REGULATORY IMPACT ANALYSIS:
BENEFITS AND COSTS OF PROPOSED NATIONAL
PRIMARY DRINKING WATER REGULATIONS
FOR INORGANIC CHEMICALS
PREPARED FOR:
Environmental Protection Agency
Office of Drinking Water
Washington, D.C. 20460
PREPARED BY:
Wade Miller Associates, Inc.
1911 North Fort Myer Drive
Arlington, VA 22209
PREPARED UNDER:
EPA Contract No. 68-03-3348
Work Assignment No. 2-12
Under Subcontract to The Cadmus Group
Mr. Brian Rourke, Project Officer
Mr. Carl Kessler, Technical Project Monitor
-------
-------
TABLE OF CONTENTS
PAGE
1. Introduction and Summary !_!
1.1 Introduction 2.-1
1.2 Problem Definition ' ^.^
1.3 Market Imperfections, The Need for Federal
Regulation, and Consideration of Regulatory
Alternatives !_3
1.4 Assessment of Total Costs 1-3
1.5 Assessment of Benefits 1_5
1.6 Regulatory Flexibility, and Paperwork Reduc-
tion Analyses 1-6
1.7 Summary of Costs and Benefits 1-7
2. Problem Definition 2-i
2.1 Introduction 2-i
2.2 Health Effects 2_4
2.3 Occurrence 2-5
2.4 Treatment Technologies 2-14
2.5 Analytical Methods 2-18
3. Market Imperfections, The Need for Federal Regula-
tion, and Consideration of Regulatory Alternatives 3-1
3.1 Introduction 3_!
3.2 The Nature of the Imperfections 3-2
3.3 The Need for Federal Regulation 3-6
3.4 Consideration of Regulatory Alternatives 3-10
-------
TABLE OF CONTEWTS (Continued)
PAGE
4. Assessment of Costs 4-1
4.1 Introduction 4-1
4.2 Major Assumptions Used in Cost Estimates 4-1
4.3 Treatment Costs for lOCs 4-2
4.4 Waste Disposal Costs 4-10
4.5 Uncertainty in Estimates of National Costs 4-10
4.6 Monitoring Costs 4-15
4.7 Costs to State Programs 4-28
5. Assessment of Benefits 5-1
5.1 “Damages Avoided” versus Total Benefits 5—1
5.2 Benefits of a Margin of Safety 5-1
5.3 Induced Efficiency Improvements in the
Water Industry 5-3
5.3 Aggregate Analysis of Health Benefits 5-4
6. Regulatory Flexibility Analysis and Paperwork
Reduction Analysis 6-1
6.1 Regulatory Flexibility Analysis 6-].
6.2 Paperwork Reduction Analysis 6-7
7. Summary of Costs, Benefits, and Uncertainty 7-1
7.1 Total Incremental National Costs 7-].
7.2 Total Incremental National Benefits 7-4
-------
LIST OF EXHIBITS
EXHIBIT PAGE
1-1 Incremental Cost of IOC Rule at Proposed
MCL Alternatives 1-4
1-2 Incremental Impacts of Proposed IOC Reg-
ulations At Proposed Alternatives 1-8
2—1 Summary of Health Effects for Inorganic
Chemicals 2-2
2—2 General Form of Delta Log - Normal Distri-
bution 2-7
2-3 Standardized Numbers of Public Water Sys-
tems and Populations Served by Source arid
Size Categories 2-9
2—4 Conversion of Occurrence Data to Standard-
ized System Size Categories (by Population
Served) 2-10
2-5 Community Water Systems with bC Contamina-
tion Greater Than or Equal to Current MCLs 2-11
2—6 Summary of Co-Occurrence in All Community
Water Systems 2-13
2-7 lOCs for Which Conventional Treatment is
Effective (Greater than 70 Percent Removal) 2-15
2-8 lOCs for Which Lime Softening is Effective
Treatment (Greater than 90 Percent Removal) 2—16
2—9 lOCs for Which Ion Exchange Treatment is
Effective (Greater than 90 Percent Removal) 2-16
2—10 Analytical Methods for lOCs 2—19
3—1 Annual Family Bills for Selected Utilities
1952—1984
- and -
Percent of Median Family Income Spent on
Selected Utilities 1952—1984 3—3
3-2 MCL Alternative for lOCs (ug/l) 3-13
-------
LIST OF EXHIBITS (Continued)
EXHIBIT PAGE
4-1 FRDS IOC Violations in Community Water
Systems Serving More than 50,000 Persons 4-3
4—2 Estimated Treatment and Waste Disposal
Costs at Most Stringent MCL Alternatives 4-4
4—3 Estimated Treatment and Waste Disposal
Costs at Proposed MCL Alternatives 4-7
4—4 Standard Errors of Independent Variables in
bC Cost Calculation (As a Percent of Mean
Values) 4—14
4-5 Standard Errors in Compliance Cost Estimates
(As Decimal Fractions of Mean Values) 4-15
4-6 Average Annual Monitoring Costs of Proposed
IOC Regulations (In Millions of 1986 Dollars) 4-16
4-7 Inorganic Chemicals: Proposed Monitoring
Requirements 4-19
4-8 Asbestos: Proposed Monitoring Requirements 4-22
4-9 Nitrate/Nitrite: Proposed Monitoring Re-
quireinents 4-23
4—10 Analytical Cost Assumptions —— lOCs 4-27
5-1 Size Distribution of Community Water Systems 5-6
5-2 Reduction in Population Exposed to Inorganic
Chemicals at Most Stringent MCL Alternative
(In Thousands) 5-8
5-3 Reduction in Population Exposed to Inorganic
Chemicals at Preferred MCLJ Alternatives
(In Thousands) 5-9
5-4 Estimated Population Exposed (Millions) to
Various Concentrations of Asbestos in Community
Drinking Water Systems 5-10
5-5 Estimated Baseline Cancer Cases Due to
Asbestos Contamination in Community Drinking
Water Systems 5-12
-------
LIST OF EXHIBITS (Continued)
EXHIBIT PAGE
6-]. Estimated Number of Small Entities Affected
by Proposed IOC Rule 6-4
6-2 Production Cost Increase (Macro—Level
Analysis) 6-5
6-3 Production Cost Increase by System Size
Category (Micro-Level Analysis) 6-6
7-1 Estimated Impact of Regulating lOCs at Most
Stringent MCL Alternatives 7-2
7-2 Estimated Impact of Regulating lOCs at
Proposed MCL Alternatives 7-3
-------
1. INTRODUCTION AND SUMMARY
1.1 Introduction
This report contains an analysis of the costs and benefits of
controlling inorganic chemical contaminants (lOCs) in drinking
water through the promulgation of regulations for maximum contam-
inant level goals (MCLGs) and maximum contaminant levels (MCLs).
This regulatory impact analysis (RIA) was prepared in accordance
with Executive Order 12291 which requires that the costs and bene-
fits of all major rules be examined and compared. The major
topical areas covered in this RIA are as follows:
o Problem Definition:
o Market Imperfections, the Need for Federal Regulation, and
Consideration of Regulatory Alternatives;
o Assessment of Total Costs;
o Assessment of Benefits;
o Regulatory Flexibility and Paperwork Reduction Analyses;
and,
o A Summary of Costs and Benefits.
This initial chapter contains a summary of results. Detailed
analyses of costs and benefits are included in Chapters 4 and 5.
1.2 Problem Definition
In accordance with the requirements of the Safe Drinking Water
Act Amendments of 1986 (SDWA) EPA is developing new or revised
drinking water regulations for the following lOCs: asbestos, bar-
ium, cadmium, chromium, mercury, nitrate, nitrite, and selenium.
Available evidence suggests that adverse health effects
attributable to bC exposure generally involve acute or chronic
sub-lethal endpoints of toxicity. The Agency has developed long-
term drinking water equivalent levels (DWELs) which serve as the
basis for determining maximum contaminant level goals (MCLCs) for
lOCs.
Three contaminants evaluated under this proposal have also
exhibited evidence carcinogenic effects; these are cadmium, chrom-
ium, and asbestos. The strength of evidence of carcinogenicity yj
drinkin water ex osure is not considered significant by the Agency
for these contaminants however. Therefore, MCLG5 are being
1—1
-------
proposed as though they had limited or no evidence of
carcinogenicity as drinking water contaminants.
Ma-icr Assumptions Used in Analysis
Three critical assumptions serve as the basis for the impacts
estimated in this analysis. First, it is necessary to avoid
misallocation of impacts of regulations currently in place to those
resulting from revised regulations. Because all contaminants
except asbestos and nitrite are currently regulated, it was assumed
that no impacts would be attributable to revised MCLs set equal to
or greater than those currently in place.
At the most stringent (i.e., lowest) MCL alternatives, an
assessment of net incremental cost impacts is necessary for all
lOCs evaluated (recognizing that significant co-occurrence may be
involved at such low concentrations). At “preferred” MCL alter-
natives (i.e., those corresponding most closely with proposed
MCLGs), it is estimated that net incremental costs will result for
only cadmium.
The second major assumption concerns the consolidation of
occurrence data across different system size categories. In
developing estimates of occurrence probabilities, raw survey data
from several size categories were typically combined and analyzed
as a group. This procedure resulted in invariant occurrence
probabilities across system size categories for most lOCs.
However, because the estimates are based on sampling data from
finished water, and since systems serving greater than 50,000
persons generally have more effective treatment processes, more
comprehensive monitoring, and a higher level of operator expertise
in place, it was assumed that bC occurrence in large systems is
probably negligible compared to that in smaller systems.
Finally, occurrence data on mercury in surface water systems
is subject to extreme uncertainty. For this reason, estimates
which include mercury occurrence as calculated, as well as esti-
mates using the assumption that mercury occurrence is zero are
presented in the analysis.
The phenomenon of co-occurrence was evaluated based on
combined probabilities of occurrence of individual contaminants at
the preferred MCLs. It was estimated that approximately one
percent of affected systems have occurrence of two contaminants.
On this basis, it was assumed that an even smaller number of
systems would be likely to have simultaneous occurrence of three
or more contaminants. Based on these results, it was assumed that
the degree of co-occurrence of lOCs (and resulting effects on the
aggregate national impacts) is negligible and no attempt was made
to adjust the calculations at the proposed MCLs.
1—2
-------
Under the assumptions discussed above, it is estimated that
the incremental impacts of the revised regulations will affect
approximately 192 coriununity and non-transient, non-community water
systems. This translates into a national population exposure
estimate of 170,000 persons.
1.3 Market Imperfections, The Need for Federal Recrulatjon, and
Consideration of Req-ulatory Alternatives
EPA has proposed the IOC regulations in accordance with the
requirements of the SDWA Amendments of 1986. The SDWA mandates
that EPA publish MCLGs and promulgate national primary drinking
water regulations (NPDWR) for 83 specific contaminants prior to
June, 1989. The inorganic contaminants evaluated in this document
are included on this list.
Since the proposed standards and monitoring requirements for
lOCs have been mandated by statute, EPA is limited in its consid-
eration of alternative regulatory approaches and strategies for
implementation of the SDWA. EPA does not have the flexibility to
consider taking “no action,” nor can the Agency propose that states
establish the necessary standards and monitoring requirements based
on Federal guidance. Under these constraints, the Agency is
committed to setting explicit MCLs for each contaminant as close
to the MCLG “as is feasible,” or in the absence of feasible
analytical methodology, a treatment requirement.
1.4 Assessment of Total Costs
Estimates of aggregate national costs for proposed alternative
MCLs for lOCs are based on individual results for each contaminant
evaluated. Because nitrite levels in drinking water are considered
to be negligible relative to nitrate levels, and due to nitrite’s
tendency to be oxidized to nitrate during disinfection, it was
assumed that incremental impacts of the nitrite regulation beyond
those calculated for nitrate would be negligible and no further
analyses were performed for this chemical. As indicated above, it
was assumed that treatment and waste disposal costs for removal of
asbestos would be subsumed by the corrosion control and surface
water treatment rules.
The basic algorithm utilized in obtaining estimates for total
national treatment costs is relatively straightforward. For each
system size category, for both ground and surface systems, the
number of systems with contamination above the MCL alternative is
estimated. The number of systems is then. merged with a decision
matrix which predicts the relative likelihood that a system of a
given size and source will choose various treatment technologies
or other compliance options. These estimates are then multiplied
1—3
-------
by the appropriate engineering costs developed for each technology
or option.
Net incremental capital costs associated with treatment and
waste disposal at the most stringent MCL alternatives in systems
serving fewer than 50,000 persons are approximately $40.8 billion,
excluding mercury. The incremental operation and maintenance (O&M)
costs are approximately $4.5 billion per year, resulting in annu-
alized costs of $7.3 billion using a three percent discount rate,
$7.8 using five percent, and $8.4 billion using a seven percent
discount rate over 20 years. These may be overestimates due to the
potential for co-occurrence at these concentrations.
Capital costs associated with the proposed MCL alternatives
for systems serving less than 50,000 persons are approximately $73
million, with O&M costs of $6 million per year. This results in
annualized treatment and waste disposal costs of $1]. million per
year (Exhibit 1-1).
Exhibit i-i
Incremental Cost of IOC Rule at
Proposed MCL Alternatives
Capital Costs 0 & N
( Million 1986 $) IMillion 1986 S/Year )
Treatment 40 4
Waste Disposal 33 2
Monitoring 0 2
State Programs o 0
Total 73 8
Total Annualized
Costs (@ 3% Over 20 years) $11 Million/Year (excluding
monitoring costs)
$13 Million/Year (including
monitoring costs)
The standard errors of the national costs of compliance were
estimated for cadmium at an MCL of 5 ugh. The estimates of
uncertainty apply to capital, O&M, and annualized costs, on the
assumption that errors in capital and O&M costs are perfectly
1—4
-------
correlated. It also applies to the compliance cost (i.e., the
cost of meeting the proposed MCL minus the cost of meeting the
NIPDWR) . The errors in cost range from 20 to 28 percent of the
mean value for cadmium.
Monitoring cost estimates for each contaminant include ana-
lytical costs averaged over an 1]. year period and were derived
using the proposed monitoring requirements outlined in the draft
Phase II proposal. Models were developed for calculating high and
low bound estimates for total annual monitoring costs for each
contaminant.
The estimated total cost of monitoring for inorganic con-
taminants under the interim regulations is reported in the Public
Water System Supervision Program (PWSSP) Information Collection
Request (ICR) as $3.3 million. It is estimated that $2.0 million
of the interim regulation’s costs would be incurred due to moni-
toring for the sub-set of contaminants evaluated in this analysis.
Incremental monitoring costs for lOCs can be estimated by
subtracting the costs associated with the interim regulations from
those calculated for the proposed revised regulations. This
results in average incremental monitoring costs of $2.1 million
per year under the low bound scenario, and $3.3 million per year
under the high bound at the proposed MCL alternatives. In addition
to these annual monitoring costs, it is estimated that there will
be a one-time cost to monitor the unregulated contaminants that
ranges from $0.2 to $1.6 million.
An estimate of the total costs to state programs for the
entire Phase II regulatory package (i.e., including both lOCs and
synthetic organic chemicals) is included in the draft Regulatory
Impact Analysis for Synthetic Organic Chemiàals.
1.5 Assessment of Benefits
Traditionally, benefits of removing contamination from
drinking water are expressed in terms of cases of disease avoided.
Since all lOCs are being regulated based on sub-lethal health
effects, and no usable dose/response data were available for these
contaminants, it was impossible to calculate the number of cases
of adverse health effects avoided. For this reason, benefits
associated with the regulation of lOCs in drinking water are
presented only in terms of a reduction in the population exposed
under the various MCL alternatives.
The net incremental reduction in exposure derived through
adoption of the most stringent MCL alternatives for each contam-
inant is approximately 102 million persons in systems serving fewer
than 50,000 persons. Excluding the systems estimated to have
occurrence of mercury, a net reduction of approximately 99 million
1—5
-------
persons exposed is estimated. The estimated net incremental
reduction in exposure derived through adoption of the proposed MCL
alternatives for cadmium is approximately 170,000 persons in
systems serving fewer than 50,000 persons. In addition to reduced
exposure, the total annual baseline cancer cases due to asbestos
exposure were calculated depending on assumptions used regarding
population exposed and average concentration in affected systems.
It is estimated that between 0.06 and 0.3 cases occur annually in
community water systems.
Note that the above estimates of health damages avoided
represent an underestimate of the total benefits because certain
intangible benefits are not included. Of particular importance is
the value of an extra margin of safety -- a warranty -- which pro-
vides assurance to consumers, and to society as a whole, that it
can be taken for granted that the water is safe to drink.
1.6 Reqillatory Flexibility, and Paperwork Reduction Analyses
Because of the health risks associated with inorganic contam-
inants, the nature of their occurrence in public water supplies,
and the generally limited treatment currently in place, the revised
standards for these contaminants are likely to affect small surface
water systems.
EPA guidelines on compliance with the Regulatory Flexibility
Act indicate that, in general, a “substantial” number of small
entities is more than 20 percent of the total. Of the estimated
199,390 public water supplies serving fewer than 50,000 persons,
192 (0.1 percent) systems will incur incremental treatment and
waste disposal costs as a result of the revised IOC regulations
(assuming MCLs are set equivalent to the MCLGs for non-carcino-
gens). Therefore, by the 20 percent rule, the proposed bC
regulations would not affect a “substantial” number of small water
utilities at the proposed MCL alternatives.
Under the RFA, annual costs of compliance are to be compared
to the existing cost of production. Agency guidance regarding the
RFA defines a percentage increase in production cost of five
percent or more as a significant impact. A macro-level, or
aggregate, analysis produced an estimate of the percentage increase
in production costs to range from 0.05 to 0.15. Alternatively, a
micro-level analysis was conducted to examine the percentage
increase in production cost for affected systems within a system
size category. The micro—level analysis indicates the incremental
production cost increase to range from 63 to 206 percent, with an
average increase of 146 percent in systems serving less than 50,000
persons.
A detailed discussion of the number of water systems affected
by monitoring and paperwork requirements associated with the
1—6
-------
proposed rules is provided in the Information Collection Request
Document.
1.7 Summary of Costs and Benefits
The “best estimates” of total national incremental impacts of
the proposed ICC regulations are summarized in Exhibit 1-2. As
described above the standard error for incremental treatment and
waste disposal costs at the proposed MCL alternatives is between
20 and 28 percent of the values presented, depending on the
assumption used regarding the correlation of errors across
contaminants.
