1
February
24,
2006
MEMORANDUM
Subject:
Considering
Household
Costs
on
an
Incremental
Basis
From:
Kira
Smith,
TAB
Jonathan
Shefftz,
Industrial
Economics,
Incorporated
Thru:
Dan
Olson
Team
Leader
To:
Record
EPA
stated
in
its
February
2006
Federal
Register
notice,
Small
Drinking
Water
Systems
Variances
 
Revision
of
Existing
National­
Level
Affordability
Methodology
and
Methodology
to
Identify
Variance
Technologies
that
are
Protective
of
Public
Health,
that
it
is
considering
using
the
incremental
costs
of
providing
water
rather
than
the
cumulative
costs
as
the
basis
for
its
affordability
determination.
This
is
a
change
from
the
Agency's
current
approach,
which
adds
incremental
costs
to
an
expenditure
baseline
to
determine
affordability.
An
incremental
approach
would
not
calculate
or
consider
household
water
bills,
nor
would
it
provide
a
ceiling
on
the
total
increase
in
household
costs
due
to
the
cumulative
effects
of
different
drinking
water
rules.
The
Agency
believes
that
an
incremental
approach
would
be
more
appropriate
than
the
current
cumulative
approach
for
the
following
reasons:
 
An
incremental
approach
is
consistent
with
EPA's
Science
Advisory
Board
(
SAB)
and
National
Drinking
Water
Advisory
Council
(
NDWAC)
recommendations.
 
An
incremental
approach
applies
only
to
the
National
Primary
Drinking
Water
Rule
(
NPDWR)
being
promulgated.
 
An
incremental
approach
avoids
the
possibility
that
the
order
in
which
EPA
promulgates
a
rule
determines
whether
a
rule
will
be
considered
affordable.
 
A
cumulative
approach
may
lead
to
large
cost
increases
at
the
system
level.
 
Doing
away
with
the
expenditure
baseline
of
a
cumulative
approach
eliminates
concerns
associated
with
the
regional
variability
of
water
bills
and
the
accuracy
of
baseline
values
over
time.
 
An
incremental
approach
allows
EPA
to
rely
on
a
more­
representative
data
source
to
calculate
median
household
income
(
MHI)
for
small
systems.

An
incremental
approach
is
consistent
with
SAB
and
NDWAC
recommendations
Currently,
EPA
determines
if
affordable
compliance
technologies
are
available
to
small
systems
by
comparing
the
current
household
cost
of
water
plus
the
estimated
additional
cost
to
comply
with
a
new
rule
to
an
affordability
threshold
equal
to
2.5
percent
of
MHI
(
about
$
1,000).
2
Both
the
SAB
and
the
NDWAC
recommended
that
EPA
eliminate
the
expenditure
baseline
and
evaluate
the
affordability
of
each
set
of
regulations
incrementally.
The
SAB
believed
that
the
use
of
an
expenditure
baseline
could
potentially
cause
early
regulations
to
be
considered
affordable,
whereas
later,
if
the
affordability
threshold
is
exceeded,
even
regulations
with
trivial
costs
could
be
found
unaffordable
to
small
systems.
The
NDWAC
recommended
that
EPA
replace
its
current
approach
with
an
incremental
approach
because
it
is
"
theoretically
sounder,
is
simpler
to
administer,
and
has
greater
transparency
than
the
current
EPA
method."

An
incremental
approach
applies
only
to
the
NPDWR
being
promulgated
An
incremental
approach
applies
only
to
the
regulation
for
which
affordability
is
being
evaluated.
In
contrast,
the
cumulative
approach
considers
not
just
the
cost
of
treatment
to
comply
with
the
new
standard
but
also
takes
into
account
the
current
household
cost
for
water.
Since
systems
ultimately
pass
water
treatment
costs
through
to
customers,
this
includes
costs
for
existing
water
system
improvements,
which
may
involve
treatment
for
odor
control,
taste,
or
other
items
not
regulated
under
NPDWRs,
as
well
as
costs
for
distributing
and
storing
water.
These
costs
may
not
be
relevant
for
determining
whether
a
system
can
afford
to
comply
with
NPDWRs.

