not
appropriate
to
scale
them
to
local
populations.
Complete
details
of
the
emissions,
air
quality,
and
benefits
modeling
conducted
for
the
HD
Engine/
Diesel
Fuel
Rule
can
be
found
at
http://
www.
epa.
gov/
otaq/
diesel.
htm
and
http://
www.
epa.
gov/
ttn/
ecas/
regdata/
tsdhddv8.
pdf.

We
examine
the
impacts
of
SO
2,
NO
x,
and
direct
PM
emissions.
Excess
emissions
of
NO
x
are
associated
with
potential
increases
in
both
ozone
and
particulate
matter
(
PM).
Excess
emissions
of
SO
2
and
direct
PM
are
primarily
associated
with
increases
in
PM.
Due
to
data
limitations,
we
are
providing
estimates
only
for
PM
related
health
impacts.
The
vast
majority
of
the
benefits
we
are
able
to
measure
are
PM
related.
Thus,
these
estimates
will
capture
most
of
the
monetized
benefits
associated
with
reductions
in
SO
2,
NO
x,
and
direct
PM
emissions.
However,
one
important
limitation
is
that
benefits
from
ozone
reductions,
visibility
improvement,
and
other
unquantifiable
health
endpoints
are
not
captured
in
these
estimates.
The
results
of
this
analysis
are
summarized
in
Table
1.

Before
applying
the
values
reported
in
this
memo
for
policy
analysis,
several
factors
should
be
considered.
These
factors
include
whether
the
sources
of
emissions,
meteorology,
transport
of
emissions,
initial
PM
concentrations,
population
density,
and
population
demographics
are
reasonably
consistent
with
those
used
in
generating
the
benefit
transfer
values.
A
major
assumption
is
that
application
of
these
results
is
for
cases
where
the
situation
is
consistent
with
the
conditions
examined
in
the
HD
Engine/
Diesel
Fuel
analysis.
A
general
rule
is
that
as
these
factors
increasingly
diverge,
the
likelihood
of
significant
error
in
the
estimated
benefits
values
will
increase.

While
the
benefit
transfer
figures
will
provide
a
reasonable
ball
park
estimate
of
the
national
benefits
of
a
nationwide
reduction
in
emissions
from
diesel
sources,
considerably
more
uncertainty
is
introduced
by
applying
the
transfer
value
for
the
nation
to
specific
emission
reductions
within
a
particular
state
or
region.
This
occurs
because
we
assume
that
all
emissions
from
diesel
sources
are
valued
at
the
same
average
$/
ton/
person
amount.
There
may
be
significant
variation
in
the
$/
ton/
person
for
any
specific
ton
of
emission
reductions.
The
magnitude
and
direction
of
this
bias
is
not
known.
This
limitation
should
be
pointed
out
to
audiences
of
your
analysis.

It
is
also
important
to
note
that
the
population
adjustments
provided
in
the
method
below
are
to
account
for
growth
in
the
national
population
in
future
years.
You
should
not
adjust
the
$/
ton
or
incidence/
ton
transfer
estimates
within
a
given
year
for
differences
in
populations
between
a
given
region
of
interest
and
the
national
population.
This
will
lead
to
a
severe
underestimation
bias
in
your
results.
This
is
because
the
averaging
process
assumes
that
all
reductions
in
emissions,
wherever
they
occur,
affect
air
quality
across
the
entire
U.
S.
population.
In
order
to
derive
appropriate
transfer
values
for
states
or
regions
within
the
U.
S.,
you
must
first
construct
incidence/
ton
and/
or
$/
ton
estimates
for
the
entire
country
for
the
year(
s)
of
interest
and
then
apply
those
to
the
emission
reductions
in
the
particular
region
(
see
below
for
an
example).
The
total
$/
ton
estimates
listed
for
each
pollutant
in
Table
1
can
be
used
to
value
tons
reduced
anywhere
in
the
U.
S.
for
2001
population
levels.
For
other
years,
you
will
have
to
recalculate
the
national
average
$/
ton
using
the
procedures
described
below.
Again,
let
me
remind
the
user
that
applying
these
average
$/
ton
estimates,
which
are
based
on
emission
reductions
that
occured
throughout
the
U.
S.,
to
emission
reductions
occuring
in
a
specific
location,
can
lead
to
potentially
signficant
biases
(
of
an
unknown
direction)
in
the
results.