1—7
-------
EXHIBIT 1-2
Incremental Impacts of Proposed IOC Regulations
At Proposed MCL Alternatives
No. Systems Affected : 192
Compliance Costs :
Treatment/Waste Disposal — Total Capital 73
(Million 1986 5)
Treatment/Waste Disposal - 0 & M 6
(Million 1986 $/Year)
Total Annualized @ 3% Over 20 Years 11
(Million 1986 S/Year)
Monitoring (Million 1986 S/Year) 2
State Programs (Million 1986 5/Year) 01
Total Annual Costs (Million 1986 $/Year) 13
(Million 1986 S/Year)
Average Increased Household Costs (1986 5/Year) 2
System Size (Pop. Served) Best Estimate Range
25—500 461 286—571
501—3300 126 97—151
3301—50,000 72 52—78
> 50,000 0 N/A
Benefits :
Reduction in Population Exposed 170,000 Persons
1 lncluded with estimates for proposed Synthetic Organic
Chemicals regulations.
2 Average increased household costs are only for systems
affected by the rule.
1—8
-------
2. PROBLEM DEFINITION
2.1 Introduction
Potential threats to public health due to contamination of
drinking water supplies with inorganic chemicals (IOCs) have long
been recognized. Along with microbiological contaminants, lOCs
were among the first substances to be regulated under the Public
Health Service Act of 1962 and the original Safe Drinking Water
Act (SDWA) of 1974. The regulatory initiatives evaluated in this
document involve the revision of existing standards and estab-
lishinent o new standards for eight IOCs determined to be of sig-
nificance.
Exhibit 2—1 summarizes the available health effects inform-
ation on the lOCs addressed in this document. Also included in
Exhibit 2-1 is a summary of the proposed Maximum Contaminant Level
Goals (MCLGs) for each contaminant. Under the SDWA, EPA must set
MCLGs “...at the level at which no known or anticipated adverse
effects on the hea th of persons occur and which allows an adequate
margin of safety.” MCLGS are unenforceable goals which serve as
the basis for setting the enforceable Maximum Contaminant Levels
(MCL5).
For substances that have exhibited strong evidence of car-
cinogenicity the Agency’s policy is that MCLGs are set at zero.
For substances exhibiting “equivocal” evidence of carcinogenicity,
MCLGs are set based on sub-lethal health effects plus an additional
uncertainty factor. For substances with inadequate or no evidence
of carcinogenicity, MCLGs are set based solely upon sub-lethal
health effects. Complicating the task of determining “safe” levels
of exposure to lOCs in drinking water is the fact that some of
these substances have been shown to be essential for proper human
nutrition as trace elements.
The Agency has developed long-term drinking water equivalent
levels (DWELs) which serve as the basis for determining thq MCLGs
for lOCs via the procedure described below.
For non-carcinogens, threshold levels at which no adverse,
sub-lethal health effects are anticipated to occur for chronic
periods of exposure are determined from human or animal studies.
This threshold is referred to as either the no observed adverse
effect level (NOAEL) or lowest observed adverse effect level
1 1n addition to the lOCs evaluated in this document, revised
standards will be proposed for arsenic, lead and copper. Impacts
of these regulation will be evaluated in separate documents.
2 SDWA Section 1412 (b) (4)
2—1
-------
EXHIBIT 2—1
SUNMAFY OF HEALTH EFFECTS FOR I NOFGANIC CHE I IICAIS
• I I
I I I
I I I
• I I
Peripheral yes- : Selenoals • gas-
cular collapse. : trvintestinal
SE Ifl IILRI lesions of heart.: distress, liver I
kidney and spleen and spines
Pu1ncna edema effects
CHILD ADULT
I I
1—DAY H A (UG/L) 1: io-n&y H A (W/L) EWEL PROPCGED ESSENTIAL NITFRIBIT CARC1NCX3E I’IICI lY
(W/L) MCER :
: /L V/ l i , RDI2(w/DAY): CLASS
:7mill ion: : : :7mil lion
N/A : N/A : H/A : N/A : N/A :fibez-s/1: N : ti/A : A 6 : fibers/i
: 1ouin: : : : (>lOum)
I I I ‘ I
I I — I — —
I I I I I • •
I I I I • •
ti/A : N/A: N/A I N/A :5.000:5.000: ti: N/A : D : N/A
I I I I I I I I I
I I I I I I I I I
I I I I I I I I
I I I I I I I I I
I I I I
I I I I
I I I I I I I I
ti/A : N/A 1,400:5 ,000 1 200 100 : Y : 50—200 : A 5 : N/A
(CrVI) : (CrVI) : (CrVI) : 1(CrUflI 1 (CrVIfl
I I I I I
I I I I I I I I
I I I I I I
N/A fl/A N/A ,ID N/A
10,000 : iii,ooo 1 10,000 : xo,ooo 6 : :
(1000) (11,000) : (1000) 1 (100t H : N I N/A D N/A
I I I I I
I I I I I I I
I I I I I I I
1 I $ • I
I I • I
I I I
I I I • I •
I I I I U I
I I I I I I I
I I I I I I I
41 1 144 : 106 5 0 : Y : 50—200 : D : N/A
I I I I I I
I I I I I I I
I I I I I I
I I I I I I
1 H A - Health Advisory
2 RDI - Recongnended daily intake
3 N/A - information not available -
4 Via inhalation exjcsure Regulated for drinking water pirposea as if it were in Group C
5 Regulated for drinking water gurposes as if it were in Group D
6 In addition, PICLG for total nitrate and nitrite z 10 mg/i (as N)
INOEGANIC HEALTH EFFECFS
OVTA MNA I IF
GifiONIC
I Possible cancer
ASI3ESTC$ N/A I of the stomach
and pancreas
Gastrointestinal I Hypertension,
BARIUH : distress, hyper- Cardiovascular
tension, neuro- effects
suscular effects
Gastrointestinal : Kidney and liver
CAEt IILI 1 I distress, liver : damage, anemia,
& kidney failure,: hypertension,
renal dysfunction: &ne damage
CHRct lllfl l
Liver and kidney
damage, intermal
bleeding, nausea,
vomiting
Respiratory dis-
orders, derma-
titis, ulceration
of the skin
Gastrointestinal : Kidney damage,
MERCURY Distress, kidney I Central nervous : N/A : N/A
failure system effects
I I I
I__ — — ,
Methescglobin-
NITRATE : enia, Neurceuscu-: Central nervous
(NITRITE) lar effects, kid- system effects : N/A : N/A
ney damage, car—
diotoxicity
N/A
N/A
-------
(LOAEL). Depending on the nature of the toxicological data avai-
lable for a given substance, an uncertainty factor between one and
1,000 j s applied to the NOAEL or I OAEL to compute a Reference Dose
(RfD).’ The DWEL 4 is then calculated by converting the RfD into
a drinking water concentration. The DWEL calculation assumes that
total exposure is through drinking water for an average person of
70 kg body weight consuming two liters of water per day. The DWELs
for the lOCs are presented in Exhibit 2-1.
Finally, to determine the MCLG, the relative contribution to
total exposure from other sources is taken into account. Because
lOCs have been comparatively well—studied, sufficient data are
generally available on the contribution from food and air. For
these contaminants, the MCLG was generally calculated by account-
ing for the contribution from food and the contribution from air
reflected in the DWEL. The above calculations can be summarized
as follows:
NOAEL (or LOAEL)
RfD (in mg/kg/day) Uncertainty Factor(s)
RfD x Body Weight (70 kg for adults )
DWEL (in mg/i) = Drinking Water Consumption (2 1/day)
Percent of total exposure from
MCLG (in mg/i) = DWEL * Drinking Water
In cases where relative source contribution data are suff i-
cient, the DWEL is multiplied by the actual percentage of total
exposure accounted for by drinking water, if actual drinking water
exposure is determined to be between 20 and 80 percent of the
total. If the data show that actual exposure is lest than 20
percent from drinking water, 20 percent will be used as a minimum.
In cases where drinking water accounts for more than 80 percent of
the total exposure, 80 percent will be used as a maximum. Where
data are insufficient, contribution from drinking water to total
exposure is assumed to be approximately 20 percent. More detailed
descriptions of the derivation of the one and 10-Day Health
Advisories, RfD, DWELs, and MCLGs can be found in the Health
Criteria Documents prepared for each contaminant, and in the draft
Federal Register proposal.
The remainder of this chapter outlines additional background
information on bc contamination in drinking water. The following
3 me reference dose was formerly referred to as the Acceptable
Daily Intake or ADI.
4 me DWEL was formerly referred to as the adjusted acceptable
daily intake level or kADI.
2—3
-------
sections summarize four major topics: health effects; occurrence;
treatment technologies; and analytical detection methods. Further
summary information on each contaminant is presented in the Fact
Sheets in Appendix I. Readers are directed to the EPA-ODW
publications cited for each of these topics for more thorough and
detailed discussions.
2.2 Health Effects 5
Available evidence suggests that adverse health effects
attributable to bc exposure involve acute or chronic sub-lethal
endpoints of toxicity. As discussed above, the threshold concen-
trations at which no sub-lethal health effects are anticipated to
occur generally form the basis for MCLGs. However, three contami-
nants evaluated under this proposal have also exhibited evidence
of carcinogenicity; these are cadmium, chromium, and asbestos. The
strength of evidence of carcinogenicity via drinking water exposure
for these substances is described below.
While cadmium and cadmium compounds have been shown to induce
tumors at several sites through injection and inhalation exposure
in animal studies, no evidence has been found linking cadmium
ingestion with carcinogenicity in animals or humans. EPA has
classified cadmium in Group Bi according to the Proposed Guidelines
for Carcinogen Risk Assessment. However, since there are inade-
quate data to characterize the presence or lack of carcinogenic
hazard through ingestion exposure, cadmium will be regulated as
though it was classified in Group D.
Chromium (CrVI) compounds were shown to be carcinogenic in
animals via injection exposure. In addition, epidemiological
studies have demonstrated a link between prolonged inhalation of
Chromium VI compounds and the incidence of lung and respiratory
tract cancers. Based on this evidence, chromium is classified in
Group A, according to EPA’s Proposed Guidelines for Carcinogen Risk
Assessment. However, since there are inadequate data to charac-
terize the presence or lack of a carcinogenic hazard from ingestion
exposure, chromium will be regulated as though it was classified
in Group D.
Although inhalation exposure to asbestos has been shown to
produce a high incidence of lung tumors and inesothelioma in lab-
oratory animals, the majority of studies involving ingestion
exposure have failed to produce carcinogenic effects. In 1984
however, the National Toxicology Program (NTP) reported that “some
evidence of carcinogenicity” was observed in one sex of one species
5 Sources of information contained in this section are gener-
ally, 50 Federal Register 46957—46975, November 13, 1985, and the
Health Effects chapters of the Criteria Documents for lOCs.
2—4
-------
exposed to intermediate range (i.e., greater than 10 urn in length)
chrysotile asbestos fibers via ingestion. 6 In this study, a
significant increase was observed in the number of epithelial neo-
plasms in the digestive tracts of male F344/N rats. In addition,
there have been several epidemiological studies which have shown
gastrointestinal cancer to be associated with occupational exposure
to asbestos. The question of whether this observed increased risk
of gastrointestinal cancer is due to ingestion of inhaled fibers
has not been fully resolved, however.
Asbestos has been classified as a Group A carcinogen accord-
ing to EPA’S Proposed Guidelines for Carcinogen Assessment, based
on evidence of human carcinogenicity via inhalation exposure.
Because of the limited evidence of carcinogenicity via ingestion,
EPA is proposing to regulate asbestos as if it were in Group C for
drinking water purposes.
Based on the evidence in the NTP report described above and
the use of the “one-hit” model, EPA has calculated estimates for
ingestion of asbestos via drinking water. The concentration w ich
corresponds to a lifetime individual excess cancer risk of l0 is
approximately 7 million intermediate range fibers per liter (MFL).
This level is also being proposed by the Agency as the MCLG.
2.3 Occurrence
For each contaminant, the basis for all calculations of
impacts is the estimated number of public water systems (and in
turn, the number of persons) affected above a given concentration.
Occurrence estimates for all lOCs except asbestos were based on one
or more of five national surveys. For groundwater systems, the
National Inorganic and Radionuclides Survey (NIRS) provided the
primary source of occurrence data for all contaminants except
nitrate. For surface water systems, no single comprehensive
national survey has been conducted for lOCs. Consequently, the
results of four surveys were pooled to provide adequate baseline
data. These surveys are the Community Water Supply Study (1969),
the Community Water Supply Survey (1978), the Rural Water Survey
(RWS), and the National Organics Monitoring Survey (NOMS).
The NIRS sampling protocol ensured that samples represented
water in the distribution systems and specified flushing for at
least three minutes to avoid the influence of corrosion products,
including cadmium.
Review and analysis of the occurrence data was performed for
EPA Office of Drinking Water by Science Applications International
Corporation (SAIC). Based on that review, SAIC developed statisti-
650 Federal Register No. 219, pg. 46962.
2—5
-------
cal distributions of the occurrence probabilities at various
concentrations for each contaminant. These probablities were
formulated by system size category and water source. A “delta”
log—normal distribution used to represent the national occurrence
distribution and forms the basis for subsequent calculations. This
methodology assumes that all positive sampling results are log-
normally distributed and was considered most appropriate since it
corresponds well with the actual survey data. Under this inethodol-
ogy, nothing is assumed about the distribution of concentrations
in results below the minimum reporting levels. The delta log-
normal model is illustrated in Exhibit 2-2 below.
At the time the occurrence analysis was conducted, all NIRS
data were not available. Estimates for groundwater systems are
therefore based on approximately 85 percent of the NIRS data.
Recent comparison of occurrence probabilities derived through
analysis of 100 percent of the data with those originally used in
the draft RIA show a maximum variation of 1.5 percent for all
contaminants. On this basis, it was assumed that the use of
updated occurrence data would result in insignificant changes and
would be well within the bounds of uncertainty. For this reason,
not because of the schedule constraints involved, impacts were not
recalculated using the updated set of occurrence data.
In arriving at estimates of systems affected/population
exposed, the following methodology was employed. A model (referred
to as “Replicate”) was developed to replicate the delta log-normal
distribution results on Lotus spreadsheets. This model utilizes
statistical parameters calculated for each contaminant by source
and system size category, and allows estimation of occurrence
probabilities at MCL alternatives not explicitly considered in the
draft occurrence documents. The model also facilitates estimation
of the average concentration levels in affected systems. The
results of the Replicate calculations for lOCs are presented in
Appendix II. It should b noted however, that estimates at
concentrations below the MQC for a contaminant are less reliable
and subject to a greater degree of uncertainty. The most stringent
MCL alternatives analyzed for barium, cadmium, mercury, and nitrate
are subject to this qualification.
7 Science Applications International Corp., 1987. Preliminary
National Drinking Water Occurrence and Exposure Estimates for Phase
II Inorganjcs .
8 The MQC is the minimum level at which quantitative analyti-
cal results are considered reliable for a given set of survey data.
This is distinct concept from the practical quantification limit
(PQL), and is explained in further detail in the draft occurrence
documents for each bC. The MQCs and PQLs are not necessarily
equivalent for these contaminants.
2—6
-------
EXHIBIT
General Form of Delta Log—Normal
.6
.5
.4
.3
.2
.1
0
Di tribution
Minimum Reporting Level MRL)
Co N CE NT RATIO N
— Represents proportion of total 4amples < MRL
for which it is not possible to describe a
itistributi n (i.e.. this represents the delta
portion of the overall listribution
PROBABILITY
of
CC [ l:I I:N E
2
4 5
2—7
-------
To obtain estimates of actual number of systems affected and
population exposed, results from the Replicate model are merged
with the data on total numbers of systems and population served by
source and system size presented in Exhibit 2-3. The standardized
data in Exhibit 2-3 are based on a FRDS run conducted in March,
1986 and have been corroborated by data collected in Conjunction
with a recent survey by the Association of State Drinking Water
Administrators. It is recognized however, that these estimates
may be subject to change due to inconsistencies in reporting
procedures, and the fact that the number and status of public water
systems (PWSs) nationwide are in a constant state of flux.
To maintain consistency between the disparate data sets used
in the calculation of impacts, it was necessary to rectify a
discrepancy between the form in which the occurrence probabilities
were originally displayed, and that required as an input to the ATm
cost model. Although both data sets were arrayed in 12 system size
categories by population served, the size categories did not match.
Since the cost model size categories were matched to those used in
the Cost and Technology Documents, it was considered most appro-
priate to convert the occurrence data accordingly. As displayed
in Exhibit 2-4 below, it was assumed that the same probabilities
calculated using the original draft SAIC occurrence information
would apply to system sizes that overlap.
9 Association of State Drinking Water Administrators, 1986.
Survey to Support Analysis of Impacts of Proposed Regulations
Concerning Filtration and Disinfection of Public Drinking Water
Supplies . September 16 draft.
2—8
-------
EXHIBIT 2—3
Standardized Nt.iiters of Public Water Systems and Populations Served by Source and Size Categories
I SizeCategory 1 2 3 4 5 6 7 8 9 10
I Pop. Served 25-100 101-500 501 1K 1K-3.3K 3.3K1OK 1OK-25K 25K-SOK 50K-75K 75K-lOOK 100K-SOOK 500K-is Over 114 Total
Conmunity Water Systems SUFACE WATER SYSTEMS I
Purchased Systems 620 1,343 741 1,197 613 308 166 63 29 38 3 2 5,123
Non Purchased Systems 833 779 754 1,040 1,156 569 328 157 108 175 43 15 5,957
Total Systems 1,453 2,122 1,495 2,237 1,769 877 494 220 137 213 46 17 11,080
Total Populations (000s) 90 570 1,280 4,330 10,200 12,640 15,910 10,310 10,090 36,770 22,380 30,090 154,660 I
Non - Cosninity Water Systems I
PurchasedSystems 6-44 177 49 30 13 0 3 0 0 0 0 0 9161
Non Purchased Systems 2,110 1,069 192 135 50 2 0 0 1 1 0 0 3,560 I
Total Systems 2,754 1,246 241 165 63 2 3 0 1 1 0 0 4,476 I
Total Populations (000s) 134 310 198 301 375 50 88 0 82 360 0 0 1,898 I
Non Transient Non Coom. Systems I
PurchasedSystems 91 25 7 4 2 0 3 0 0 0 0 0 1321
NonPurchasedSystems 298 151 27 19 7 0 0 0 1 1 0 0 5041
TotalSystems 389 176 34 23 9 0 3 0 1 1 0 0 636 1
Total Populations (000s) 19 44 28 42 53 7 88 0 82 360 0 0 723
Conmiunity Water Systems GROUND WAlER SYSTEMS I
Purchased Systems 221 540 244 268 92 37 13 3 0 3 0 0 1,421 p
Non Purchased Systems 17,079 15,354 5,038 5,185 2,308 823 278 77 17 39 4 0 46,202 I
Total Systems 17,300 15,894 5,282 5,453 2,400 860 291 80 17 42 4 0 47,623 I
Total Populations (000s) 950 3,850 3,910 10,000 13,310 13,110 9,540 4,710 1,360 7,600 2,760 0 71,160
Non - Coniunity Water Systems
PurchasedSystems 385 60 9 5 2 2 0 0 0 0 0 0 4631
Non Purchased Systems 98,453 28,078 8,155 1,455 349 25 1 1 6 2 7 - 0 0 136,541 I
Total Systems 98,838 28,138 8,164 1,460 351 27 11 6 2 7 0 0 137,004
Total Populations (000s) 4,355 6,637 7,193 2,485 1,934 503 275 398 180 1,096 0 0 25,056
Non Transient Non Corn. Systems I
PurchasedSystems 54 8 1 1 0 0 0 0 0 0 0 0 65 1
Non Purchased Systems 13,899 3,964 1,151 205 49 4 11 6 2 7 0 0 19,298
Total Systems 13,953 3,972 1,153 206 50 4 11 6 2 7 0 0 19,364 I
Total Populations (000s) 615 937 1 .015 351 273 71 275 398 180 1,096 0 0 5,211 I
Total populations include those served by purchased and non-purchased water systems.