An
incremental
approach
avoids
the
possibility
that
the
order
in
which
EPA
promulgates
a
rule
determines
whether
a
rule
will
be
considered
affordable
An
incremental
approach
(
does
not
include
baseline
expenditures)
avoids
the
possibility
that
the
order
in
which
EPA
promulgates
NPDWRs
is
a
key
determinant
in
whether
a
rule
will
be
considered
affordable.
Exhibit
1
illustrates
the
difference
over
time,
and
for
each
size
class,
between
the
current
methodology
and
an
incremental
methodology
that
sets
affordability
at
0.75
percent
of
median
household
income,
i.
e.,
about
$
300.
In
each
case
the
incremental
threshold
is
less
than
half
of
the
cumulative
expenditure
margin.
(
Note
that
the
incremental
thresholds
in
Exhibit
1
are
applied
to
incremental
treatment
costs
for
the
median­
sized
system;
applying
them
instead
to
the
10th­
percentile
system
would
further
decrease
affordability.)
Conceptually,
as
EPA
promulgates
additional
rules
a
cumulative
approach
might
act
as
a
brake"
on
future
water
bill
increases
whereas
an
incremental
approach
would
not
have
this
effect.
Data
that
describe
projected
future
income
levels
and
water
rates
suggest
that
this
difference
could
become
a
factor
in
future
decision
making.
3
Exhibit
1.
Projected
Affordability
Thresholds
Size
Class
Median
Household
Income
(
Nominal)
Median
Water
Bill
(
Nominal)
Expenditure
Margin
Based
on
2.5%
Cumulative
Threshold
0.75%
Incremental
Threshold
2005:
25­
500
$
44,544
$
299
$
810
$
330
501­
3,300
$
40,872
$
294
$
730
$
310
3,301­
10,000
$
42,459
$
285
$
780
$
320
2010:
25­
500
$
52,516
$
369
$
940
$
390
501­
3,300
$
48,976
$
386
$
840
$
370
3,301­
10,000
$
50,103
$
321
$
930
$
380
2015:
25­
500
$
62,036
$
455
$
1,100
$
470
501­
3,300
$
58,803
$
508
$
960
$
440
3,301­
10,000
$
59,239
$
363
$
1,100
$
440
Note:
1.
All
income
figures
in
this
table
are
inflated
to
the
dates
listed
from
the
2000
U.
S.
Census
by
first
adjusting
them
at
the
ratio
between
the
1999
and
2004
national
MHI
as
estimated
here:
http://
www.
census.
gov/
hhes/
www/
income/
histinc/
h06ar.
html.
They
are
then
adjusted
from
2004
at
the
ratio
of
the
CPI
values
for
the
respective
dates.
2.
Expenditure
Margin
and
Incremental
Threshold
estimates
are
rounded
to
nearest
ten.

Exhibit
2
lists
projected
MHI
and
water
bill
growth
rates.
Exhibit
3
illustrates
projected
water
costs
relative
to
income.
1
If
these
projected
trends
continue,
the
two
smaller
size
classes
might
expect
water
bills
to
grow
faster
than
income.
This
means
that
a
cumulative
affordability
methodology
would
be
more
likely
to
find
future
rules
unaffordable
the
further
in
the
future
they
are
promulgated,
as
more
of
that
methodology's
expenditure
baseline
would
be
consumed
by
rising
water
bills,
leaving
less
room
for
compliance
costs.