Finally,
it
should
be
noted
that
benefit
transfer
values
should
be
used
only
as
an
input
to
a
screening
level
benefits
analysis
or
analyses
intended
to
provide
ballpark
estimates
of
what
benefits
could
be
provided
by
reductions
in
diesel
emissions.
For
policy
support
purposes,
such
as
a
regulatory
impact
analysis,
a
full
benefits
analysis
should
be
conducted,
using
best
practices
including
emissions
modeling,
air
quality
modeling,
and
application
of
health
and
economic
valuation
functions
at
an
appropriate
geographic
scale.
Table
1.
Summary
of
Health
Effects
and
Economic
Cost
Estimates
for
Transfer
Health
Effect
Incidence/
ton
in
2001
based
on
U.
S.
population
of
277
million
Estimated
$/
ton
economic
costs
in
2001
based
on
U.
S.
population
of
277
million
(
1999$)

SO2
NOx
PM
SO2
NOx
PM
All­
cause
Premature
Mortality
from
Long­
term
Exposure
0.0025
0.0016
0.0221
$
15,528
$
9,726
$
136,164
Chronic
Bronchitis
0.0016
0.0010
0.0143
$
564
$
350
$
5,012
Hospital
Admissions
­
COPD
0.0003
0.0002
0.0024
$
3
$
2
$
30
Hospital
Admissions
­
Pneumonia
0.0003
0.0002
0.0030
$
5
$
3
$
44
Hospital
Admissions
­
Asthma
0.0002
0.0002
0.0023
$
2
$
1
$
15
Hospital
Admissions
­
Total
Cardiovascular
0.0008
0.0005
0.0072
$
15
$
10
$
132
Asthma­
Related
ER
Visits
0.0006
0.0004
0.0053
$
0
$
0
$
2
Asthma
Attacks
0.0517
0.0324
0.4566
$
2
$
1
$
19
Acute
Bronchitis
0.0054
0.0034
0.0479
<$
1
<$
1
$
3
Upper
Respiratory
Symptoms
0.0588
0.0368
0.5188
$
1
$
1
$
13
Lower
Respiratory
Symptoms
0.0598
0.0373
0.5270
$
1
$
1
$
8
Work
Loss
Days
0.4554
0.2849
4.0180
$
48
$
30
$
402
Minor
Restricted
Activity
Days
(
minus
asthma
attacks)
2.3681
1.3875
20.9184
$
116
$
68
$
1,023
Totals
$
16,621
$
10,193
$
142,867
Note
that
the
wide
discrepancy
between
the
per
ton
values
of
NO
x,
SO
2,
and
direct
PM
is
due
to
differences
in
their
contributions
to
ambient
concentrations
of
PM
2.5.

Example
of
estimating
total
national
program
benefits
When
applying
these
estimates,
please
make
sure
to
scale
these
estimates
to
the
appropriate
populations
(
see
Attachment
1).
Also
note
that
application
of
economic
values
to
the
estimates
of
incidences
should
be
done
with
caution.
Incidences
in
different
future
years
will
have
different
values
based
on
adjustments
to
WTP
for
growth
in
income
over
time.
See
Attachment
2
for
a
schedule
of
adjustment
factors
and
adjusted
WTP
values
to
be
applied
for
each
year
out
to
2030.
Adjustment
factors
should
not
be
applied
to
the
values
for
avoided
hospital
admissions,
as
these
are
cost­
of­
illness
estimates
and
not
WTP
estimates.
Likewise,
adjustment
factors
should
not
be
applied
to
the
value
of
work
loss
days,
as
this
is
a
wage­
based
estimate,
not
WTP.

While
the
base
value
for
a
mortality
incidence
is
$
6.1
million
(
1999$),
this
number
is
always
adjusted
downward
to
reflect
the
impact
of
discounting
over
the
assumed
5
year
lag
period
between
reductions
in
PM
concentrations
and
full
realization
of
reduced
mortality.
The
lagadjusted
base
VSL
is
$
5.8
million
(
1999$)
when
a
3%
discount
rate
is
assumed.
Thus
the
attached
table
reflects
income
adjustments
applied
to
these
lag
adjusted
base
values.