Non purchased water systems conprise of systems treating their own source water supplies and exclude systems purchasing water from other sources.
Totals may not tally due to independent rounding
11 12
oI
-------
EXHIBIT 2—4
Conversion of Occurrence Data to Standardized
System Size Categories (by Population Served)
System Size
1
2
3
4
5
6
7
8
9
10
11
12
Cateaorv
Draft SAIC
Occurrence Document
25—100
101—500
501—1000
1001—2500
2 5 01—3 3 00
3301—5000
5001—10,000
10, 001—25, 000
25,001—50,000
50, 001—75, 000
75,001—100,000
100, 000+
Standardized
f or Draft RIA
2 5—100
101—500
501—1000
1001—33 00
3301—10,000
10, 001—25,000
25,001—50,000
50, 001—75, 000
75, 001—100, 000
100,001—500,000
500, 000—1, 000, 000
1,000,000+
The data 1 ° summarized in Exhibit 2-5 illustrate another
potential source of error in calculating impacts for the proposed
IOC rule. Comparison of the FRDS violation data with estimates of
number of community systems affected at the interim MCLs calculated
using the “Replicate” methodology described above i11ust tes that
some error is inherent in one or both of the data sets, 1 - despite
their agreement in the total number of systems affected for several
contaminants.
10 FRDS violation data were compiled in the draft occurrence
documents for each contaminant.
11 1t should be noted that the “number of systems affected” in
Exhibit 2-4 does not correspond precisely to the “number of systems
incurring treatment costs” (presented in Chapter 4 and in Appendix
III) since both purchased and non-purchased water systems are
included in the former calculation, and only non-purchased systems
in the latter.
2—10
-------
EXHIBIT 2-5
Community Water Systems
Contamination Greater
or Equal to Current
with bC
Than
MCLs
IOC
(current MCL
in ug/l)
FRDS Violations 12
RIA Estimate of Tot
Systems Affected
Ground Surface
Water Water Total
Ground Surface
Water Water
Total
As noted above, it was necessary to combine several data sets
to derive occurrence estimates for surface water systems. The
quality and quantity of data in these surveys (especially the RWS)
varied significantly. As a result, the accuracy of these estimates
may be questionable. Although the majority of occurrence seems to
be in groundwater systems, significant surface water occurrence is
estimated for some contaminants. For example, mercury is estimated
to have an average occurrence probability of 6.62 percent in
surface water systems at a concentration greater than the proposed
MCLG of 2 ugh. Intuitively, this seems extremely high and may not
be an accurate representation of field conditions. This is
supported by the analysis of FRDS violations data which indicate
12 Source: SAIC draft occurrence documents, data current as
13 Includes both purchased and non-purchased systems.
Barium (1000)
73
1
74
Cadmium (10)
31
6
37
Chromium (50)
25
3
28
Mercury (2)
38
19
57
Nitrate (10,000)
726
75
801
Selenium (10)
Total
208
17
225
848
1191
121
1222
Total less
Mercury SW
1191
102
1203
848
14
0
48
0
6
26
34
733
767
669
22
691
929
196
1777
1044
of FY—1986.
2—il
-------
that there were only 19 violations of the r teriin mercury MCL of
2.0 ugh in surface water systems in 1986 (see Exhibit 2-5)
Exhibit 2-5 also illustrates inconsistencies for contaminants such
as selenium where most occurrence estimated using the Replicate
model is in surface systems whereas the FRDS violation data suggest
the opposite.
In addition to the uncertainty associated with the raw data
used to estimate national occurrence, at least two other potential
sources of error are inherent in the occurrence estimates.
First, because of the methodology used to consolidate survey
data, some small probability of occurrence is assigned to all
system size categories, including the largest systems. Thus, while
the model may project only a “fraction” of a system in the larger
size categories to be affected, this may still result in an
estimate of significant impacts due to the high unit costs and
large populations served in these systems. Since more extensive
and sophisticated monitoring and treatment practices are generally
in place at systems serving greater than 50,000 persons, the
likelihood of bC occurrence at levels of concern relative to that
in smaller systems is probably negligible. Based on this ration-
ale, it is assumed that occurrence of lOCs is negligible in the
larger system size categories ‘for purposes of calculating incremen-
tal impacts (see Chapters 4 and 5).
Second, the phenomenon of co-occurrence was evaluated based
the assumption of joint, independent probabilities of occurrence
for individual contaminants. It was estimated that approximately
one percent of affected systems have occurrence of two contaminants
(see Exhibit 2-6). On this basis, it was assumed that a much
smaller number of systems would have simultaneous occurrence of
three or more contaminants. Based on these r-esults, it was assumed
that the degree of co—occurrence of lOCs (and resulting effects on
the aggregate national impacts) is negligible and no analyses
beyond those presented in Exhibit 2-6 were conducted.
Comprehensive national surveys on occurrence of asbestos in
drinking water supplies have not been undertaken. However, it is
known that asbestos fibers may contaminate water sources through
natural erosion of mineral deposits and through anthropogenic
sources such as industrial discharges and runoff from mine tail-
ings. In addition, corrosion and tapping of asbestos/cement pipes
may contaminate water in distribution systems, particularly in
14 Science Applications International Corporation, 1987.
Estimated National Occurrence and Exposure to Mercury in Public
Drinkin Water Supplies (Revised Draft) . Section 2.2.
2—12
-------
DRAFT
EXHIBIT 2—6
Summary of Co—Occurrence in All Community Water Systems
JUN 301991
- C ’ VT
0.03Z 0.04Z
0.4 5t 0.63X
Contam i nant
AFFECTED
AFFECTED
AF FEC TED
TOTAL
POPULATION
TOTAL
POPULATION
POPULATION
SURFACE
SURFACE
GROUND
GROUND
ALL
ALL
Proposed
Co-cant-
Preposed
WATER
WATER
WATER
WATER
WATER
WATER
NCL
am:nant
MCL
SYSTEMS
SYSTEMS
SYSTEMS
SYSTEMS
SYSTEMS
SYSTEMS
Arsenic 50
Barium
1500
0
0
0
0
0
0
Arsenic 50
Cadmium
5
0
0
0
43
0
43
Arsenic 50
Chromium
120
0
20
0
0
0
2 f l
Arsenic 50
Capper
[ 300
0
22
0
7
0
29
Arsenic 50
lead
20
0
42
143
186
Arsenic 50
Mercury
3
1
17,933
0
1
1
1798 ’
Arsenic 50
Nitrate
10
0
1,482
0
124
0
1606
Arsenic 50
Selenium
45
0
0
0
0
0
Barium 1500
Cadmium
5
0
0
0
4
0
4
Barium 1500
Chromium
120
0
0
0
0
0
?arium 15fl0
Copper
13 f l0
0
0
0
1
:
Eariu:i 1500
Lead
20
0
0
12
‘
C
ar:um 1500
Mercury
3
I l
i i
u
i l
Barin 1500
Nitrate
kO
0
0
0
i
.1
?ariuu 130O
Selenium
15
0
0
‘ 1
“
Cadmium 5
Chromium
120
0
0
0
1
0
1
Cadmium 5
Copper
l3 l 0
9
0
0
414
0
l4
Cadm ium 5
Lead
20
0
0
5
7,997
5
1.997
Cadmium 5
Mercury
3
0
0
41
0
47
Cadmium 5
Nitrate
10
0
0
4
6911
4
E91 1
Cadmium 5
Selenium
45
0
0
ii
0
0
,
Chromium 120
Copper
1300
0
1
0
0
0
1
Chromium 120
Lead
20
Q
I
A
4
5
Chromium 120
Mercury
3
0
411
0
0
0
‘71
Chromium 120
Nitrate
10
0
39
1
3
0
42
Chromium 120
Selenium
45
fl
0
0
0
0
0
Copper 1300
Lead
20
0
1
1
1,372
1
LP’
Copper 1300
Mercury
3
fl
506
0
9
0
511
Copper 1300
Nitrate
10
0
42
1
1 ,196
I
1,223
Copper k300
Seleniun
45
0
0
0
0
‘
‘
Lead 20
Mercury
3
1
9 93
:56
, 0
Lead 20
Nitrate
10
0
81
13
22,921
13
:, ) 2
Lead 20
Selenium
£5
0
0
0
0
0
0
Mercur 3
Nitrate
10
1
34,462
t)
ITS
1
34.5
Mercury 3
Selenium
45
1
0
A
i l
0
Nitrate 10
Selenium
45
U
)
0
‘)
0
Total Co—occurrence
Percent of all systems
Percent of Total Occurrence
25
41,501
0. 06Z
39
.i5 t
tl•
4X
1.29t
i.37t
1.’)7t
0 .927.
-------
areas with highly aggressive natural waters. 15 Depending on the
location and other specific characteristics of the water system,
the size, type, and concentration of asbestos fibers may be highly
variable.
Based on a compilation of several studies of limited geo-
graphic scope, EPA estimates that as many as 3,234 public water
systems nationwide may have occurrence of asbestos fibers greater
than 10 urn in length in concentrations exceeding 0.071 MFL. of
these systems, t least 609 may have levels which exceed the
proposed MCLG. 1 This occurrence is generally restricted to
systems with some asbestos/cement pipe in their distribution
systems, or in surface water systems in regions of the country
having significant asbestos-containing mineral deposits (i.e.,
California and Washington state). Twenty-six (4.3%) of the systems
estimated to have occurrence in excess of the proposed MCLG are
surface water systems without asbestos/cement pipe.
2.4 Treatment Technologies
The most appropriate centralized treatment technologies for
removal of lOCs from drinking water are varied and generally depend
on a number of physical and chemical parameters of the contaminants
and source waters. In general, metals which form insoluble
precipitates lend themselves to removal by conventional filtration
processes or lime softening. Metals which complex in aqueous
solutions will require more technologically intensive processes.
Non-metals (e.g., nitrate and nitrite) generally do not form salts
with low enough solubilities for removal by conventional processes
and therefore may also require more advanced types of treatments
such as ion exchange. However, because it is readily oxidized to
nitrate, any instances of nitrite contamination of source water
would generally be rectified through disinfection of the water
supply prior to distribution.
Applicable technologies for IOC removal are summarized below.
Detailed descriptions of bC treatment technologies are provided
l5 EPA, Office of Drinking Water, Criteria and Standards
Division, 1987. Drinking Water Criteria Document for Asbestos
( revised Draft) . PP. IV-2 — IV—3.
16 Ibid., pg. IV-12. It should be noted that at the time the
asbestos occurrence analysis was conducted, an MCLG and proposed
MCL of 7.1 MFL was assumed and therefore may represent a slight
underestimate of the number of systems with occurrence above the
current proposed MCL of 7 MFL.
2—14
-------
in the appropriate draft “Cost ar Technology Documents” published
by the Office of Drinking Water.
Conventional Treatment Processes
This process is generally used for removal of color or tur-
bidity from surface waters. The water treatment process usually
involves the following steps: mixing with coagulants, floccula-
tion, sedimentation, filtration, and disinfection. Removal of
inorganics occurs through adsorption to or enmeshment in the floc.
This process is generally effective for removal of cationjc
inorganics (see Exhibit 2-7). For nitrate and barium however, the
process is virtually ineffective.
Lime Softening
This process has traditionally been used to remove hardness
from ground and surface waters. The process consists of the fol-
lowing steps: softening with lime or soda ash, sedimentation,
filtration, and disinfection. Like conventional treatment, inor-
ganics removal occurs through adsorption on or enmeshment in the
floc. In general the process is effective on cations but not on
chromium VI, mercury, selenium, or nitrate (see Exhibit 2-8).
EXHIBIT 2—7
lOCs for Which Conventional Treatment is Effective
(Greater than 70 Percent Removal)
With Alum Coaqulatjon With Iron Coaqu1atior
Asbestos Asbestos
Cadmium (at pH > 8.5) Cadmium (at pH > 8.0)
Chromium III Chromium III
Chromium VI (with ferrous salts)
Mercury
Selenium IV
‘ 7 u.s. Environmental Protection Agency - Office of Drinking
Water, Criteria and Standards Division. 1986. Technologies and
Costs for Removal of Inorganic Chemical Contaminants from Public
Drinking Water Supplies . Draft documents prepared for each
contaminant.
2—15
-------
EXHIBIT 2—8
lOCs
for
Which Lime Softening is Effective
(Greater than 90 Percent Removal)
Treatment
Barium (at pH 9.5 - 10.8)
Cadmium
Chromium III (at pH > 10.5)
Ion Exchange
This process is used to remove hardness arid nitrates from
groundwaters. Removal of inorganics occurs through adsorption to
exchange sites on the resin. Typical unit processes include
prefiltration (the process is effective only with extremely low
turbidity waters), ion exchange, and disinfection. With the use
of cationic resins, the process is effective on cationic morgan-
ics, while the use of anionic resins is effective for nitrate and
selenium (see Exhibit 2-9).
EXHIBIT 2-9
lOCs for Which Ion Exchange Treatment is Effective
(Greater than 90 Percent Removal)
Anion Exchange Cation Exchange
Chromium VI Barium
Nitrate Cadmium
Chromium
Reverse Osmosis
This process has traditionally been used in the desalination
of seawater and brackish groundwaters. It involves forcing water
through a filter membrane under pressure. Inorganics are removed
by retention in the brine by the membrane. The process generally
involves the following steps: pretreatment, membrane contact, and
disinfection. This process is effective in treating for all lOCs
at greater than 70 percent removal.
2—16
-------
Activated Alumina
This process is used to remove inorganics from groundwaters
through adsorption on the activated alumina. The process consists
of the following steps: pretreatment, activated alumina contact,
and disinfection. The process is effective in removing selenium
IV and VI, but it is not effective in removing most cations.
Activated Carbon
Activated carbon (granular or powdered) is generally used for
the removal of tastes, odors, and organic chemicals from surface
waters. Inorganic chemical removal takes place through adsorption
to the activated carbon. Typically, activated carbon use occurs
as a unit process of conventional or lime softening treatment.
Bone char activated carbon is effective in removing arsenic while
granular or powdered activated carbon is effective on organic
mercury.
Other Alternatives
Several alternatives to centralized treatment, including
regionalization, welifield management, and alternate source, are
available. For very small systems, exemptions to the requirements
based on economic considerations may be available. Specific
alternatives to centralized treatment are discussed below.
Regionalization involves interconnection of one water system
to an existing, adjacent system with excess capacity and acceptable
quality. The affected community then usually purchases sufficient
water to meet its needs. This technique i most frequently used
by small water systems.
Welifield management consists of pumping to waste from contam-
inated wells, and increasing production from remaining wells. This
technique is feasible only for systems that have enough remaining
capacity to supply needs without drilling new wells (e.g., large
groundwater systems).
Alternate source involves developing a new water source and
installing pipeline to replace a contaminated ground or surface
water system. In general, this alternative is feasible only for
small systems which have the option of drilling a new well and
utilizing a groundwater source for which no treatment beyond that
already in place is required.
2—17
-------
2.5 Analytical Methods. 18
The Agency has proposed that analytical methods are available
for all lOCs for which MCLGs have been proposed. A number of
generally available detection methods exist such as furnace atomic
absorption (AA), flame atomic absorption, and inductively coupled
plasma (ICP) atomic emission spectrometry techniques. Exhibit 2-
10 lists proposed analytical methodologies for the lOCs, and
practical quatitation limits (PQLs). Because impacts of the
monitoring requirements for arsenic, lead, and copper are not
addressed in this document, the analytical techniques for these
substances are not included in this summary. This information will
be included in separate ODW reports.
For most lOCs use an AA technique or ICP are the accepted
methods of detection. Two exceptions are mercury and nitrate.
Analytical methods for mercury in drinking water include the manual
and the automated cold vapor techniques. Nitrate detection is
accomplished through calorimetric brucine, spectrometric cadmium
reduction, and automated hydrazine and cadmium reductionteci -iniques.
More details regarding analytical methodologies are provided in the
Federal Register Volume 50, November 1985.
18 Federal Register 46957—46959. November 13, 1985.
2—18
-------
EXHIBIT 2—10
ANALYTICMJ METHODS FOR IOCS
Inorganic Contaminants Detection I Practical I EPA Method
and Methods : Limit (ugh) I Quantitation I Number
Level (ugh)
I U
I I I
I I I
Asbestos ‘ N/A 1 1
I I
Transmission Electron Microscopy 0.01 NFL I EPA
I I I
U U I
150
• Barium
AA Flame 1 100 1 206.1
AA Furnace ‘ 2 I 208.2
• I I
ICP • 2 ‘ S 200.7
I U I
I I
U I
‘Cadmium ‘ U 21
I U
AA Furnace 1 0.1 1 213.2
IC? 1 1 I 200.7
U I I I
U U I
Chromium ‘ ‘ 10 1
* I
AA Furnace 1 1 1 1 218.2
IC? 7 U 200.7
I I
U I I I
S * I
‘Mercury ‘ ‘ 05’
$ U . I U
Manual Cold Yapor Technique U 0 2 ‘ 245.1
• • I
02 1 245.2’
Automated Cold Vapor Technique
U U I I
S I U U
Nitrate 1 400 ‘
I U
• Ion Chromatography U 10 ‘ 300.0
I I $
Automated Hydrazine Reduction I 10 ‘ I 353.1
U I - I
Automated Cadmium Reduction 1 50 ‘ ‘ 353.2 1
U I
Manual Cadmium Reduction U 10 ‘ U 353.3
I I S
Ion Selective Electrode 1 1000 1
U U U I
• U I I
Nitrite 400
Spectrophotometric U 10 ‘ I 354.1
U I U
• Automated Cadmium Reduction 50 I 353.2
Manual Cadmium Reduction 10 ‘ 1 353 3
• I
Ion Chromatography • 4 ‘ U 300.0
I U
U I U I
U U I I
• Selenium I U 10 ‘
• U I U
AA; Gaseous Hydride I 2 ‘ ‘ 270.3
I I I
AA Furnace ‘ 2 I 270.2 1
I I I
icP 1 10 1 1 200.7
-------
3. MARKET IMPERFECTIONS, THE NEED FOR FEDERAL REGULATION, AND
CONSIDERATION OF REGULATORY ALTERNATIVES
3.1 Introduction
The “Public” Nature of Water Supply
Public water supply is an example of a “natural monopoly;” it
would not be efficient to have multiple suppliers competing to
build, operate, and maintain multiple systems of pipelines, res-
ervoirs, wells and other facilities. Instead, a single entity
performs these functions under public control.