1
A
few
caveats
apply
to
these
data
and
to
any
conclusions
to
be
drawn
from
them:
°
The
future
projection
of
water
bills
is
extrapolated
from
only
a
five­
year
historical
period;
°
The
future
projection
is
based
upon
a
relatively
small
sample
size
that
may
not
be
representative
of
all
small
water
systems;
and,
°
Small
system
MHI
growth
may
not
equate
to
Consumer
Price
Index
(
CPI)
growth
in
the
future.
From
2000
to
2001,
the
national
average
MHI
increased
by
approximately
0.6
percent,
while
the
CPI
increased
by
approximately
2.8
percent.
4
Exhibit
2.
Projected
Growth
Rates
for
Water
Bills
and
Income
Size
Class
Nominal
Cumulative
Growth
Rate
Nominal
Annual
Growth
Rate
Annual
Rate
of
Inflation
Real
Annual
Growth
Rate
Projected
Growth
Rate
2005­
10
MHI
(
Between
1990
and
2000
U.
S.
Census)
25­
500
49.35%
4.09%
3.00%
1.06%
3.35%
501­
3,300
54.29%
4.43%
3.00%
1.39%
3.68%
3,301­
10,000
49.62%
4.11%
3.00%
1.08%
3.37%
Water
Bills
(
Between
1995
and
2000
CWSS)
25­
500
24.40%
4.46%
2.47%
1.94%
4.25%
501­
3,300
32.81%
5.84%
2.47%
3.29%
5.62%
3,301­
10,000
13.76%
2.61%
2.47%
0.14%
2.40%
Notes:
1.
Real
annual
growth
rate
is
based
on
nominal
annual
growth
rate
net
of
annual
rate
of
inflation,
which
in
turn
is
based
on
CPI
growth
over
the
relevant
historical
period.
2.
Projected
growth
rate
is
based
on
combining
the
real
annual
growth
rate
with
an
average
of
estimated
future
CPI
growth,
which
in
turn
is
based
on
an
average
of
forecasts
from
the
Congressional
Budget
Office
and
the
Office
of
Management
and
Budget.
Sources:
1.
1990
U.
S.
Census
2.
2000
U.
S.
Census
3.
1995
Community
Water
Systems
Survey
4.
2000
Community
Water
Systems
Survey
5
The
cumulative
approach
may
lead
to
large
cost
increases
The
cumulative
methodology
could
lead
to
larger
cost
increases
at
the
system
level,
and
thereby
fail
in
its
intended
role
to
act
as
a
"
brake"
on
water
bill
increases
for
these
systems.
This
outcome
could
occur
because
the
expenditure
baseline
increases
as
a
result
of
a
new
regulation
by
an
amount
equal
to
the
average
cost
of
the
regulation
across
all
households
in
a
particular
size
category.
This
cost­
averaging
occurs
because
of
the
timing
with
which
EPA
promulgates
new
rules
and
the
frequency
with
which
EPA
currently
administers
the
Community
Water
System
Survey
(
CWSS,
the
baseline's
primary
data
source).
The
CWSS
is
a
national
survey
that
EPA
has
collected
fairly
regularly
since
1977.
The
survey
requests
operating
and
financial
information
from
a
sample
of
community
water
systems
in
the
United
States.
EPA
conducted
its
last
CWSS
in
2000,
although
the
results
were
not
fully
tabulated
and
available
until
several
years
later.
Since
2000,
however,
EPA
has
promulgated
a
number
of
new
drinking
water
rules.
As
a
result,
except
in
the
period
immediately
following
the
publication
of
CWSS,
the
water
bill
data
from
the
CWSS
would
understate
the
actual
baseline
because
it
would
not
reflect
the
costs
of
new
rules.
Therefore,
EPA
calculates
the
approximate
number
of
households
in
systems
serving
less
than
10,000
people
and
averages
costs
for
new
rules
across
these
households.
EPA
must
average
across
all
households
instead
of
just
those
in
systems
affected
by
each
rule
because
of
an
inability
to
identify
each
system
whose
costs
are
affected
by
each
rule.
Exhibit
4
illustrates
how
updating
the
baseline
resulted
in
relatively
modest
increases
in
most
cases
following
the
promulgation
of
five
recent
drinking
water
regulations.
However,
new
regulations
typically
impose
much
higher
costs
on
the
relatively
small
subset
of
the
approximately
50,000
small
systems
in
the
United
States
that
must
modify
or
install
treatment
technology
to
comply
with
the
regulations.
For
example,
only
5.5
percent
of
the
systems
in
each
small
system
size
category
are
expected
to
modify
or
install
treatment
technology
in
response
to
the
recently
established
maximum
contaminant
level
for
arsenic.
In
other
words,
unless
a
large
percentage
of
small
systems
incur
higher
treatment­
related
costs,
the
increase
in
the
expenditure
Exhibit
3.
Water
Costs
Relative
to
Income
over
Time
(
Nominal
Dollars)

0.0%
0.2%
0.4%
0.6%
0.8%
1.0%
1.2%

2005
2006
2007
2008
2009
2010
2011
2012
2013
201
Year
Nominal
Water
Bill
as
a
%
of
Nominal
MHI
25­
500
501­
3,300
3,301­
10,000
6
baseline
will
remain
relatively
modest
and
the
cumulative
methodology's
"
braking"
potential,
at
least
for
the
subset
of
systems
that
require
additional
treatment,
will
be
constrained.