For
example,
if
one
were
to
estimate
the
value
of
a
nationwide
reduction
in
NO
x
emissions
of
1
million
tons
in
the
year
2005,
the
avoided
incidences
of
premature
mortality
would
be:

0.00196
(
from
Table
2)
*
0.829
(
from
Attachment
1)*
1,000,000
=
1,625
And
the
value
of
those
avoided
incidences
would
be:

1,625*$
6.27
million
(
from
Attachment
2)
=
$
10,189
million
Example
of
estimating
local
area's
contribution
to
national
benefits
If
the
reduction
in
NO
x
emissions
were
only
in
Seattle,
for
example,
and
the
reductions
were
only
10,000
tons
in
the
year
2001,
then
you
would
calculate
the
incidences
per
ton
of
NO
x
as
0.00196*
0.829*
10,000=
16.
Recall
however,
that
this
is
assuming
that
a
ton
of
NOx
reduced
in
Seattle
affects
PM
concentrations
everywhere
in
the
U.
S.
in
the
same
average
way.
Not
a
very
realistic
assumption,
which
is
why
it
is
better
to
apply
these
calculations
at
the
national
level.
The
important
thing
to
remember
is
that
populations
from
Attachment
1
should
only
be
used
to
adjust
for
differences
in
national
populations
across
years,
not
to
adjust
for
differences
in
regional
populations.

Details
on
Generation
of
Per
Ton
Estimates
HDD
with
REMSAD
Modeling
The
HDD
analysis
examined
the
impacts
in
2030
of
reducing
SO
2
emissions
by
141,000
tons
and
NO
x
emissions
by
2,570,000
tons,
as
well
as
a
109,000
ton
reduction
in
direct
PM
emissions.

SO2
Screening
Analysis
°
Input
the
population
weighted
change
in
annual
mean
sulfate
of
­
0.034
micrograms
per
cubic
meter
into
each
of
the
concentration­
response
functions
used
in
the
HD
Engine/
Diesel
Fuel
rule
benefits
analysis.
This
yields
estimates
of
the
health
effects
associated
with
the
SO
2
emission
reductions.
Based
on
2030
populations,
this
change
leads
to
the
estimated
reductions
in
health
effects
listed
in
the
second
column
of
Table
2.
Note
that
for
C­
R
functions
that
use
daily
average
PM
2.5
or
PM
10
levels,
use
of
the
annual
mean
as
a
proxy
for
daily
averages
will
over
or
underestimate
the
annual
incidence
by
a
small
amount
(
less
than
five
percent).
°
Divide
attributable
incidences
by
SO
2
tons
reduced
resulting
in
incidences
per
ton
of
SO
2
reduced
in
2030
as
listed
in
the
third
column
of
Table
2.

°
Multiply
incidences
per
ton
by
the
ratio
of
population
in
the
year
of
analysis
to
population
in
2030.
For
example,
multiplying
by
0.803
gives
the
correct
incidence
per
ton
estimate
for
2001
(
277/
345=
0.803),
reported
in
Table
1.

NOx
Screening
Analysis
°
Input
the
population
weighted
change
in
nitrate
of
­
0.388
micrograms
per
cubic
meter
into
each
of
the
concentration­
response
functions
used
in
the
HD
Engine/
Diesel
Fuel
rule
benefits
analysis.
This
yields
estimates
of
the
health
effects
associated
with
the
NO
x
emission
reductions.
Based
on
2030
populations,
this
change
leads
to
the
estimated
reductions
in
health
effects
listed
in
the
fourth
column
of
Table
2.
Note
that
for
C­
R
functions
that
use
daily
average
PM
2.5
or
PM
10
levels,
use
of
the
annual
mean
as
a
proxy
for
daily
averages
will
over
or
underestimate
the
annual
incidence
by
a
small
amount
(
less
than
five
percent).

°
Divide
attributable
incidences
by
NO
x
tons
reduced
resulting
in
incidences
per
ton
of
NO
x
reduced
in
2030
as
listed
in
the
fifth
column
of
Table
2.

°
Multiply
incidences
per
ton
by
the
ratio
of
population
in
the
year
of
analysis
to
population
in
2030.
For
example,
multiplying
by
0.803
gives
the
correct
incidence
per
ton
estimate
for
2001
(
277/
345=
0.803),
reported
in
Table
1.
°
Direct
PM
Screening
Analysis
°
Input
the
population
weighted
change
in
primary
PM
of
­
0.232
micrograms
per
cubic
meter
into
each
of
the
concentration­
response
functions
used
in
the
HD
Engine/
Diesel
Fuel
rule
benefits
analysis.
This
yields
estimates
of
the
health
effects
associated
with
the
PM
emission
reductions.
Based
on
2030
populations,
this
change
leads
to
the
estimated
reductions
in
health
effects
listed
in
the
sixth
column
of
Table
2.
Note
that
for
C­
R
functions
that
use
daily
average
PM
2.5
or
PM
10
levels,
use
of
the
annual
mean
as
a
proxy
for
daily
averages
will
over
or
underestimate
the
annual
incidence
by
a
small
amount
(
less
than
five
percent).