W1iile not a pure “public good” in the economic sense, drinking
water is nonetheless a “publicly provided good” in that there is
a significant government role in the pricing and production
decisions of the industry. “Public” water supplies are typically
either publicly owned and operated as a routine function of local
government, or privately owned and publicly regulated as a routine
function of state government.
The concept of a “production function” is used to specify
combinations of inputs to a production process which can be
employed to produce alternative combinations of outputs. In
competitive markets, producers attempt to find the optimal
combinations of inputs and outputs to suit prevailing market
conditions. Thus, the physical options in production are fixed by
technical factors, but the choice between them is market driven.
A natural monopoly such as water supply is an example of
“market failure;” the competitive forces that would normally shape
production decisions “fail” to perform. Therefore, the production
function or publicly provided goods (i.e., the “social production
function” ) embodies not only physical relationships between input
and output combinations, but also a method for choosing among them.
There are two essential questions that must be answered in orde
to make optimal production decisions for publicly provided goods:
o What price should be charged to ensure optimal utilization
of any given level of service?
o What is the optimal level of service to provide?
1 Schultze, C.L., The Politics And Economics of Public
Spending , 1968, The Brookings Institution, Washington, D.C., ninth
printing, 1977, p.56.
2 Layard and Walters, Microecorioniic Theory , 1978, McGraw Hill,
New York, NY, p.171 and p.196.
3—1
-------
To maximize efficiency, prices would be set equal to marginal
cost and decisions regarding the level of ervice would be based
on comparisons of benefits and costs. In other words,
institutions responsible for publicly provided goods should strive
to mimic the processes of competitive markets.
In competitive markets, these two questions are answered
simultaneously and continuously through the presence of multiple
suppliers. This is a successful way of operating because, across
a range of price levels, consumer preferences for different levels
of service are revealed in the transactions taking place in the
market. There are two essential differences in the case of
publicly provided goods. First, differential preferences regard-
ing the level of service are not fully revealed in the face of a
single monopoly price. Second, depending on the institutional
arrangements in place, there may or may not be a direct relation-
ship between the price charged and the level of service provided.
3.2 The Nature of the Imperfections
Imperfect revelation of consumer preferences and flawed
pricing policy are the two fundamental imperfections affecting the
provision of public water supplies. The effects of these
imperfections are manifest in the history of “underpricing” that
characterizes the water industry. Water has been underpriced in
one sense because the relative abundance and purity of available
sources have long been taken for granted and not reflected in the
price signal or the demand response. In another sense, water has
been underpriced because the historical cost of water was so low
that pricing practices evolved in ways unrelated to the cost of
service in many places.
The effect of these two concepts of underpricing are evident
in the historical trend of water utility bills. The top panel of
Exhibit 3-1 presents a comparison of annual family utility bills
over the period from 1952 to l984. The bottom panel presents the
same comparison in terms of the percentage of median family income.
It is clear from these diagrams that rates for other utilities have
grown at a much faster pace than those for sewer and water. In
fact, it has been confirmed that the real price of water supply has
3 lbid, p. 174
4 BLS data.
3—2
-------
EXHIBIT 3—1
Percent of Median Family Income
Spent on Selected Utilities 1952—1984
Annual
for Selected
Electric
Phone
— —
Natural Gas
Water (and Sewer)
1980 1990
Family Bills
Utilities 1952—1984
$717
/ 608
1
500— ‘
I 497
, /
400— ,/ (
‘-I ,
00—. I,
‘ /
/
200H$18& . . /
I /
I - ___
1950 1960 1970
Year
6 _ i
5. 4Z __________________________
I * S
3 J
2.ZX
.65X
.54X
1950 1960 1970
Year
3—3
Electric
Phone
—
— -
Natural Gas
Water (and Sewer)
0
(J
—
8
C
a
V
0
V
U
I.
V
0
1980
1990
-------
actually declined over time. 5 ’ 6
Flaws In The Revelation of Consumer Preferences
Centrally supplied potable water is a “multi-attribute good”
which also has multiple uses. There are two major classes of
attributes: quantity features and quality features. These may also
be referred to as “pressure” and “purity.”
Keeping pressure in the pipes is achieved by maintaining ade-
quate flows and reliable performance throughout the water system -
— from the water source(s) through treatment, storage, and
distribution. In addition to the economic benefits of having a
central water supply for a multitude of residential, commercial,
and industrial uses, adequate capacity and pressure also serve a
public safety purpose in providing fire fighting capability.
Overlaid on these use-specific attributes is the more general
attribute of reliability. As in all categories of infrastructure,
there is an implied warranty that the system will not fail.
Reliability of water service is taken for granted to the extent
that the public reaches for the tap with the same confidence exhi-
bited in flipping a light switch.
The purity of the water delivered to water system customers
is assured by performance of the treatment facilities. The purity
attribute has four important elements: 1) aesthetic appeal (i.e.,
taste, odor, and appearance); 2) safety from acute health risks;
3) safety from chronic health risks; and, 4) public confidence that
the water is safe to drink. This last attribute constitutes
another implied warranty. Similar to other categories of
infrastructure that affect public safety, the safety of potable
water is largely taken for granted. For the most part, people fill
a coffee pot with the same nonchalance they exhibit in driving
across a bridge or stepping onto the elevator of an office
building.
To ensure provision of optimal levels of each of the multiple
attributes of public water supply, each must be given appropriate
weight in the production decisions by local water systems. Despite
the fact most consumers have taken all of these attributes for
granted for many years, the weights assigned in local decision-
making processes have not always been optimal. The performance of
5 Mann, P.C., and LeFrancois, P.R., “The Real Price of Urban
Water,” Journal of the American Water Works Association , January
1982, Vol. 74, No. 1.
6 Beattie, B.R. and Foster, H.S., “Can Prices Tame The
Inflationary Tiger?” Journal of the American Water Works Associ-
ation , August 1980, Vol 72, No. 8.
3—4
-------
water systems is commonly regarded as adequate as long as
uninterupted delivery of the most readily noticeable attributes
(e.g., pressure, aesthetic appeal, and protection from acute health
risks) is maintained. Thus, these most “visible” attributes are
accorded the greatest weight in decisionmaking. Problems involving
planning for long term needs (e.g., infrastructure maintenance and
replacement, chronic health risks) or low probability events (e.g.,
drought, waterborne disease outbreaks) tend to be underweighed in
the local “public choice” environment.
Flaws in Pricing and Capacity Planning
Provision of water supply has historically been regarded as
a “service delivery function.” Recently however, it has been
acknowledged that an era of relative scarcity may be beginning
which will force recognition of water as a “commodity.” It is
generally agreed that, although the United States is not actually
“running out of water,” it is “approaching the limits of inexpen-
sive water.” 7 Within this context, water supply is both a service
and a commodity; both characteristics are present in the quantity
and quality attributes of the good.
As alluded to above, a service orientation of “meeting
capacity requirements” has traditionally predominated in local
decisionmaking processes. Another facet of the service orientation
is the belief that water supply should be provided at an affordable
price as a publc service and, if necessary, supported by subsidy.
To adjust to conditions of relative scarcity, however, a commodity
orientation towards pricing and cap city planning must also be
incorporated in local decisioninaking.
In the future, there is likely to be a convergence of the
factors which have historically been underweighted in local
decisionmaking. Increased relative scarcity will make the untreat-
ed source water (the basic commodity) more expensive. Treatment
requirements imposed by the Safe Drinking Water Act Amendments of
1986 (SDWA) will increase the cost of producing “finished” water
at the treatment plant. Deteriorating infrastructure, exemplified
by leakage in the distribution system, will increase the cost of
“delivered” water either through continued leakage of increasingly
7 Frederick, K., “The Legacy of Cheap Water,” Resources ,
Resources for the Future, Inc., spring, 1986.
8 For example: Lamm, Richard D., “Kicking The Cheap Water
Habit: A New Era In Water Management,” in Water Values and Markets:
Emerging Management Tools , The Freshwater Foundation, Navarre,
Minnesota, 1986.
:3—5
-------
valuable treated water or through the cost of making overdue
repairs.
Underpricing always implies excess capacity. The excess
capacity in water supply may result in increased costs of
compliance with regulations for water systems. One result of
drinking water regulation, therefore, will be a reduction in the
extent of this over capacity. Numerous water systems have already
initiated more aggressive capacity management procedures in an
attempt to meet demand more efficiently while, minimizing system
loss. These measures are intended to reduce their anticipated
costs of compliance with drinking water regulations. The economic
efficiency of this adjustment cannot be precisely quantif led, but
the net effect should be enhanced attention to resource allocation
principles in the decisionmaking process. As a result, the overall
change in society’s welfare would be positive. This is relevant
to the evaluation of econmonic impacts of drinking water
regulations because costs of compliance are currently calculated
based on existing notions of capacity.
It is likely that significant increases in consumers’ monthly
water bills will result as the above noted convergence of factors
affecting capacity costs occurs. The water supply industry has
stressed the need to educate the public regarding the true value
of potable water in order to mitigate adverse reactions to such
increases. Given the market imperfections that exist, such public
education would seem to be an efficient intervention. Since the
historically low price levels of drinking water supplies may
distort the response of consumers to corrective changes in prices,
it is not clear that this response implies anything about the con-
sumer’s true willingness to pay. When price signals are distort-
ed, the demand response cannot be literally interpreted.
3.3 The Need for Federal Regulation
The “market failure” induced by conditions of natural monopoly
is not a failure of the mechanism of “private choice” to reveal
preferences through consumer responses to prices. The under-
weighting of certain attributes of public water supply in local
decisionmaking processes reflects a failure of the “social pro-
duction function” -- a failure of “public choice.” This failure
cannot be fully corrected at the state or local level and is
evident in both publicly owned and privately owned water systems
as discussed below.
Publicly Owned Water Systems
Many publicly owned water systems exist in institutional
settings in which water system revenues and costs are commingled
with other functions of local government. Where the operations of
3—6
-------
water systems are not fiscally discrete within multi-purpose local
governments, there is no means of ensuring optimal pricing and
production decisions for water supply. In the commingled setting,
outcomes will be optimal only in cases where the weights assigned
to the multiple attributes of public water supply are the same as
those that would be produced when the water supply “objective
function” is considered by itself.
Commingled budgeting precludes establishment of a rational
relationship between the reve ues generated by the water system
and its level of expenditure. When revenues are contributed to
a jurisdiction’s general fund, the water system is left to compete
for funding along with other public needs through a process
unrelated to the amount of revenue generated by water rates. The
separation of revenues from expenditures produces not only
arbitrary and suboptimal patterns of expenditure, but arbitrary
pricing polici as well. General fund financing creates an air
of uncertainty- which fosters a misperception that a relationship
exists be?Jeen the cost of the service and the level of service
provided. In this regard, water supply is generally regarded as
an excellent bargain because of the historical trend towards
underpricing.
Privately Owned Water Systems
In 45 states, privately owned water systems must obtain
approval for rate increases from state public utility commissions
(PUCs). Conventional principles of public utility regulation would
be expected to establish a full-cost basis in the relationship
between prices, costs, and revenues. Ideally, this would permit
privately owned water systems to apply appropriate weights to the
multiple attributes of water supply in their production decisions.
In practice, however, there are flaws in the rate regulation
process and, in particular, flaws in its application to water
utilities.
There are two major classes of privately owned water systems:
1) investor owned systems which have professional management; and
2) small systems without professional management belonging to
homeowner associations, trailer parks, and similar non-municipal
9 Goldstein, J., “Full-Cost Water Pricing,” Journal of the
American Water Works Association , February 1986.
10 Buchanan, J.M., Public Finance In A Democratic Process ,
Chapel Hill, University of North Carolina Press, 1967.
11 Mueller, D.C., Public Choice , Cambridge University Press,
1979., p. 90.
3—7
-------
entities. In general, investor-owned systems tend to be successful
in negotiating rate increases with the PUCs, while the smaller
private systems are poorly equipped to prepare or defend rate
proposals and, as a result, often do not even apply for rate
increases.
The problems inherent in PUC regulation of small water systems
have been the subject of a number of recent studies by the National
Regulatory Research Institute (NRRI) under sponsorship of the
National Association of Regulatory Utility Commissioners (NARUc).
One set of findings confirms that the PUC regulatory process has
presented a number of barriers in its application to small water
utilities. First, the total dollar value of water utilities under
commission jurisdiction is est ated to be less than one percent
of the total for all utilities. The procedures required of large
gas, electric, and telephone utilities in rate cases are clearly
out of proportion to small water utilitie 60 percent of which
have annual revenues of less than $15,000. Yet water utilities
are present in significant numbers, accounting for 34 percent of
all regulated utilities and 43 percent of all rate cases in 1981.
The N RRI studies have identif led strategies for improving the regu-
latory process used by numerous states.
A second finding of the NRRI studies, however, is that small
water system problems do not result entirely from complex rate
setting. Many small systems are simply not viable economic enti-
ties. Preventing future creation of such systems and encouraging
existing ones to be absorbed by larger systems may be the only
long-term solutions.
A Model of “Perfect” Public Choice
In evaluating the performance of markets, economic theory
relies on the hypothetical concept of “perfect” markets, or
“perfectly competitive” markets. Among other conditions, perfect
competition requires that there be no monopoly power among
suppliers and that all participants possess “perfect information.”
It is useful to envision how a “perfect” public choice process
might apply to water supply.
An appropriate model of perfect public choice for delivery
12 Lawton, R. and Davis, V., Commission Regulation f Small
Water Utilities: Some Issues and Solutions , May 1983, The National
Regulatory Research Institute, Columbus, Ohio.
13 Mann, P., Dreese, R., Tucker, M., Commission Regulation of
Small Water Utilities: Mergers and Acquisitions , October 1986, The
National Regulatory Research Institute, Columbus, Ohio.
3—8
-------
of local public services has been presented. 14 Assuming perfect
information, it might be possible to accurately ascertain consumer
preferences through the vehicle of an insurance sale. A water
system would offer consumers a choice between two types of
insurance policies relating to the need to install a new treatment
process for removal of contaminants.
The first type of policy would insure those concerned about
increases in their water bills against such losses in the event
that the decision is made to purchase a new treatment process.
The second type of policy would insure those concerned, about their
health against damages incurred from the contaminants if the water
utility decides not to install the new treatment process.
Consumers wishing to hedge could purchase some of each type of
insurance, thus revealing their mixed preferences. Those electing
not to participate would reveal their indifference; they would not
be “free riders,” however, as their abstinence would directly
affect their welfare. In arriving at its decision, the water
system would determine which insurance policy generated the
greatest revenue. The proceeds of the insurance sale would be used
to pay the claims of the losers, making the outcome “pareto
optimal” in that no one would be made worse off.
Upon closer scrutiny, flaws exist in this model of “perfect”
public choice. Determining the correct insurance premiums to
charge is a difficult problem, for example. However, the insurance
strategy for revealing consumer preferences highlights a crucial
point. The optimal level of service in public water supply depends
ultimately upon the level of certainty desired by consumers and the
extent of their willingness to pay an extra premium for certainty -
- for the privilege of being able to “take it for granted” that the
water is safe to drink.
It is the willingness to pay for an extra margin of safety
that is critical to optimal decisioninaking. Not only is it
difficult to assess preferences at such a margin, local
decisioninaking processes are not capable of excluding other
external factors from affecting this public choice at the margin.
Thus, regulatory intervention is necessary to establish uniform
goals for the level of safety to be achieved.
The SDWA instructs EPA to establish drinking water standards
at levels that will avoid adverse effects on the health of persons
and allow for a margin of safety. This regulatory mandate is thus
well-designed to address the central flaw that has been identified
in local public choice processes.
14 Thompson, E. A., “A Pareto Optimal Group Decision Process,”
in G. Tu].lock, ed. Papers on Non-Market Decision Makin , Univ. of
Virginia: Charlottesville, 1966, pp.133—40.
3—9
-------
3.4 Consideration of Requlatory Alternatives
EPA has proposed the IOC regulations in accordance with the
requirements of the SDWA Amendments of 1986. The SDWA mandates
that EPA publish maximum contaminant level goals (MCLGs) and
promulgate national prim r drinking water regul ions (NPDWR) for
83 specific contaminants prior to June, 1989. All inorganic
contaminants evaluated in this document are included or have been
added to this list.
National Primary Drinking Water Regulations under the SDWA
are to also include monitoring requirements. Specifically, the
Act requires that “...there must be criteria and standards to
assure a supply of drinking water which dependably complies with
such maximum contaminant levels; including quality control and
testing procedures to insure compliance with such leve1 7 and to
insure proper operation and maintenance of the system...”
Protection of Water Sources Versus Remedial Action
As an alternative to setting standards governing the level to
which contaminants must be removed prior to potable use, EPA could
devise actions to protect water sources from contamination by
inorganic chemicals. Such a strategy, emphasizing prevention
rather than remediation, could be developed using various EPA,
USDA, and FDA authorities. However, it is unlikely that any
strategy for protecting water sources could be implemented
effectively without relying on some reference levels of contam-
ination such as MCLGs and MCLs, related to the risk of adverse
health effects.
Nevertheless, considerable attention has been focused on
groundwater protection recently. EPA has developed a Groundwater
Protection Strategy and a national Wellhead Protection Program.
In addition, the need for comprehensive legislation to address this
issue has been recognized. The general structure of these initia-
tives are discussed below.
15 The 83 contaminants are listed in the Advance Notices of
Proposed Rulemaking at 47 . 45502 (March 4, 1982) and 48
Fed. .g. 45502 (October 5, 1983).
Section 1412 (b).
17 SDWA Section 1401 (1) (D).
3—10
-------
The Groundwater Protection Strateqy
In an August 1984 report entitled Groundwater Protection
Strateqy , the Office of Groundwater Protection (OGWP) outlined the
Agency’s overall strategy for addressing groundwater policy issues.
Thus far, the most important effect of the groundwater protection
strategy has been to increase the quality and quantity of
information available regarding the condition of the nation’s
groundwater resources. This information has been used extensively
in creating the framework for guiding EPA groundwater programs.
The Agency recognizes the need for consistency, as well as a need
to take current quality and vulnerability into account when
developing protection and clean-up programs to be used by the
various EPA programs. As a first step toward achieving these
goals, the Agency published a report entitled Guidelines for
Groundwater Classification Under the EPA Groundwater Protection
Strateqy in December, 1986. The classification system outlined
in the document consists of three levels representing a ground-
water’s relative “value” based on its current quality, vulner-
ability to contamination, and uses.