Exhibit
4.
Updating
the
Expenditure
Baseline
Size
Class
25­
500
501­
3,300
3,301­
10,000
Households
1,985,492
7,695,205
9,982,561
Arsenic
Total
Cost
$
21,983,158
$
39,665,263
$
33,333,158
Update
to
Baseline
$
11.07
$
5.15
$
3.34
DBP­
I
Total
Cost
$
31,780,878
$
113,142,868
$
168,788,558
Update
to
Baseline
$
16.01
$
14.70
$
16.91
Filter
Backwash
Total
Cost
$
202,551
$
550,520
$
594,064
Update
to
Baseline
$
0.10
$
0.07
$
0.06
Long
Term
1
Total
Cost
$
3,360,836
$
11,333,536
$
9,422,178
Update
to
Baseline
$
1.69
$
1.47
$
0.94
Uranium
Total
Cost
$
7,664,974
$
5,757,186
$
6,666,514
Update
to
Baseline
$
3.86
$
0.75
$
0.67
Notes:
1.
Total
costs
are
reported
as
calculated
in
the
economic
analysis
and
then
updated
to
September
2005
dollars.
2.
The
increase
to
the
baseline
is
the
total
cost
divided
by
the
number
of
households
in
a
given
size
class
(
estimated
by
the
May
2001
Drinking
Water
Baseline
Handbook).
Therefore,
changes
in
the
demographic
estimates
(
i.
e.,
number
of
systems
and
number
of
households
per
system)
do
not
affect
total
cost,
but
do
affect
the
baseline
updates.

Eliminating
a
baseline
eliminates
regional
water
bill
variability
Doing
away
with
the
expenditure
baseline
eliminates
concerns
associated
with
the
regional
variability
of
water
bills
and
the
accuracy
of
baseline
values
over
time.
As
Exhibit
5
illustrates
(
with
figures
inflated
to
the
September
2005
Producer
Price
Index),
water
bills
vary
across
geographic
regions
in
all
small
system
size
classes.
These
differences,
ranging
from
about
40
to
80
percent
depending
on
the
size
class,
illustrate
the
challenge
of
setting
a
baseline
for
nationwide
application.

Exhibit
5.
Regional
Median
Water
Bills
Size
Class
Midwest
Northeast
South
West
25­
500
$
233
$
288
$
311
$
359
501­
3,300
$
219
$
320
$
288
$
394
3,301­
10,000
$
248
$
351
$
276
$
298
An
incremental
approach
allows
EPA
to
rely
on
a
more
representative
data
source
to
calculate
median
household
income
for
small
systems
EPA
currently
calculates
the
cumulative
affordability
threshold
based
on
MHI
using
the
results
of
the
1995
CWSS.
The
CWSS
database
includes
the
necessary
information
to
calculate
household
water
bills,
which
is
a
key
component
under
the
current
cumulative
affordability
7
methodology.
The
CWSS
does
not
provide
income
measures
for
small
drinking
water
systems,
but
it
does
provide
information
on
the
geographical
area
that
the
system
serves.
Therefore,
the
Agency
can
link
U.
S.
Census
data
to
each
system
in
the
CWSS.
Each
CWSS
collects
data
from
several
hundred
small
systems,
which
represent
about
one
percent
of
the
entire
universe
of
small
systems.
Furthermore,
the
2000
CWSS
sampling
was
not
intended
to
be
randomly
distributed.
Instead,
the
goal
was
to
achieve
a
higher
response
rate
by
dispatching
interviewers
to
individual
systems
for
on­
site
data
collection
(
rather
than
relying
on
mailed
forms).
The
need
to
maximize
the
efficiency
of
in­
person
data
collection
dictated
a
cluster
pattern.
By
contrast,
the
Safe
Drinking
Water
Information
System
(
SDWIS)
is
EPA's
national
regulatory
compliance
database.
SDWIS
is
a
census
of
all
public
water
systems
that
is
updated
continually,
containing
contact,
location,
inventory,
and
violation
information.
SDWIS
therefore
covers
the
entire
universe
of
small
drinking
water
systems
across
the
nation.
Just
like
the
CWSS,
SDWIS
does
not
provide
income
measures
for
small
drinking
water
systems,
but
it
does
provide
information
on
each
system's
location.
U.
S.
Census
data
can
thereby
be
linked
to
each
system
in
SDWIS.
Unlike
the
CWSS,
SDWIS
does
not
contain
the
information
required
to
estimate
household
water
bills.
For
the
cumulative
affordability
methodology,
the
CWSS
was
necessary
for
its
water
bill
data,
but
EPA's
option
for
public
comment
would
change
its
small
system
affordability
determination
to
an
incremental
methodology.
Under
such
a
methodology,
water
bill
data
are
no
longer
necessary,
eliminating
the
need
to
rely
on
CWSS.
Therefore,
adopting
SDWIS
for
the
income
estimation
component
of
the
affordability
methodology
is
now
feasible,
as
well
as
preferable
given
its
greater
accuracy
from
the
universal
coverage
of
small
systems.
Exhibit
6
calculates
MHI
for
systems
that
appear
in
both
the
2000
CWSS
and
January
2004
SDWIS.
The
goal
is
to
determine
whether
income
calculations
for
the
same
subset
of
systems
are
similar
when
based
upon
different
techniques
for
estimating
each
system's
service
area.
The
exhibit's
first
column
(
i.
e.,
"
Median
CWSS")
reports
the
MHI
for
all
systems
in
both
the
2000
CWSS
and
SDWIS
based
on
the
census
tracts
served
by
the
system,
as
identified
in
the
2000
CWSS.
The
third
column
(
i.
e.,
"
Median
SDWIS")
calculates
MHI
for
the
same
systems,
yet
uses
the
zip
code
contact
information
(
as
provided
in
earlier
SDWIS
freezes),
as
a
proxy
for
the
area
served.
In
absolute
terms,
and
especially
relative
to
standard
deviation,
Exhibit
6
suggests
that
the
SDWIS
zip
codes
(
instead
of
the
2000
CWSS
census
tracts)
can
reasonably
approximate
MHI.
8
Exhibit
6.
Systems
in
Both
2000
CWSS
and
SDWIS
Size
Class
Median
CWSS
(
by
tract)
Standard
Deviation
Median
SDWIS
(
by
zip)
Standard
Deviation
Sample
Size
25­
500
$
45,921
$
19,000
$
44,657
$
17,783
212
501­
3,300
$
40,280
$
13,948
$
39,668
$
13,238
152
3,301­
10,000
$
41,594
$
14,680
$
41,839
$
14,967
123
Notes:
1.
A
total
of
521
small
systems
appear
in
both
SDWIS
and
2000
CWSS,
but
only
487
had
zip
codes
that
could
be
linked
to
the
U.
S.
Census
MHI
tables
(
with
the
remainder
having
unavailable
or
inaccurate
zip
code
information).
2.
SDWIS
figures
are
based
on
the
"
freeze"
as
of
January
2004,
reflecting
data
from
fourth
quarter
2003.
3.
All
income
figures
in
this
table
are
inflated
to
the
present
from
the
2000
U.
S.
Census
by
first
adjusting
them
at
the
ratio
between
the
1999
and
2004
national
MHI
as
estimated
here:
http://
www.
census.
gov/
hhes/
www/
income/
histinc/
h06ar.
html.
They
are
then
adjusted
from
2004
to
September
2005
at
the
ratio
of
the
CPI
values
for
the
respective
dates.