°
Divide
attributable
incidences
by
primary
PM
tons
reduced
resulting
in
incidences
per
ton
of
primary
PM
reduced
in
2030
as
listed
in
the
seventh
column
of
Table
2.
°
Multiply
incidences
per
ton
by
the
ratio
of
population
in
the
year
of
analysis
to
population
in
2030.
For
example,
multiplying
by
0.803
gives
the
correct
incidence
per
ton
estimate
for
2001
(
277/
345=
0.803),
reported
in
Table
1.

Background
on
REMSAD
2030
Modeling
The
Regulatory
Model
System
for
Aerosols
and
Deposition
(
REMSAD)
is
a
threedimensional
grid­
based
Eulerian
air
quality
model
designed
to
estimate
annual
particulate
concentrations
and
deposition
over
large
spatial
scales
(
e.
g.,
over
the
contiguous
U.
S.).
The
RIA
benefits
analysis
applies
the
modeling
system
to
the
entire
U.
S.
for
two
future­
year
scenarios:
a
2030
base
case
and
a
2030
HD
Engine/
Diesel
Fuel
control
scenario.
The
particulate
matter
species
modeled
by
REMSAD
include
a
primary
fine
fraction
(
corresponding
to
particulates
less
than
2.5
microns
in
diameter)
and
several
secondary
particulates
(
e.
g.,
sulfates,
nitrates,
and
organics).
PM
2.5
is
calculated
as
the
sum
of
the
primary
fine
fraction
and
all
of
the
secondary
particulates.