Development of Welihead Protection Programs
Designating protection areas around drinking water wells is
one way to protect groundwater supplies. Application of this
concept is common in Europe. At least 11 European countries,
including Germany, Switzerland, and the Netherlands have designated
zones around their public water supplies. Within these zones,
special controls are imposed on any number of potential hazards.
A growing number of states and municipalities in this country also
are beginning to create such welihead protection areas.
Depending on the nature of the contaminants involved, welihead
protection areas range in size anywhere from a distance of a few
hundred feet to several miles from wells. The hydrogeologic
characteristics of the aquifer, the extent of pumping, and the
degree of development and activity surrounding the well are the
primary criteria by which protection areas are delineated.
Management actilties commonly employed within these protection
areas include regulation of land use through special ordinances
and permits, prohibition of specified activities, and acquisition
of land. Provisions for welihead protection were adopted as part
of the reauthorization of the SDWA Amendments of 1986. This
legislation established a nationwide program to encourage states
to develop systematic and comprehensive programs to prevent
contamination of public water supply wells and weilfields within
their jurisdictions.
Historically, states have primary responsibility for ground-
water management. The Wellhead Protection Program was enacted to
both enhance state programs already underway, and to encourage
other states to begin such protection programs by providing
3—il
-------
financial and technical assistance. The SDWA specifies that all
states will participate; however, the EPA has no authority to
establish a Wellhead Protection Program if a state chooses to forgo
action on its own. There are no EPA sanctions against states that
do not participate.
The Need for Comprehensive Groundwater Legislation
In a 1984 report entitled Protecting the Nation’s Groundwater
from Contamination , the Office of Technology Assessment (OTA)
identified some 16 Federal statutes as containing groundwater
protection language but dismissed these laws as generally
inadequate. Inefficient regulations and a historical lack of
program funding were cited by OTA as contributing to a non-
preventive approach to groundwater protection. Due to the high
costs associated with groundwater clean-up technologies, OTA
advised the Federal government to take a preventive approach.
The National Groundwater Policy Forum, an organization
sponsored jointly by the Conservation Foundation and the National
Governors’ Association, released a report in November, 1985
describing Federal and state efforts to protect groundwater as
fragmented. The group stated that “...Because many of the laws
were written at different times and for different purposes, they
often dd up to a program of groundwater protection that is neither
coherent nor consistent, even if those laws are implemented to the
limits of the enacted authority.” Like OTA, the Forum has called
for a more preventive approach to groundwater protection.
Alternatives Considered in Developing Inorganic Chemical
Regulations
Since the proposed standards and monitoring requirements have
been mandated by statute, EPA is limited in its consideration of
alternative regulatory approaches and strategies for implementation
of the SDWA. EPA does not have the flexibility to consider taking
“no action,” nor can the Agency propose that states establish the
necessary standards and monitoring requirements based on Federal
guidance. Under these constraints, the Agency is comint ted to
setting explicit MCLs as close to the MCLG “as is feasible” since
18 ”Feasjble” is defined in Section 1412 (b) (5) of the SDWA as
achievable through “...the use of best technology, treatment
techniques and other means which the Administrator finds. . .are
available (taking costs into consideration).” In addition, the
Administrator must determine if “...it is economically and techno-
logically feasible to ascertain the level of a contaminant in
public water supplies” before setting an MCL (SDWA Section 1401
(1) (C))
3—12
-------
analytical methods have been identified and deemed available for
all contaminants.
At least two MCL alternatives were evaluated for each contami-
nant. Generally, these alternatives corresponded to concentrations
equal to the MCLG, more, stringent than the MCLG, and/or less
stringent than the MCLG. 9 In addition, for those contaminants
which are currently regulated, MCL alternatives equivalent to the
interim standards were evaluated. Exhibit 3-2 summarizes the MCL
alternatives which were evaluated for each inorganic chemical
contaminant.
E IBIT 3-2
MCL Alternatives for lOCs (ug/l)
More Less Interim
Containin 8 t Stringent Proposed Stringent Standard
Asbestos N/A 7 nil. N/A N/A
Barium 100,1500,2000 5000 N/A 1000
Cadmium 0.1 5 50 10
Chromium 10 100 1000 50
Mercury 21 0.2,3 2 4,10 2
Nitrate 100,1000 10,000 N/A 10,000
Selenium 5 50 100 10
Alternatives and Basis for Monitoring Requirements
EPA developed and examined several alternative sets of mon-
itoring and reporting requirements for the IOC regulations. The
monitoring and reporting alternatives and the basis for the options
selected for each rule are discussed in the Information Collection
Request (ICR) submitted for the proposed rule. The ICRs are
19 Alternatjve MCLs were obtained from Memorandum from Joe
Cotruvo dated June 13, 1985 and entitled “Phase II Regulatory
Impact Analysis Assumptions.”
20 Concentration in fibers 10 un or longer/liter.
21 No impacts were evaluated for nitrite standards under the
assumption that additional incremental impacts beyond those
calculated for nitrate would be negligible.
3—13
-------
submitted in accordance with the Paperwork Reduction Act of 1980.
Costs associated with the preferred monitoring alternative were
estimated and are presented in Chapter 4.
In developing monitoring requirements for lOCs, EPA considered
which level of government was most appropriate and provided con-
siderable discretion to state regulatory agencies for determining
precise monitoring standards based on local conditions. Further,
to take into account the complexities of collecting representative
samples in individual systems, and the potential for variability
of water quality over time in surface and ground water systems, the
following variables were considered in developing monitoring
options:
o Nu4er of samples to be taken:
o Frequency of sampling;
o Location of sampling points;
o Which systems should sample;
o When compositing of samples should be allowed; and,
o When the sampling should be performed.
As noted in Chapter 2, impacts of monitoring requirements for
contaminants related to corrosion of plumbing and distribution
system materials ..e., lead and copper) are evaluated in a
separate document.
EPA, Office of Drinking Water, 1988. Draft Re u1atory
Impact Analysis of Proposed National Primary Drinking Water
Recrulatjons for Lead and Copper .
3—14
-------
4. ASSESSMENT OF COSTS
4.1 Introduction
This chapter presents the results of analyses of cost impacts
of regulatory alternatives for controlling lOCs in drinking water.
The sections below describe the major assumptions that were used
in the analyses, and present estimates for treatment and waste
disposal costs along with a discussion of the uncertainty assoc-
iated with those estimates. Finally, cost estimates for compliance
monitoring and unregulated contaminant monitoring are presented
along with a discussion of the monitoring requirements for public
water supplies.
4.2 Major Assum tjons Used in Cost Estimates
•Three critical assumptions serve as the basis for the cost
estimates presented in this chapter. First, it is necessary to
avoid misallocation of costs associated with current regulations
to those resulting from the proposed revisions. With the exception
of asbestos and nitrite, all contaminants evaluated in this
document are currently regulated under interim standards. It was
assumed that no impacts are attributable to revised drinking water
MCLs set equal to or greater than those currently in place. Under
this assumption, it is therefore necessary to determine incremental
impacts for contaminants having MCL alternatives below the current
standards. That is, to estimate the impacts to only those systems
affected at concentrations between the proposed revised MCL and the
current, interim MCL.
At the most stringent (i.e., lowest) MCL alternatives, an
assessment of net incremental cost impacts is necessary for all
lOCs evaluated. At proposed MCL alternatives (ike., those cor-
responding most closely with proposed MCLG5 for most contaminants),
it is estimated that net incremental costs will result for only
cadmium.
The second major assumption concerns the consolidation of
occurrence data across different system size categories. In
developing estimates of occurrence probabilities, raw survey data
from several (or in some cases all) size categories were combined
and analyzed as a group. As mentioned in Chapter 2, this procedure
resulted in invariant occurrence probabilities across system size
categories for lOCs (see Appendix II). However, because the esti-
mates are based on sampling data from finished water, and since
systems serving greater than 50,000 persons generally have more
effective treatment processes, more comprehensive monitoring, and
a much higher level of operator expertise in place, it may be
inappropriate to assume that larger systems have the same
likelihood of occurrence as smaller systems.
4— ].
-------
Assuming even minute occurrence probabilities in large systems
may have a large impact on cost estimates. The cost model is
structured such that the application of any occurrence probability
greater than zero results in the calculation of at least some
“fraction” of a large system being affected. This, in turn, may
lead to the overestimation of aggregate impacts since unit costs
for treatment technologies and average population served in large
systems are high.
For these reasons, it was assumed that bC occurrence in large
systems is probably negligible relative to that in smaller systems.
This is supported by FRDS data which indicate that as of 1986, only
18 large systems were re orted out of compliance for all interim
bc regulations combined (see Exhibit 4-1). The summary tables
presented in Appendix III for each contaminant provide estimates
of total costs both including and excluding occurrence in systems
serving greater than 50,000 persons.
Finally, as discussed in Chapter 2, occurrence data on mercury
in surface water systems is subject to extreme uncertainty. Fo
this reason, cost estimates which include mercury occurrence
estimates, as well as estimates using the assumption that mercury
occurrence is zero are presented in Exhibit 4-2, and in the summary
table in Appendix III.
4.3 Treatment Costs for bOCs
Draft estimates of aggregate national treatment costs for
proposed alternative MCLs for lOCs are based on individual results
for six contaminants: barium, cadmium, chromium, mercury, nitrate,
and selenium.
Because “nitrite levels in tdr nking] water are considered to
be negligible relative to nitrate,” and due to nitrite’s tendency
to be oxidized to nitrate during disinfection, it was assumed that
incremental impacts of the nitrite regulation beyond those
calculated for nitrate would be negligible and no further analyses
were performed for this chemical.
Although approximately 609 systems nationwide were estimated
to have asbestos concentrations in excess of 7 MFL, it is assumed
that treatment and waste disposal costs for this contaminant will
not be attributable to this rule based on the following rationale.
Natural and anthropogenic contamination of source waters is
1 0f these, 12 were violations of the nitrate standard.
2 Science Applications International Corporation, 1987.
Estimated National Occurrence and Exposure to Nitrate/Nitrite in
public Drinking Water Supplies (Revised Draft) .
4—2
-------
EXHIBIT 4—i
FRDS IOC Violations in Community Water
Systems Serving More than 50,000 persons
Number of FRDS Violations
bC Groundwater Surface Water Total
Barium o 0 0
Cadmium 0 1 1
Chromium 1 1 2
Mercury 0 2 2
Nitrate 4 8 12
Selenium 0 1 1
Total 5 13 18
* Source: SAIC Draft Occurrence Documents for Inorganic Chemicals
4—3
-------
EXHIBIT 4—2
Emtimated Ireateent and Waste Disposal Costs at Most Stringent MCI Alternatives 1 .2 .1
Treatsent Costs Waite Costs lotal Cost5 Asnsaliied Costs it mfyrl
Womber 0
c c l : Systems 1 Capital Ott Capital DIM Capital OtC
Ccntamisaot tog/Il : Altucted I 45 m l U m/yrl 8 ml 15 u/yr t I I m l 1 m/yrt 4 32 51 4 12
Estimated Cotts at Moat Strinqeot MCI.s
Barium I C C 6 ,713 2,090 206 : 2,032 I i ? - 4,122 313 - - — - - -
Cadmium 0 I l 7 60 : 196 lB 121 7 323 23 : - - - - — -
Chromium 10 : 1323 765 76 572 46 1 1,337 (74 - - - - - -
:Mercarr 02: 1,376: 1,206 137 234 16 1,440 iss: -- - - — -
Witratetfitrile too : 108,227 22,355 3,456 : (2,536 611 : 34611 4,057 - - - - - -
Setenism S : 1713 : 4 09 4) 617 42 : loll 69 : - - - - — -
Octal : 120114 3 27,001 3,942 0 16,113 651 43,114 4,793 1 - - - - — -
total less Ileecury 1 118,73.6 1 25,795 3,605 1 15,939 633 1 41,754 4636 I — - - - - -
Estimated Costs ol torrent MCLs Promulgated lJsder NIPOWI
:Oariuu 1,000: 19: 5 3 : 5 0 1 10 i: - - - - - -
:Cadoium I I I 166: 39 4: 31 2: io i:-- - - - —
thromios 3 0: 20: 19 21 9 I l 2) 21-- - — - -
I Mercury 2 : 445 : 553 67 1 99 9 : 652 71 : - - - - - -
Wmtrateflti trite 10,000 2,577 1 469 6) : 264 12 1 733 79 1 - - — - - —
iSelesiom it: (851 45 5 : 3 0 1 46 5: - - - - - —
total : 3,364 1 1,130 141 : 410 24 1 1,540 (65 1 — — — — - —
total less Mercury : 2,939: 577 79 1 311 IS : 686 94 - — - - - —
Emtimated Wet Incremental Costs at Most Stringent MCL5
Oarioa : 6,694 1 2,065 205 : 7,027 137 1 4,112 342 1 6 (6 671 130
Cadmium : 5 ) 2: 157 14: 96 5: 253 19: 36 39 43:
Chro.iom 1 1,303 1 141 76 : 564 43 : 1,310 121 1 209 276 745 1
Mercury 9351 653 1 5: 135 9: 706 641 (37 :47 156:
Witratu/Witnte 1 105,700 21,866 3,389 12,212 589 1 34,156 3,9)6 1 6,273 6,717 7,700
Selesins 1 1,529 1 364 42 : o9 42 1 1,033 64 1 153 :67 62
total 1 116,130 1 25,871 3901 1 15,165 627 1 41,634 4,628 7,426 7,969 8,556
total lean rcory 115,797 I 25,216 3,126 15,628 I I I 1 40,846 4,544 t 7,289 7,622 6,400 I
I lnciudes community and san-transient son-community mater eyntemi serving tune than 50,000 people fur elI coituminasts
emcept #itralelWitrite. Witeate/Witrite estimatui include all psblec mater muppliet serving temer than 50,000 people.
2 Costs eipsemsed in 1966 Dollars and ore asnualeued over 20 yeaei.
3 61 most strio tnt MCL altereativem therm- may be significant co—occurrence ol lOCu
-------
expected only in surface water systems in restricted areas of the
country (i.e., primarily California and Washington state). These
systems will also be subject to the requirements of the Surface
Water Treatment Rule (SWTR) which will be in place prior to
promulgation of the IOC rule. Filtration technologies installed
to comply with the SW’TR will also be effective for removal of
asbestos from source waters and would therefore largely subsume any
treatment costs for removal which would otherwise be incurred under
the IOC Rule.
Corrosion of asbestos/cement (A/C) pipes in distribution
systems has been shown to be another significant source of asbestos
contamination in drinking water. Although a large number of public
water supplies use A/C pipe, corrosivity problems in virtually all
water systems nationwide will be mitigated through implementation
of the corrosion control provisions promulgated in conjunction with
the NPDWR for Lead/Copper/Corrosion Control. It is assumed that
reducing the corrosivity of the water toward lead and copper will
also be effective in eliminating corrosion of A/C pipe. While it
is not possible to predict precisely, it is assumed that the number
of systems for which additional, centralized treatment will be
necessary (beyond that adopted for the SWTR and the corrosion
control rule) for control of asbestos will be negligible and no
further cost analyses were conducted for this contaminant.
The basic algorithm utilized in obtaining estimates for total
national costs associated with treatment is relatively
straightforward. For each system size category, for both ground
and surface water systems, the number of systems with contamination
above a given MCL alternative is estimated according to the
methodology described in Chapter 2.
These estimates are then merged with a compliance decision
matrix which predicts the relative likelihood of a system of a
given size and source choosing various treatment technologies or
other compliance options (see Appendix IV). These matrices
represent a consensus based on the best judgement of professional
engineers, WMA analysts, and EPA staff.
The result of these procedures is to provide an estimate of
the number of affected systems that would use each treatment/com-
pliance option nationwide. These estimates are then mult4.plied by
the appropriate unit costs for each technology or option detailed
in the CIT documents (see Appendix V).
Exhibit 4-2 presents a summary of net incremental treatment
and waste disposal costs associated with the most stringent MCL
alternative for each contaminant in systems serving less than
4—5
-------
50,000 persons. 3 Cost estimates are in 1986 dollars and were
annualized using social discount rates of three, five, and seven
percent over 20 years.
Net incremental capital costs associated with the most
stringent MCL alternatives would be approximately $40.8 billion,
excluding mercury. The incremental operation and maintenance costs
are approximately $4.5 billion per year, resulting in annualized
costs of between $7.2 billion and $8.4 billion. Two important
factors should be noted regarding the cost estimates at the most
stringent MCL alternatives, however. First, the concentrations
used for regulatory alternatives are less than the PQLs for several
contaminants. As such, these estimates reflect impacts that would
occur only if standards were set below thresholds of readily
achievable quantification and discernable adverse health effects.
Second, these costs may represent an overestimate insofar as the
“double counting” of costs in systems with multiple occurrence has
not been taken into account. Due to these two caveats, these
estimates should be used for comparison purposes only.
Because of its ubiquitous presence at low concentrations, and
because all CWS and NCWS are assumed to be subject to the
nitrate/nitrite regulation, the vast majority of the annualized
costs (83 percent) are attributable to these substances at the most
stringent MCL alternatives.
Net incremental costs associated with the proposed MCL
alternatives for systems serving less than 50,000 persons are
presented in Exhibit 4—3. Capital costs at the preferred MCL
alternatives are approximately $73 million, with O&M costs of $6
million per year. This results in annualized costs of between $11
million and $13 million per year depending on the discount rate
applied. Appendix III contains cost summary tables for each
contaminant at all MCL alternatives at which occurrence was noted.
It should be noted that the estimates provided in the Appendix
III tables are cumulative and therefore include costs associated
with systems having contaminant levels exceeding the current stand-
ards. For this reason, the costs presented in the tables for the
interim MCLs (i.e., cadmium at 10 ugh) must be subtracted from the
corresponding costs presented in the tables for the proposed
3 As described in Chapter 2, it is probable that occurrence
estimated in systems serving more than 50,000 persons was an
artifact of the method used to estimate occurrence. Under the
assumption that occurrence in these systems is zero, the estimates
in Exhibits 4—2 and 4—3 represent WMA’s “best” estimates.
4 Unhike the co-occurrence estimates presented in Chapter 2,
it is possible that co-occurrence at the most stringent MCL
alternatives may be significant.