The
similarity
of
these
income
estimates
supports
using
the
zip
code
identifier
in
SDWIS
as
the
basis
for
pairing
U.
S.
Census
income
data
with
small
drinking
water
systems,
for
which
SDWIS
has
almost
universal
coverage.
Exhibit
7
compares
the
MHI
estimates
using
the
subset
of
systems
both
in
the
2000
CWSS
and
SDWIS.
The
second
MHI
column
("
All
SDWIS")
represents
the
median
household
in
the
median
system,
as
measured
using
SDWIS
data.

Exhibit
7.
Comparison
of
SDWIS
MHI
for
Systems
in
2000
CWSS
to
All
SDWIS
Systems
Size
Class
SDWIS
in
CWSS
CWSS
Sample
Size
All
SDWIS
SDWIS
Sample
Size
25­
500
$
44,657
212
$
44,544
27,178
501­
3,300
$
39,668
152
$
40,871
13,265
3,301­
10,000
$
41,839
123
$
42,459
4,308
Notes:
1.
SDWIS
reports
(
as
of
the
January
2004
"
freeze")
a
total
of
30,132,
14,263,
and
4,661
systems
in
each
of
the
three
smallest
size
categories
(
respectively),
but
those
systems
that
were
unable
to
have
income
estimates
developed
for
them
(
because
of
unavailable
or
inaccurate
zip
code
information)
were
dropped
from
the
calculations
for
this
exhibit.
2.
All
income
figures
in
this
table
are
inflated
to
the
present
from
the
2000
U.
S.
Census
by
first
adjusting
them
at
the
ratio
between
the
1999
and
2004
national
MHI
as
estimated
here:
http://
www.
census.
gov/
hhes/
www/
income/
histinc/
h06ar.
html.
They
are
then
adjusted
from
2004
to
September
2005
at
the
ratio
of
the
CPI
values
for
the
respective
dates.