For
the
purposes
of
this
analysis,
we
separated
the
predicted
2030
change
in
the
primary
and
secondarily­
formed
components
of
PM
2.5
(
i.
e.,
sulfates
and
nitrates)
to
provide
attributable
health
effects
for
SO
2
and
NO
x.
The
change
in
primary
concentrations
is
represented
by
the
sum
of
PMFINE,
particulate
elemental
carbon
(
PEC),
and
total
organic
aerosols
(
TOA),
the
secondary
sulfate
change
is
represented
by
total
sulfate
(
TSO
4),
and
the
secondary
nitrate
change
is
represented
by
particulate
nitrate
(
PNO
3).
It
is
deemed
appropriate
for
this
"
scoping"
analysis
to
separate
these
predicted
concentrations
because
of
the
limited
interactions
of
secondary
sulfate
and
nitrates
within
the
modeling
system
and
the
limited
contribution
of
secondary
organic
aerosols
(
SOA)
to
TOA
(
i.
e.,
since
there
little
or
no
change
in
VOCs).
Table
2.
Summary
of
Results
from
2030
HD
Engine/
Diesel
Fuel
Health
Benefits
Analysis
Health
Outcome
SO
2
NO
x
PM
Avoided
Incidences
Avoided
Incidences
Avoided
Incidences
Total
Per
Ton
Total
Per
Ton
Total
Per
Ton
Premature
Mortality
All­
cause
premature
mortality
from
long­
term
exposure
441
0.00313
5,027
0.00196
3,007
0.02759
Chronic
Illness
Chronic
Bronchitis
(
pooled
estimate)
285
0.00202
3,243
0.00126
1,941
0.01781
Hospital
Admissions
COPD
49
0.00034
554
0.00022
331
0.00304
Pneumonia
59
0.00042
676
0.00026
404
0.00371
Asthma
46
0.00032
523
0.00020
313
0.00289
Total
Cardiovascular
143
0.00102
1,635
0.00064
978
0.00897
Asthma­
Related
ER
Visits
106
0.00075
1,209
0.00047
723
0.00663
Other
Effects
Asthma
Attacks
9,107
0.06459
103,905
0.04043
62,135
0.57005
Acute
Bronchitis
957
0.00679
10,874
0.00423
6,515
0.05977
Upper
Respiratory
Symptoms
10,348
0.07339
118,063
0.04594
70,601
0.64771
Lower
Respiratory
Symptoms
10,528
0.07467
119,760
0.04660
71,711
0.65790
Work
Loss
Days
80,163
0.56853
914,055
0.35566
546,744
5.01600
Minor
Restricted
Activity
Days
(
minus
asthma
attacks)
416,836
2.95628
4,763,239
1.85300
2,846,434
26.11407
Attachment
1.
Continental
United
States
Population
Estimates
for
Use
in
Analysis
of
Alternative
Years
Year
Population
(
millions)
Ratio
of
Analysis
Year
Population
to
2030
Population
1990
249
0.722
1991
252
0.730
1992
255
0.739
1993
258
0.748
1994
260
0.754
1995
263
0.762
1996
265
0.768
1997
268
0.777
1998
270
0.783
1999
272
0.788
2000
275
0.797
2001
277
0.803
2002
279
0.809
2003
281
0.814
2004
284
0.823
2005
286
0.829
2006
288
0.835
2007
291
0.843
2008
293
0.849
2009
295
0.855
2010
298
0.864
2011
302
0.875
2012
305
0.884
2013
307
0.890
2014
310
0.899
2015
312
0.904
2016
315
0.913
2017
317
0.919
2018
320
0.928
2019
322
0.933
2020
324
0.939
2021
327
0.948
2022
330
0.957
2023
333
0.965
2024
335
0.971
2030
345
1.000
Note:
Population
covers
the
48
contiguous
states
and
the
District
of
Columbia.
Attachment
2.
Longitudinal
Income
Adjustment
Factors
for
Value
of
Statistical
Life
and
Other
Health
Effect
Values
Year
VSL
Adjustment
Factor
Adjusted
VSL
(
million
1999$)
Chronic
Bronchitis
Value
Adjustment
Factor
Adjusted
Chronic
Bronchitis
Value
(
thousand
1999$)
Minor
Health
Effect
Value
Adjustment
Factor
1990
1.0000
$
5.80
1.0000
$
331.00
1.0000
1991
0.9920
$
5.75
0.9910
$
328.02
0.9972
1992
0.9981
$
5.79
0.9979
$
330.30
0.9993
1993
1.0031
$
5.82
1.0035
$
332.16
1.0011
1994
1.0128
$
5.87
1.0145
$
335.80
1.0045
1995
1.0170
$
5.90
1.0191
$
337.32
1.0059
1996
1.0244
$
5.94
1.0275
$
340.10
1.0085
1997
1.0343
$
6.00
1.0387
$
343.81
1.0119
1998
1.0393
$
6.03
1.0443
$
345.66
1.0136
1999
1.0436
$
6.05
1.0491
$
347.25
1.0150
2000
1.0487
$
6.08
1.0550
$
349.21
1.0168
2001
1.0544
$
6.12
1.0614
$
351.32
1.0187
2002
1.0609
$
6.15
1.0688
$
353.77
1.0209
2003
1.0675
$
6.19
1.0763
$
356.26
1.0231
2004
1.0742
$
6.23
1.0839
$
358.77
1.0254
2005
1.0805
$
6.27
1.0911
$
361.15
1.0275
2006
1.0869
$
6.30
1.0983
$
363.54
1.0296
2007
1.0933
$
6.34
1.1055
$
365.92
1.0317
2008
1.0996
$
6.38
1.1128
$
368.34
1.0338
2009
1.1060
$
6.41
1.1201
$
370.75
1.0359
2010
1.1124
$
6.45
1.1274
$
373.17
1.0380
2011
1.1965
$
6.94
1.2239
$
405.11
1.0647
2012
1.2041
$
6.98
1.2326
$
407.99
1.0670
2013
1.2130
$
7.04
1.2429
$
411.40
1.0697
2014
1.2229
$
7.09
1.2544
$
415.21
1.0728
2015
1.2326
$
7.15
1.2655
$
418.88
1.0757
2016
1.2414
$
7.20
1.2757
$
422.26
1.0783
2017
1.2498
$
7.25
1.2855
$
425.50
1.0809
2018
1.2574
$
7.29
1.2943
$
428.41
1.0831
2019
1.2646
$
7.33
1.3027
$
431.19
1.0853
2020
1.2783
$
7.41
1.3186
$
436.46
1.0893
2021
1.2784
$
7.41
1.3188
$
436.52
1.0894
2022
1.2847
$
7.45
1.3261
$
438.94
1.0912
2023
1.2910
$
7.49
1.3335
$
441.39
1.0931
2024
1.2972
$
7.52
1.3407
$
443.77
1.0949