4—6
-------
EXIIIBIT 4—3
Wotiaated treataest sod Waite Phapoaai Costa at Proposed MCI. literaatie,a
Teeniest Cost. Waste Cost. 3 : total Coats 3 ;oonualised Coats ($ a/pr) 3 :
Pushers :: : :
MC I Spates, Capital 008 : Capital O I l : Capital Oil
Costoaisaot lug/I) : Iffeeted I 1 ii i i a/pr) : ii . 1$ a/pr) : a .) i i a/pr) : S 30 S 5% S 7%
fat iaated Coat. of Proposed ICLa
Cadai na : 380: 79 8 61 1 1(3 i i :-- - - - -
lstlaated Coats of Carreat WCLa Proazigated Qoder WiPDWi
4 ; Cadajua l0 i sa: 30 i: 3) 2: to 6:-- - - - -
ktiaated Wet iscreseotal Costa of Proposed ICLa
:Cadaioa : 192: 40 o: 33 2 73 6 Ii 12 i3
I the preferred ICLo for barioa, chroaasa, aercarp, nitrate, sod eeieeisa are greater than or .qaai to the latersa WCLo,
hence the Iscreaootai lapact . of reaoel ,g these coztaaioosta are aaaeaed to he zero Nitrite occorreoce is aoeuaod to he
oeglagihie, therefore so heoefsta or teeataeot aad aaate disposal coata oouid he incurred for this cootaaasaat Benefits
and treataeot disposal ceat. for eabe.toa are aobaaaed bp othar drinking aster regsiatioas
2 include coaaooitp and aos-traaaieot aoa-coaauaitp aater apatesa serving less thao 50,000 people
3 Cooto are eaproooed in 1906 Dsiiara and are annualized over 21 pears
-------
revised MCLs (i.e., cadmium at 5 ug/l) to obtain the incremental
costs described above.
In addition to the assumptions discussed in Section 4.1,
several significant sources of uncertainty and a number of other
assumptions made in the analysis contribute to the error inherent
in the bC cost estimates. These assumptions arid sources of
uncertainty are discussed below.
In the Cost and Technology (C/T) documents, the Criteria and
Standards Division (CSD) assumed that some contaminants would occur
predominantly in either ground or surçace waters, and capital and
O&M costs were developed accordingly. However, occurrence data
show that some contaminants assumed (in the C/T document) to be
prevalent in groundwater were in fact more predominant in surface
water systems and vice versa. To compensate for this discrepancy,
unit treatment cost estimates were adjusted (where necessary)
according to the following procedure. Capital costs, provided as
total costs in thousands of dollars, were divided by maximum daily
production values provided for each system size to obtain costs per
million gallons of capacity. Similarly, 0&M costs, provided as
thousands of dollars per year, are divided by the product of
average daily production values (in million gallons per day), 365
(days per year), and a conversion factor of 1000 to obtain costs
in million dollars per million gallons of production. These adjus-
ted, “per unit” costs were then incorporated into the cost model
and multiplied by the applicable design flow parameters.
Based on estimates of average influent concentrations of lOCs
in affected systems and the proposed MCL alternatives, the
percentage removal required (see Appendix II) for compliance in
most syst ms will generally be less than that assumed in the C/T
documents to derive unit treatment cost estimates. The cost
estimates are based on the removal efficiencies which corresponded
most closely to the average concentrations estimated using the
Replicate model and the proposed MCL alternatives.
Unit costs associated with regionalization and alternate
source development were derived in the C/T documents using
assumptions that resulted in unreasonably high estimates for these
compliance options. These cost estimates, in turn, were used in
5 This distinction is important because costs of constructing
and maintaining treatment systems may vary significantly according
to source due to dissimilar design flow parameters and other
factors.
6 USEPA Office of Drinking Water, Criteria and Standards
Division, 1987. Cost Supplement to Technologies and Costs for
Removal of Inorganic Chemicals from Potable Water Supplies .
Separate document developed for each contaminant.
4—8
-------
developing compliance decision trees. Given the high cost
presented, it was not considered likely that small systems would
frequently choose these options. Instead, higher probabilities
were assigned to centralized treatment. Lower cost assumptions for
alternate source (e.g., the cost of drilling a well), and
corresponding adjustment of the compliance decision trees would
result in a decrease in total national costs for the smallest four
system size categories. A sensitivity analysis was conducted to
evaluate the effect of this parameter on annualized treatment cost
estimates. It was determined that altering these inputs to reflect
more realistic assumptions will decrease annualized treatment costs
by less than ten percent and would therefore be well within the
bounds of uncertainty described in Section 4.4 below.
Two sets of unit cost estimates were developed by CSD for most
central treatment options. The low cost estimates, which assume
that no special pretreatment of source water is necessary for
effective removal of lOCs, were used as inputs for the ATm runs.
The high cost estimates incorporate costs for such practices as
softening and filtration as pretreatment to optimize bC removal
in waters with high turbidity and/or hardness. Because the struc-
ture of the decision trees already incorporates the likelihood of
installation of filtration technologies for compliance with the
Surface Water Treatment Rule, it was assumed that the portion of
the “high” unit costs associated with filtration as a pretreatment
would not apply to surface water systems. For groundwater systems,
a similar assumption was made since turbidity is naturally very
low. It was then possible to estimate the portion of the high cos
estimates associated only with hardness removal as a pretreatment.
These costs were assumed to represent the net difference between
the high and low unit cost estimates. Although it was not possible
to estimate the number of systems to which these “high” unit cost
estimates would apply, compliance costs may be slightly higher in
systems affected with bC contamination that also have to pretreat
to control hardness.
Some uncertainty is also associated with costs for the use of
point-of-use (POTJ) devices as a compliance option for lOCs.
Although the cost and technology documents provide estimates for
POU, the decision trees assume that the only acceptable non-
centralized compliance option will be point-of-entry (POE) devices.
Nevertheless, the POU costs were used as a surrogate for this
option. Finally, the cost estimates associated with the least
expensive POU technology were used in this analysis. Because a
very small portion of total costs are associated with the possible
use of POLl/POE devices as compliance options, the error introduced
by this assumption is probably negligible.
7 Malcolm Pirnie, Inc. Personal Communication, June, 1987.
4—9
-------
4.4 Waste Disposal Costs
Estimates of waste disposal costs for lOCs are included in
Exhibits 4-2 and 4-3 and in Appendix III. The methodology employed
to derive these estimates is similar to that used in calculating
treatment costs. The number of systems using each water treatment
option was applied to a decision tree for waste treatment/disposal
options (see Appendix IV). Through this procedure, an estimate of
the number of affected systems which would choose each waste treat-
ment option was derived nd multiplied by the unit waste treatment
costs published by CSD. This procedure was repeated for each
contaminant at each MCL alternative. Based on a review of the
provisions in the current Federal regulations pertaining to
hazardous waste disposal, it was assumed that process wastes are
not subject to requirements for hazardous wastes under current
Resource Conservation and Recovery Act (RCRA) regulations.
Disposal methods included direct discharge to sanitary sewers,
direct discharge into receiving waters, sanitary landfill, and land
application.
Since the structure of the cost model does not allow differ-
entiation by source in systems affected, unit waste disposal costs
for groundwater were used if the majority of systems affected were
groundwater systems, and vice versa. The unit waste disposal cost
data are presented as curves (see Appendix VI) and had to be tran-
scribed in tabular form; some error is inherent in this procedure.
In addition, no waste disposal costs were estimated for systems
which would opt to modify existing central treatment for compliance
since cost curves were not generated for “modification” technol-
ogies. Finally, zero waste disposal costs were assumed for systems
using POU, alternate source, regionalization, and GAC as
treatment/compliance options.
4.5 Uncertainty in Estimates of National Costs
Introduction
Uncertainty calculations discussed in this section have been
conducted for impacts associated only with contaminants for which
incremental treatment and waste disposal costs will be incurred at
the proposed MCL alternatives (i.e., those cost estimates presented
in Exhibit 4-3 above). The magnitude of uncertainty associated
with other cost estimates (e.g., those calculated at the most
stringent MCL alternatives) may vary significantly due to the
possibility of co-occurrence and other factors.
8 USEPA Office of Drinking Water Criteria and Standards
Division, 1986. Technologies and Costs for the Treatment and
Removal of Waste By-Products from Water Treatments for the Removal
of Inorganic and Radioactive Contaminants . Revised draft.
4—10
-------
Methodoloqy
There is a degree of uncertainty inherent in any estimate of
compliance costs. This uncertainty arises from the fact that only
a small number of observations are available for many of the
factors that contribute to overall costs. These factors include
the following:
o The frequency and degree of contamination from any given
chemical:
o The likelihood that a system will select a particular
treatment technology; and,
o The unit costs of available treatment technologies.
The errors in the estimates of these input variables form the basis
of the overall uncertainty computed for compliance costs.
The form of the fundamental equation underlying the cost
calculation facilitated the estimation of the magnitude of the
errors. This equation, as embodied in the ATm model is as follows:
C = P 1 * N 1 * * jk * Cik
ijk
Where:
I designates the category of water system (12 sizes times
two water sources);
j designates the treatment already in place;
k designates the type of additional treatment selected to
meet the IOC standard;
P 1 is the probability that a system in the ith category
exceeds the MCL for the bC in question;
N 1 is the number of water systems in the ith category;
is the probability that a system in the ith category will
choose treatment j to meet the filtration requirement;
is the probability that a system with treatment j already
in place will select treatment k to meet the bC
standard; and,
Cik is the unit cost (typically, dollars per system) of
treatment k for a system in category i. (This term may
represent either capital or O&M costs).
4—li
-------
All of these variables except N are treated as random variables,
each having a distribution with a mean value (the deterministic
value used in the ATm model) and a standard error.
To evaluate the product inside the summation, the following
simplifying assumptions were adopted:
o All random variables are approximately normally distri-
buted; and,
o All independent variables are uncorrelated with each other.
These assumptions allow the expression of the standard error of a
multiplicative equation such that if
a=x*y * z
where x, y, and z are approximately normally distributed with means
X, Y, and Z and with standard errors Sy and S 1 then Sa can be
approximated by:
Sa = [ x 22 )*(sy 2 2)*(sz2+ 2) —
Performing the summation requires a method for accumulating
the standard errors of the internal products. To accomplish this,
it was assumed that random variables to be summed were either per-
fectly correlated or perfectly uncorrelated, depending on the
variables in question. Although an oversimplification, this
assumption eliminates the need to consider cross-correlation terms,
which make the calculations extremely complex and for which no data
are available. Specifically, the procedure involved the following
assumptions.
o Errors in the costs of similar treatments for a given bc
are perfectly correlated. For example, lime softening
could be used to remove a contaminant in (a) systems having
no filtration in place (provided that conventional
treatment or direct filtration were added), (b) those
already having direct filtration, or (C) those having
conventional treatment in place. This assumption means
that if the cost estimate in case (a) is of f by +10
percent, then the estimates in the other two cases are also
off by +10 percent.
o Errors in the costs of dissimilar treatments for a given
bC are perfectly uncorrelated. That is, a +10 percent
error in the cost of lime softening gives no information
about the error in the cost of reverse osmosis.
4—12
-------
o Within a given water source (surface or ground), errors n
cost estimates across “small” system size categories (de-
fined as all systems serving 3,300 people or fewer)
are perfectly correlated. Similarly, errors across all
“large” size categories (i.e., systems serving more that
3,300 persons) are perfectly correlated.
o Errors across water sources and between “small” and “large”
size categories (as defined above) are perfectly
uncorrelated.
o For a given system size and water source, errors in total
capital costs are perfectly correlated with errors in
operating and maintenance costs.
These assumptions allowed the calculation of the standard error of
the summed random variables x, y, and z through the following
expression:
a = x + y + z.
The equation for the case where the errors in x, y, and z are
perfectly correlated is:
Sa = + S , + S .
For the perfectly uncorre].ated case, the expression is:
Sa = + Sy 2 + Sz 2 )’/ 2
Standard Errors of Independent Variables
The formulas for estimating standard errors were implemented
on a spreadsheet. Mean values for independent variables were taken
from the input files for the ATh model. The errors in the
probabilities of exceeding the MCLs were estimated by SAIC on the
basis of sample data and a curve—fitting procedure. The other
errors were estimated by staff of WMA and Malcolm Pirnie, Inc. who
were responsible for generating the decision trees and unit
treatment costs. The magnitudes of the standard errors of the
independent variables, expressed as a percentage of their mean
values, are summarized in Exhibit 4-4 below. In the interest of
simplifying the calculation, the five percent error in the prob-
ability of selecting a particular filtration method was ignored;
a sensitivity analysis showed that it contributed less than one
percent to the estimate of the standard error of national cost.
4—13
-------
EXHIBIT 4—4
Standard Errors of Independent Variables in bC Cost Calculation
(As a Percent of Mean Values)
Standard Error as a
Decimal Fraction of the Mean
Probability of a system
exceeding the MCL for a
given bC (Pd): Surface Ground
Cadmium (MCL = 5 ug/1) (No occurrence) .43
Probability of selecting a 0.05 0.05
particular filtration
treatment (Pjj)
Probability of selecting a 0.15 0.15
particular ICC treatment
Unit treatment cost (cik) 0.25 0.25
Standard Errors of National Costs
Using the procedure described above, the standard errors of
the national costs of compliance were estimated for cadmium at 5
ugh. The results are given in Exhibit 4—5. The proportions shown
in the Exhibit apply to capital, O&N, and annualized costs, on the
assumption that errors in capital and O&M costs are perfectly
correlated. It also applies to the compliance cost (the cost
of meeting the proposed MCL minus the cost of meeting the NIPDWR)
under the assumption that errors in the costs of meeting the two
sets of standards are perfectly correlated.
Results presented in Exhibit 4-5 involve two assumptions: a)
that errors between “small” and “large” systems are perfectly
uncorrelated; and, b) that these errors are perfectly correlated.
In the absence of better information, there is no way to choose
between these two cases. The exhibit shows that the standard
errors in cost range from 20 to 28 percent of the mean values for
cadmium.
4—14
-------
EXHIBIT 4—5
Standard Errors in Compliance Cost Estimates
(As Decimal Fractions of Mean Values)
Assumption: Errors between
large and small sy stems are:
Compliance Cost by Contaminant: tJncorrelated Correlated
Cadmium (MCL = 5 ugh) 0.204 0.284
4.6 Monitorjn Costs
Monitoring cost estimates for contaminants addressed in this
document are summarized in Exhi1 jt 4-6 (more detailed estimates
are displayed in Appendix VII). Estimates include analytical
costs for compliance monitoring averaged over an 11 year period
and were derived for asbestos, barium, cadmium, chromium, mercury,
nitrate, nitrite, and selenium. In addition to the compliance
monitoring, costs were calculated for monitoring for the following
unregulated contaminants: antimony, beryllium, cyanide, nickel,
sulfate, and thallium.
Monitoring cost estimates for lOCs were derived using the
proposed requirements outlined in the following sources:
o National Primary Drinking Water Regulations for Organic
and Inorganic Contaminants and Monitoring for Unregulated
Contaminants - Draft Proposal, Office of Drinking Water,
January 18, 1989.
o Briefing on Monitoring Requirements for Public Water Sys-
tems, Office of Drinking Water, March 6, 1987.
o Joseph Cotruvo, memo to Arnold Kuzmack, April 13, 1987.
o Maria Gomez-Taylor memo to David Schnare, September 30,
1986.
o Personal communications with Office of Drinking Water,
Criteria and Standards Division.
9 As mentioned above, monitoring requirements for contaminants
generally considered to be of concern due to by products of
corrosion will be analyzed separately.
4—15
-------
EXHIBIT 4—6
NOTE: ALL IOC monitoring cost modeLs extend over a 11 year period. This aLLows for one
initiaL round and one repeat round for aLL water systems subject to reguLation.
CONTAMINANT I
I ASBESTOS I
I BARIUM
CADMIUM I
CHROMIUM
MERCURY I
I NITRATE I
NITRITE
I SELENIUM I
COST OF PROPOSED
MCL ALTERNATIVES
(In MiLLions of 1956
DoLLars)
COST
OF
MCL
MOST STRINGENT
ALTERNATIVES
I
MCL
(ug/L)
LOW
HIGH
MCL
(ugIL)
LOW
HIGH
7 MFL
1.0
1.7
7 MFL
1 0
1.7
100
0.2
0.4
5,000
0.1
0.2
0.1
0.1
0.2 j
5
0 1
0 2
10
0.1
0.3
100
0.1
0.2
0.2
0.3
0.3
2
0.2
0 3
1,000
3 0
1. 4
10,000
1.3
1.4
1 ,000
1.1
1.1 I
1,000
1.1
Li
5
0.3
0.4 I
50
0 2
0.2
I
I TOTAL I $6.1
CURRENT COSTS $2.0
I INCREMENTAL COSTS I $4.1
$8.8 I $4 1 $5.3 I
$2.0 $2.0 $2.0 I
$6.8 $2.1 $3.3 I
4—16
-------
Based on the requirements outlined in these sources, models
were constructed for calculating high and low range estimates for
average annual monitoring costs for each contaminant. The results
of all calculations are summarized in Appendix VII for all
contaminants at the various MCL alternatives evaluated.
At the proposed MCL alternatives, the total average annual
low bound costs (considered the “best” estimate) of compliance
monitoring for these bC contaminants are estimated at $4.1 million
per year while the high bound costs are estimated to be $5.3
million per year. Total costs of unregulated contaminant
monitoring under the rule are estimated to range from $0.2 million
to $1.6 million as a one time cost.
Like the procedure for estimating treatment and waste disposal
costs, it was necessary to subtract current monitoring costs
attributable to the NIPDWRs for these contaminants to obtain the
estimates of incremental cost. The estimated total cost of
monitoring for ten inorganic contaminants under the interim
regulations is reported in the Pub ic Water System Supervision
Program (PWSSP) ICR as $3.3 million. Of the ten inorganics which
are currently regulated, six contaminants are addressed in the
proposed regulation and evaluated in this document. It is esti-
mated that, of the total costs of IOC compliance monitoring under
the interim regulations, $2.0 m lion is attributable to monitor-
ing for these six contaminants. Using this figure, incremental
monitoring costs for lOCs are estimated at $2.1 million per year
under the low bound scenario, and $3.3 million per year under the
high bound at the proposed MCL alternatives.
Description of Monjtorjn Requirements
The EPA’s goal is to establish monitoring requirements for
lOCs that will ensure compliance with the proposed NPDWRs in the
most efficient manner possible. The program is intended to target
monitoring efforts on contaminants most likely to be present in
individual systems. This program is based on several general
concepts, including:
o Requiring states to conduct regular vulnerability
assessments to account for changes in the potential for
contamination over time;
10 lnformatjon Collection Request for the Public Water System
Supervision Program , Appendix H, U.S. Environmental Protection
Agency, Office of Drinking Water, March 13, 1986.
11 The cost estimate for the six contaminants was calculated
by apportioning the total cost per sample for inorganic analyses
reported in Appendix H of the PWSSP ICR for the ten contaminants.
4—17
-------
o Providing extensive discretion to states to tailor
monitoring requirements based on systems’ vulnerability to
various contain inants;
o Allowing the use of recent monitoring data in lieu of new
data, where applicable;
o Allowing the use of historical monitoring data in making
vulnerability determinations;
o Designation of sampling schemes that allow for simultaneous
monitoring for all regulated contaminants; and,
o Focusing sampling on times when vulnerability is highest
for contaminants that fluctuate seasonally.
Under the proposed NPDWRS for inorganic chemicals, all com-
munity water systems and non-transient, non—community systems would
be required to perform both compliance and unregulated contaminant
monitoring. In addition, all non-community water supplies would
be required to perform monitoring for nitrate/nitrite. Six of the
lOCs to be regulated in Phase II (i.e., asbestos, barium, cadmium,
chromium, mercury, and selenium) are classified as “Tier II” con-
taminants under the proposal. Tier II contaminants are of suff i-
cient concern to warrant national regulation, but occur in a
predictable fashion, justifying flexible national monitoring
requirements to be applied by state authorities.
Nitrate and nitrite are classified as “Tier I” contaminants.
Tier I contaminants occur with sufficient frequency and are of
sufficient concern to warrant national regulation and consistent
monitoring and reporting. The following sections discuss the
proposed monitoring requirements for each of the two contaminant
tiers and for the unregulated contaminants.
A) Tier II
The minimum monitoring requirements for barium, cadmium,
chromium, mercury and selenium are as follows. Ground water
systems must monitor every three years and surface water systems
must monitor annually. States may reduce the monitoring
frequencies to no less than every ten years if all results of at
least three rounds of monitoring are less than 50 percent of the
MCL. The first round of monitoring for these contaminants must be
initiated within 30 days from publication of the final regulation
and must be completed within 18 months from publication. Tier II
monitoring requirements for metals are summarized graphically in
Exhibit 4—7.
States will base their decision on monitoring frequencies for
each system on various factors, including: 1) reported levels from
4—18
-------
EXHIBIT 4—7
INORGANIC CHEMICALS*
Proposed Monitoring Requirements
I I
Surface Water
Annually
* Regulations cover Barium, Cadmium.
Chromium. Mercury, and Selenium
One or More Previous
Analytical Results
50% of MCL
Ground Water
Every 3 Years
Initial
Three Rounds
of Monitoring
All Previous Analytical
Results < 50% of MCL
-‘-
Repeat Monitoring No
Less Than Every 10
(State Discretion)
T Ground Water
Every 3 Years
I
Surface Water
Annually
I
-------
previous monitoring data; 2) the degree of variation reported in
the source waters; and 3) other factors that may affect contaminant
levels, such as changes in pumping rates for ground water supplies
and changes in stream flows.
Surface water systems will sample at points in the distribu-
tion system that are representative of each source or at each entry
point to the distribution system that is located after any
treatment. The number of samples will be determined by the number
of sources or treatment plants. Sampling will be done at entry
points to the distribution systems for ground water systems and the
number of samples will be determined by the number of entry points.
This approach will make it easier to pinpoint possible contaminated
sources within a system. In both surface and ground water systems,
the proposed sampling locations are such that they may be used for
the simultaneous collection of samples for other source related
contaminants.
The total number of samples may be reduced at state discretion
by the use of composite samples. Composite samples of up to five
sources or entry points would be allowed. If the concentration in
the composite sample indicates that one or more of the individual
samples may exceed the MCL, follow-up sampling would be required
at each sampling point included in the composite.
If the result of one analysis exceeds the MCL for a given
contaminant, the system is in non-compliance and procedures for
public notification must be followed. States have the discretion
to require that a confirmation sample be collected within two weeks
at the same sampling location to verify the original finding. In
this case, if the average of the two samples analyzed exceed the
MCL, the system is in noncompliance and procedures for public
notification must be followed. The state may specify additional
monitoring beyond the federally mandated minimum requirements for
such systems.
B) Tier II: Asbestos
The EPA is proposing that only vulnerable systems that may
have high levels of asbestos fibers greater than 10 urn in length
monitor for asbestos. States will have the discretion to determine
which systems are considered vulnerable for asbestos based on: 1)
potential contamination of the water source; 2) the use of asbes-
tos-cement pipes for finished water distribution; 3) the corro-
sivity of the water; and, 4) the potential that the concentration
of fibers greater than 10 urn might approximate 7 MFL.
The compliance monitoring requirements for asbestos will
include a one-time monitoring round for all vulnerable systems.
Sampling will be conducted at the entry points to the distribution
system if the contamination is due to raw water quality, or at
4—20
-------
representative points in the distribution system if contamination
is due to asbestos-cement pipe. Both potential sources of
contamination may be considered for some systems, requiring that
samples be collected both at entry points to the distribution
system and at the tap.
The minimum number of samples collected will be one sample
per source or treatment plant for surface water supplies and one
sample per entry point to the distribution system for groundwater
supplies. The repeat monitoring requirements for systems with
results not exceeding 50 percent of the MCL will be required at the
discretion of the states. For systems with results exceeding 50
percent of the MCL, repeat monitoring will be required every three
years for ground water systems and annually for surface water
systems. The confirmation of positive results will be the same as
previously discussed for other Tier II inorganics. EPA is also
proposing that the completion period for the initial round of
compliance monitoring for asbestos be extended to five years,
because of the limited laboratory capabilities currently available
to conduct asbestos analysis. Asbestos monitoring requirements are
summarized in Exhibit 4-8.
C) Tier I
EPA is proposing minimum monitoring requirements for Tier I
lOCs (i.e., nitrate and nitrite) that are more stringent than the
proposed monitoring requirements for Tier II inorganics. Mon-
itoring requirements for nitrate/nitrite monitoring are summarized
in Exhibit 4-9.
Initially, the sampling frequency for community and non-
transient non-community systems will be quarterly for surface water
systems and annually for ground water systems. Both surface and
ground water systems must monitor quarterly whenever any previous
result exceeded 50 percent of the MCL. Quarterly monitoring may
be reduced to annual monitoring when results from four consecutive
quarters are less than 50 percent of the MCL.
Transient non-community water systems must monitor every 3
years for ground water systems and annually for surface water
systems.
Sampling should be conducted during periods of high
vulnerability (e.g., after rainfall or fertilizer application).
The sampling locations and the minimum number of samples would be
as previously described for Tier II, source-related inorganics.
If the result of any analysis exceeds the MCL for nitrate or
nitrite, a second sample must be collected within 24 hours and
analyzed within two weeks. If the average of the two samples
exceeds the MCL, or if the system fails to analyze a follow-up
4—21
-------
EXHIBIT 4—8
ASBESTOS
Proposed Monitoring Requirements
<50% of MCL
Repeat Monitoring
at State Discretion
4—22
-------
EXHIBIT 4-9
NITRATE I NITRITE
Proposed Monitoring Requirements
Community and Non-Transient Non-Community Systems :
Ground Water
Systems
Surface Water
Systems
Monitor
Quarterly
Monitor
Annually
Any result
? 50% of
MCL
All results
<50% of
MCL -
Transient Non-Community Systems :
Surface Water
Systems
Monrtor Annually
4 Consecutive
Quarters with
Results
<50% of MCL
No 4 Consecutive’
Quarters with
Results
<50% of MCL
Ground Water
Systems
Monitor Every
3 Years
4—23
-------
sample, the system will be deemed in non- ompliance and procedures
for public notification must be followed.
At state discretion, cornpositinq of up to five source or entry
points may be allowed. Like Tier II contaminants, if the
concentration in the composite sample indicates that one or more
of the individual samples may exceed the MCL, follow-up sampling
would be required at each sampling point included in the composite.
D) Unrequlated Contaminants
EPA is proposing monitoring requirements for six inorganic
chemicals (i.e., antimony, beryllium, cyanide, nickel, sulfate,
and thallium) for which MCLs have not yet been proposed. Monit-
oring requirements for these unregulated contaminants will apply
only to those systems the state finds vulnerable to contamination.
States will have the discretion to apply these monitoring re-
quirements based on local concerns and priorities. The proposed
regulation involves one round of sampling in which groundwater
systems will collect one sample at each entry point which is
located after any treatment to the distribution system. Surface
water systems may sample at points in the distribution system that
are representative of each source or at each entry point to the
distribution system which is located after any treatment.
Composite samples representative of up to five entry points for
groundwater systems or five sources for surface water systems are
allowed at state discretion.
Assumptions Used in the Monitoring Cost Models
A number of assumptions apply to monitoring costs for
nitrate/nitrite, barium, cadmium, chromium, mercury, and selenium,
as outlined below.
o Average number of entry points for groundwater systems:
System Size
( Population Served Entry Points
<500 2
500—10,000 4
> 10,000 6
o Surface water systems average one source per system.
o One sample is taken per source or entry point.
4—24
-------
o Models are broken down by 12 standardized system size cate-
gories (see Chapter 2), by community and non-transient non-
community systems (for nitrate and nitrite, all non-
community water systems are included), and by source.
o There will be no phase-in for the initial round of moni-
toring on the basis of system size.
o Systems would be required to take one confirmation sample
each time a source or entry point tested above the MCL.
o Systems without occurrence of Tier II contaminants above
the MCLs would be allowed to conduct repeat monitoring
every 10 years.
o systems with occurrence of Tier II contaminants above the
MCLs would be required to conduct repeat monitoring
annually for surface water and every three years for ground
water systems.
o At the most stringent MCL alternatives for the Tier II
contaminants, it was assumed that all systems with any
occurrence would be required to monitor annually for
surface water systems and every three years for ground
water systems.
o All systems with resuJ . s less than the practical
quantitation limit (PQL) would be allowed to submit
composite samples each time monitoring was conducted.
o Monitoring for Tier I contaminants was modeled based on a
simplifying assumption that the initial monitoring results
would determine frequencies in subsequent rounds of
monitoring. Thus, a ground water system with initial
results greater than 50 percent of the MCL would be assumed
to monitor quarterly in all subsequent rounds of
monitoring.
Because more current information on the average number of
entry points for ground water systems was obtained after the
monitoring calculations were completed, the assumption regarding
entry points was updated for nitrates and nitrites (which represent
the majority of monitoring costs) as presented below, but was not
updated for the Tier II contaminants. The new assumption is not
expected to have a significant effect on monitoring costs for the
‘I ier II contaminants; costs for these contaminants will, however,
be revised for the final rule as necessary.
12 PQLs for each Phase II contaminant were provided in Exhibit
2—10.
4—25
-------
o Average number of entry points for groundwater systerrts: 13
System Size
( Population Served) Entry Points
25—100 1.0
101—500 1.5
501—1,000 2.0
1,001—3,300 2.5
3,301—10K 3.0
10,001—25K 5.0
25,001—50K 6.0
50,001—75K 8.0
75,001—500K 10.0
500,001—1M 20.0
Over iN 50.0
High and low bound models were designed for each contaminant
f or the proposed and most stringent MCL’S. These models are
similar except for the following assumptions.
o For the low bound model, it was assumed that systems with
occurrence above the MCL would be positive at only one
source or entry point, and would exceed the MCL only once
during the sampling period.
o The high bound models assume that systems with occurrence
greater than the MCL will have occurrence at each source
or entry point each time monitoring is conducted.
o The Tier II high bound models assume 10 percent of all
systems would have inadequate historical monitoring data
and would therefore be required to conduct an initial round
of monitoring. The low bound model assumes five pe ent
of all systems would be subject to this requirement.
Analytical cost assumptions may also differ between high and
low bound mode-is, as described in Exhibit 4-10 below.
13 Based on survey results included in the Final Descriptive
Summary: 1986 Survey of Community Water Systems , US EPA, Office
of Drinking Water, October 23, 1987.
14 Persorial Communication, Association of State Drinking Water
Administrators, October, 1987.
4—26
-------
EXHIBIT 4-10
Analytical Cost Assumptions -- lOCs
IOC Low High
Asbestos $300 $500
Mercury, Selenium 20 25
Cadmium, Barium, Chromium 9 20
and Unregulated Contaminants
Nitrate/Nitrite 10 10
The analysis of asbestos monitoring costs differs from the
other lOCs as follows:
o Only vulnerable systems are required to monitor.
o Occurrence data by system size were not available.
o Costs are estimated for community water systems only.
o Only one MCL alternative exists for asbestos, therefore
the cost estimate under “most stringent” is the same as
“proposed”.
o Monitoring will be phased in based on system size over a
five year period.
o The difference between the high and low bound model is
based on the analytical cost only. -
o The average source/entry point per system is assumed to be
one.
The analysis of unregulated contaminants also differs from
the other lOCs as follows:
o Only vulnerable systems monitor.
o Monitoring is limited to one single sample.
o Repeat monitoring or confirmation monitoring is entirely
left to state discretion and is not estimated.
4—27
-------
o The number of entry points, sources and samples per source
are assumed to be the same as those of the Tier I and II
monitoring cost models.
The critical assumption needed for estimating costs for
unregulated contaminant monitoring is the number of systems that
would be found vulnerable and would therefore be required to
collect samples. Based on NIRS data for these contaminants, it
was assumed that the sum of occurrence probabilities for the six
lOCs considered in the analysis represented the low bound of the
number of systems that the states would classify as vulnerable.
This summation procedure yields an estimate of 10 percent of
systems. The high bound estimate is represented by assuming that
all contaminants occur as frequently as the contaminant with the
highest probability of occurrence in NIRS (i.e., nickel at
approximately five percent). This procedure yields an estimate of
30 percent of systems being found vulnerable.
Several factors regarding the estimate of unregulated con-
taminant monitoring costs should be noted. First, this methodology
entails considerable uncertainty. For example, when making
vulnerability determinations, how would states know which ten to
30 percent of systems have occurrence prior to an initial round of
comprehensive testing? While no discrete data sets exist for
estimating this component of the total national costs, EPA believes
that the ten to 30 percent range represents a realistic estimate
of the number of systems likely to be found vulnerable to
unregulated lOCs. Second, all systems assumed to be vulnerable
were assumed vulnerable for all six unregulated lOCs. This is
probably an over—estimate of impacts on an individual system basis
since there would be few cases where water systems would test
positive for all six contaminants. Finally, since unregulated
contaminant monitoring is treated as a one—time, “capital”
expenditure, the total annual costs for the rule are not affected
significantly regardless of whether high or low assumptions
regarding occurrence probabilities are accepted.
4.7 Costs to State Programs
An estimate of the total costs to state programs for the
entire Phase II regulatory package (i.e., including both lOCs and
synthetic organic chemicals) is included in the draft Regulatory
Impart Analysis for Synthetic Organic Chemicals.
4—28
-------
5. ASSESSMENT OF BENEFITS
5.1 “ Damages Avoided” Versus Total Benefits
Computation of avoided health damages has been the conventional
approach used in quantifying benefits of improved drinking water
regulations. In considering the “total benefits” of improved drinking
water regulations, it is appropriate to incorporate a broader range
of effects on social welfare than the quantifiable avoidance of adverse
health effects. There are two additional categories of benefits
that cannot be assessed in a quantitative fashion but must nonetheless
be recognized as benefits, even if only as “intangibles.” First,
there are benefits which result from the fact that drinking water
standards incorporate a margin of safety. Secondly, indirect benefits
appear likely to result from efficiency improvements in the water
industry that may be induced by the expense of SDWA compliance.
This chapter addresses the economic benefits of controlling
bC contamination in public drinking water supplies. Section 5.2
and 5.3 discuss the above identified categories of “intangible”
benefits. Section 5.4 presents estimates of quantifiable benefits.
5.2 Benefits of A Margin of Safety
The safety of public water supplies has historically been
“taken for granted,” implying the risk perceived by consumers is
negligibly small. Thus, there is an implied “warranty” inherent
in the nature of public water supply which provides additional
benefits beyond the expected value of avoided adverse health effects.
There are two types of additional benefits provided by the warranty:
benefits in consumption and benefits in production. Consumption
benefits are those whose immediate beneficiaries are individuals
in their capacities as consumers; production benefits are those whose
immediate ben ficiaries are economic units engaged in production
for a market.
Benefits In Consumption
The value of a reduction in risk to an indiv 2 idual is the amount
that person would be willing to pay to achieve it. Economic research
has not produced a definitive quantitative assessment of willingness
to pay for risk reduction in drinking water.
1 Arrow, K.J., “Criteria for Social Investment,” Water Resources
Research , Vol. 1, No.1, 1965.
2 Freeman, A.M., The Benefits of Environmental Improvement ,
Resources for the Future, Washington, D.C., 1979, P. 168.
5—1
-------
Analytical effort has often focused instead on the more
tractable problem of computing the value of health-related damages
that would be avoided. It is usually assumed that this represents
an amount of compensation sufficient to restore an individual whose
health has been damaged to their original level of utility. Stated
otherwise, the underlying presumption is that, faced with no
alternative choices, people would be willing to pay up to the amount
of their expected losses in order to avoid those losses. Hence,
there is an assumed equivalence between “damages avoided” and
“willingness to pay.”
This equivalence is incomplete, however, when the valuation
of damages is limited to the most readily quantifiable items such
as the value of medical costs and lost income. This limited concept
of damages ignores the fact that people would not be fully compensated
(i.e., not truly indifferent) unless there is also some compensation
for the pain, suffering, and inconvenience associated with illness.
The presence of these additional damages means there must be a
corresponding amount of additional willingness to pay. The additional
willingness to pay represents a demand for an additional margin of
safety to offset the individual’s fears and anxieties regarding the
undesirable consequences of the health risks. It reflects the benefit
in consumption provided by the “warranty” --the extra margin of safety
-- in public water supplies.
Benefits In Production
The benefit in production derived from having a margin of
safety —— a “warranty” -— in public water supplies consists of the
addition to total economic output made possible by the fact that
public confidence in the integrity of the water supply facilitates
the production and exchange of goods.
Without the implied “warranty” in infrastructure systems, the
participation of more risk averse individuals in both production
and consumption might be limited, constrained, or encumbered by
compensating expenditures. Constraints and distortions of this
type can result in a lower level of economic output than would
otherwise be achievable. The presence of a substantial margin of
safety in other categories of infrastructure suggests that the
welfare gained by removing such risks as impediments to economic
activity significantly outweighs the additional cost.
Our expectations of public drinking water supplies are no
different from what we expect of other categories of physical
infrastructure such as roads, bridges, and buildings. In the design
of bridges and buildings, engineers and architects follow a time-
honored practice of multiplying their structural calculations by
a “safety factor” of two or more. These safety factors are specified
in highway construction standards and building codes. Despite the
5—2
-------
fact we have always operated on the presumption that water supplies
are safe beyond question, there was no uniform national mechanism
for providing a guarantee of the safety of drinking water until the
passage of the Safe Drinking Water Act (SDWA) in 1974. The 1986
amendments to the SDWA are intended to complete the process.
5.3 Induced Efficiency Improvements In The Water Industry
Water has historically been very abundant, pure (or presumed
pure), and inexpensive to produce. In fact, Adam Smith was quite
perplexed that something of such obvious value should have such a
low price and be taken for granted almost as though it were valueless.
This state of relative abundance prevailed from the time of Adam
Smith into the early part of the Twentieth Century and our institutions
and attitudes have been shaped by it. Even in the arid west, we
made water abundant through large federally subsidized water projects.
In the absence of compelling cost pressures, there was no need
to “economize.” Various non—economic forces have filled the void
and evolved as the controlling influences in how we produce, price,
and consume water. As a result, many water systems are plagued by
a legacy of inadequate maintenance and replacement, and inefficient
pricing and capacity management (see expanded discussion of these
topics in Chapter 3).
Cumulatively, the regulations forthcoming under the 1986
amendments to the Safe Drinking Water Act will have the effect of
increasing the per gallon cost of treated water by a sufficient
amount that water may no longer be regarded as being inexpensive
to produce in many places. The recognition of this new cost
environment may induce a variety of efficient improvements in both
demand management and supply management in many water systems.
The SDWA regulations will increase the incentive for water
systems to do a better job of optimizing their maintenance and
replacement efforts to fix leaks in distribution systems. The SDWA
will provide ihcentive for water systems to adopt more efficient
rate structures that encourage more efficient use of capacity. The
result of both of these types of improvements is to reduce the
treatment capacity requirement.
Indirect benefits may also be reaped from improved pricing
practices in the area of drought management. The economic costs
of shortages are made worse by the inefficiencies of present pricing
practices. Environmental costs of inefficient levels of peak period
water demands on lakes, groundwater resources, and estuaries are
becoming increasingly severe across the country. To the extent that
more sophisticated approaches to demand and supply management are
encouraged, both economic and environmental benefits will result.
Even without rate reforms, higher water rates needed to pay for SDWA
improvements will marginally reduce water demands and their resultant
5—3
-------
pressures on the environment. As a final note, the higher water
rates needed to pay for SDWA improvements will also induce marginal
reductions in waste flows.
The benefits derived from induced efficiency improvements may
seem, at first blush, to be just a collection of miscellaneous
“intangibles.” Lists of miscellaneous items are often of lesser
importance. For this reason, the potential importance of the above
list must be underscored.
This raises three questions that must be addressed. First,
how certain is it that the cost pressures of the SDWA program will
be sufficient to overcome the inertia of well-entrenched institutional
practices and induce the types of efficiency improvements envis-
ioned? The answer to this depends on the answer to a second question:
how significant will the penalty for inefficiency become in complying
with SDWA standards? Answering from the bias of pricing disciplines
followed in most other public utilities, it seems likely that when
the principles of proper pricing have been so thoroughly ignored,
the potential efficiency gains of reform would be large. If the
potential gains are large, the penalty for continued inefficiency
would be large also. But, this leads to the third question: will
the penalty for inefficient practices be recognized, or will it be
unknowingly absorbed by water system customers as it has in the past?
As the SDWA program is implemented in water systems across the
country, the answer to this last question will affect the outcome
of every resulting rate increase proposal. Where inefficiencies
are allowed to persist, and consumers are unaware of their existence,
consumers will not only pay more than they should to achieve SDWA
compliance (by the amount of the penalty imposed by the remaining
inefficiencies), but will also be given the impression that the entire
amount of the rate increase is attributable to SDWA compliance.
Disguised in this manner, the inefficiencies will diminish the total
reserve of consumer willingness to pay for additional safety in
drinking water. This raid on the consumers surplus may ultimately
hinder implementation of the SDWA program and result in a lower level
of safety than is truly desired.
The inefficiencies that have evolved in water system management
and water pricing must be recognized if the benefits of safety
improvements are to be fairly appreciated. Consumers must be made
aware that only a portion of SDWA-induced rate increases relates
to safety improvement. The other portion is either a premium required
to reverse past inefficiencies or a penalty required to perpetuate
them.
5.4 Agc regate Analysis of Health Benefits
The calculation of aggregate national level benefits is
addressed through assessment of avoidance of direct damages in the
5—4
-------
form of adverse health effects due to a reduction in exposure. As
discussed in Chapter 2, exposure to the lOCs evaluated in this document
results in various sublethal health effects. Since the EPA has yet
to finalize dose/response functions for these sublethal effects,
an accurate estimate of the actual number of cases of disease avoided
was not possible. Instead, the number of persons that would derive
some reduction in exposure to each contaminant as a result of the
proposed rule was estimated.
In addition to calculation of the reduction in population
exposed, an analysis of the potential baseline cancer cases occurring
due to asbestos exposure in drinking water was conducted. It should
be noted however, that because it is assumed that costs of controlling
asbestos are subsumed by other regulations, no benefits are counted
for this contaminant in this analysis. Instead, benefits, in the
form of a reduction in population exposed, will be quantified in
subsequent drafts of the RIAs for the Surface Water Treatment Rule,
and the Lead/Corrosion Control Rule.
In the process of developing regulatory impact analyses, an
assessment of the aggregate benefits is required. Due to the
structure of the water supply industry (as outlined in Exhibit
5—1), gross comparisons of national aggregate impacts tend to produce
a deceiving picture of the true relationship between costs and benefits
at the individual water system level. Aggregate comparisons average
together an excess of positive net benefits in large systems with
an excess of negative net benefits in small and very small systems.
For this reason, aggregate analyses should be viewed only as roughly
illustrative of the magnitude of the problem at the national level
and caution should be exercised in interpreting these results.
A) Sublethal Health Effects Reduction in Population Exposed
As suggested above, benefits of removing contamination from
drinking water are expressed in terms of cases of disease avoided.
To facilitate such calculations, the Agency develops dose/ response
relationships or lifetime risk estimates for sub—lethal health effects
at chronic, low level exposure whenever possible. However, for
the lOCs considered in this analysis, such data were not available.
Since all lOCs except asbestos are being regulated on the basis of
sublethal health effects, and no usable dose/response data were avai-
lable for these contaminants, it was not possible to calculate the
number of cases of adverse health effects avoided. For this reason,
sublethal health effects benefits associated with the regulation
of lOCs in drinking water are presented only in terms of a reduction
in the population exposed under the various MCL alternatives. The
3 See Chapter VIII, “Quantification of Toxicological Effects,”
in Drinking Water Criteria Document for Inorganic Chemicals (separate
document for each contaminant).
5—5
-------
11
EXHIBIT 5—1
I)i t ii hnt iou
of ( oiiiiiiiiiiit y \Va I r Kv - t ( II1H
2.4
23.9
1 0.1
(;.9
I
:L3K— 1 OK
: i .1
4.fl
10K— I 0 0K
,
1001 <
Popiti a 11011 i Z( ( a I ego i ie
Total f y t( I11
L1 i TI 1 P 1 ). ( FV( d
(33.9
70
60
50
In
c
1 0
30
20
10
0
25— So o
501 —3.3h.
-------
results of these calculations are presented in Exhibits 5-2 and 5-
3 and in Appendix VIII.
The net incremental reduction in exposure derived through
adoption of the most stringent MCL alternatives would be approx-
imately k° 2 million persons in systems serving fewer than 50,000
persons. Excluding the systems estimated to have occurrence of
mercury (see Chapter 2 discussion of mercury occurrence estimates),
a net reduction of approximately 99 million persons exposed is esti-
mated.
Benefits of removing cadmium were evaluated at the proposed
MCL alternatives. For all other lOCs evaluated, all MCL alterna-
tives except the most stringent are either equal to or greater than
the existing standard and are therefore assumed to have no net im-
pact. The estimated net incremental reduction in exposure derived
through adoption of the preferred NCL alternative for cadmium is
approximately 170,000 persons in systems serving less than 50,000
persons.
B) Baseline Cancer Cases Due to Asbestos Exposure
As discussed in Chapter 2, asbestos is being regulated by the
Agency as a carcinogen via ingestion exposure. Because it is
assumed that asbestos treatment costs will be subsumed by other
drinking water regulations, it is most appropriate to attribute
any benefits accrued due to asbestos removal to those regulations.
To illustrate the probable magnitude of health benefits of controlling
asbestos however, an analysis was conducted to estimate the total
baseline cancer cases currently attributable to asbestos exposure
in drinking water. The results of that analysis are presented below.
Estimates of baseline annual cancer cases attributable to
asbestos exposure in drinking water were also calculated. Based
on the assumption that a concentration of 7.1 million fibers (gre-
ater than 10 urn in length) peg liter (MFL) is equal to a io6
individual lifetime cancer risk, the following conversion was made
to derive annual individual risk per unit concentration.
i x io cases/person/lifetime / 70 years/lifetime
7.1 MFL
2.01 x 10 cases/person/MFL/year
4 At concentrations equivalent to the most stringent MCLs, a
significant degree of co-occurrence may be expected.
50 Federal Re ister No. 219, p. 46963.
5—7
-------
EXHIBIT 5—2
REIXrPION 111 P’F’tILATJ(*I ED(F SE1) TO INORGANIC CHEMICALS AT MWT STRIR;ENT MCI ALTERNATIVE (IN THOUSANIS)
SYSTE2I SIZE CATIfl4W ’
(RIPIIIATI(t4 SFRVED)
MCL : ALL S(IALIY
:(W/L) 1 2 3 4 5 6 7 8 9 10 11 12 SYSTDISSYSTO’IS
25-100 101 -5( 10 5( 11-1K 1K—3 3K 3 3K— bK IOK-25K 25K—SOK SfIK-75K 75k-WOK lOOK-BOOK 500K -3M iN ’
ESTIMATED NUtIDF.R (IF PERSONS EXF(SEI) A&JVE MCGT STRI ICEUI’ MCI. ALTERNATIVE (IN ‘DIOUSAUDS) 3
‘R irium : ‘. ooo : I
10: 5
thus : s o : 1
Mercury : 2 : 8
Nltiat’/Nitrjte :io,ooo ir .
se letiun : 1( 1 : 4
Total
Total less Metctir-y
1 1 3
14 14 30
2 5 14
44 90 297
149 158 182
16 25 75
226 293 601
182 2( 13 304
4 4 3
39 38 28
24 30 37
688 847 1,061
192 3( 17 247
159 191 229
1,106 1,417 1,605
418 570 544
1,726
32
2,294
2,573
16. 046
4,933
27,604
25,031
1,959 : 17,541 8,622
0 892 682
3,016 16,181 4.891
3.420 : 18.467 5.763
19.108 : 158,599 76.753
6,611 : 34,580 10.288
34,114 246,26( 1 106.999
‘30,754 227.793 1( 11.236
(XI11TA?1I tiAlif
TOTALS
:oaritim : 100 : 160 507 566 1,295 1.993 2,114 1,987 1.156 812 3,266
0 1 18 56 58 122 159 155 114 58 18 102
chrc . iu u s 10 : 30 122 171 525 1.075 1,320 1,648 1,064 1,043 3,813
mer cury : 0 2 : 29 121 201 608 1,310 1,580 1,914 1,225 1,173 4,313
N itrate/N itr ite : 100 : 3,147 6.403 7,006 9,129 17,230 17,261 16,577 9.850 7.483 29,359
:Se lenium : 5 : 33 164 317 1,023 2,334 2,860 3 ,557 2.295 2.244 8,2 ( 19
Total 3,417 7,373 8,319 12,702 24,101 25,290 25,797 15,648 12,773 49,062
Total less Mei-cur-y : 3,388 7,252 8,118 12,094 22,791 23,710 23,883 14,423 11,600 44,749
ESTIMATED NI QIBER OF PERSONS EXF1 ED AI3DVE INTERIM STANDARDS (iN THOUSANDS)
94
86
14
24
686
136
146
1 .007
321
I t S
Cachulu urn
du n ‘liii
:I4er’’u y
Nit r ate/fl it t j I c .
Se let ii UI
T o In]
:Tot a1 t.’ s (ler’uiy
(I 3
5 25
24 87
674 2.463
74 324
139 5 1 1
916 3.413
242 950
159
13
29
21
3,072
2 )
3,323
3,30”
ESTIMATED 1111’ RELUCFION IN TI-fE NIJI-IBER OF PERSONS EXF GE1) AT M(6T STI ?IIflFNT MCI ALTERNATIVE (IN ‘IlI(*JSANLS)
1 0
8
52 71
1,483 1,991
151 136
3(14 402
1,999 2.600
516 609
22
220
371
1 1 )332
2,131
2, 2 (11
15,277
4,945
506 565 1,292
42 44 92
120 166 511
77 111 311
6.254 6.848 8,947
148 292 948
1.147 8,026 12,101
7,070 7,915 11,790
1,989
120
1,051
622
17,038
2.175
22.995
22,373
2.110 1,984
117 86
1,290 1,611
733 853
16,954 16,330
2.669 3.328
23,873 24.192
23,141 1 23.339
I . 155
44
1 ,(4(I
539
9,714
2.149
14.641
14, 102
812
13
1,019
499
7.4( 19
2, m c
11.857
11.150
17
168
113
3,035
1.310
699
5,342
2,307
8 .605
514
4,778
2 .728
75 .443
9 .589
101,657
90.929
I u nits cc-4nimu1 Ly triter systtru. iii ion-transient, non-conmisity waLer cy t.cm’, for all cc’rtami,anL5
E’Vct pL rIt rnt’/ni t i -i t o For I I . ‘ , nrutaniraxrt.w a El pub ] Ic water systems are in’ I i f r i
2 Sn-uI 1 .y’it’ m arc tti’-’s C’ rv I us “II, (10( 1 ;ei soot, or fewer
3 A l II’’ ne , l ‘-ti urigeuui 1*1. all-rI iv .- ., thore urO y Ix) sigulfic—nit co—’e’a ic’nn of l i 5
3,263
77
3,726
1,85(1
29,( ’35
7,690
45,649
‘13,799
I .725
24
2,242
I ,(190
15,895
4.629
25,605
:4,515
1,959 : 17,519
a: 672
3.005 : 15.810
1,429 : 8.135
18.972 : 456,468
6,209 : 32,379
31,574 : 230,983
30.145 : 222,848
-------
EXIIIBrT 5—3
RFJACTJC*4 IN FOFIJLATICtI EXF(SED TO !t&WGAHIC CIIE?IICALS AT PREFERRED t EL ALTERNATIVE (IN Th($JSANDS)
I I
SXSTE]1 SIZE CATWORY TOTALS
(POF IJLATIC 4I SERVED) 2
ONTANItJANT HCt : ALL StALL
( ( lU/L) 1 2 3 4 5 6 7 8 9 10 11 12 SYSTE IIS SYSTE IIS
25-1(0 101—5(0 501-1K 1K—3 3K 3 3K— bK 10K-25K 25K—50K 50K-75K 75k-lOOK 100K-bOOK SOOK—IM 111-.
ESTIMATED MEMBER OF PEESCES EXREED ABOVE PRCflEEI) IICL ALTERNATIVES (IN ThC*JSANL$)
:cadmiiun 5 9 28 29 61 80 76 57 29 9 51 16 0 445 340:
• — I I I
ESTIMATED NUMBER OF PERS(*4S FIRGEI) ABOVE INTERIM SFANDAALG (IN T HCIOSANIX)
:Cadm lum : 1( 1; 5 14 14 30 39 38 28 14 5 25 8 0 220 168
I I I I
ESTIMATED NET Rk]XKJTIC*l IN THE NUMBER OF PERSC*45 EXF(SE]) AT PROFUSE!) MCL ALTERNATIVES (IN 1 1IC AISANIS)
lID
:Uadm lum 4 14 15 31 41 38 29 15 4 26 8 0 225 172
• I I — —I
1 Includes cccriunity water systems and non-transient, non-coon*nlty water systems
2 Sanl 1 systems are those servIng 50 .0(10 ( rsons or fewer
-------
The EPA has prepared draft estimates of the populati 6 on exposed
to various levels of asbestos in community water systems and these
served as the basis for estimates of baseline cancer cases. Exhibit
5-4 summarizes estimates of the total population exposed to various
concentrations of asbestos in CWS.
EXHIBIT 5-4
Estimated Population Exposed (Millions) to Various Concentrations
of Asbestos in Community Drinking Water Systems
> 7.1 MFL > 0.71 MFL > 0.071 MFL 0 — 0.071 MFL
LOW 0.73 7.1 16.0 199.0
HI 1.2 7.5 20.0 203.0
No data were available regarding average asbestos concentra-
tions in affected systems. Therefore, to facilitate calculation
of baseline cases, it was necessary to make assumptions regarding
the concentrations to which the various segments of the population
are exposed. For the population exposed above 7.1 MFL), estimates
were ca1cu1 ted using average concentrations ranging between 7.5
and 70 MFL. For the segment of the population exposed to levels
below 7.1 MFL but above 0.071 MFL, a low scenario assuming average
influent concentrations equal to the geometric mean of the range
(e.g., for the population exposed between 0.071 MFL and 0.71 MFL
the average concentration was assumed to be 0.22 MFL). A corresponding
high scenario assumed maximum average concentrations within each
range (e.g., for the population exposed between 0.071 MFL and 0.71
MFL the average was assumed to be 0.70 MFL).
Finally, for the remainder of the population (i.e., those
exposed to less than 0.071 MFL), an average concentration of 0.355
MFL was assumed for the low exposure scenario (i.e., one-half the
6 Science Applications International Corporation, 1986, Estimated
National Occurrence and Exposure to Asbestos in Public Drinking Water
Supplies . December 23 draft. It should be noted that occurrence
analyses for asbestos were conducted under the assumption that the
proposed MCL would be 7.1 MFL. As such, estimates of systems affected
and populations exposed may represent a slight underestimate from
that expected at 7 MFL.
7 me occurrence data indicate that there is no exposure above
71 MFL.
5—10
-------
minimum reporting level), and a concentration of 0.07 MFL was assumed
for the high scenario.
Under all exposure scenarios evaluated, the annual baseline
cases of cancer were estimat ed to be 0.3 or less, using the annual
risk factor described above. Due to the small baseline estimates,
and the error inherent in those estimates, it was not considered
meaningful to calculate cases of cancer avoided and consequently
no further analysis was performed for asbestos. The full range of
baseline cancer cases under all exposure scenarios is presented in
Exhibit 5—5 below.
8 1t should be noted that it was not possible to discern the
proportion of exposure attributable to large water systems from
these data. However, because the vast majority of asbestos occurrence
is though to result from corrosion of asbestos/cement (A/C) pipe
(see Chapter 2), and because many large water systems have exten-
sive amounts of A/C pipe in their distribution systems, it may be
inappropriate to “assume away” the exposure in large systems as was
done for arsenic.
5—11
-------
EXHIBIT 5—5
Estimated Baseline Cancer Cases Due to Asbestos
Contamination in Community Drinking Water Systems
Assumed Avg.
Concentration Estimated Assumed Average Concentration for
for population Number of Populations Exposed Above 7.1 MFL
Exposed B low Persons
7.1 MFL 1 ’ Exposed 3 7.5 MFL 15 MFL 30 MFL 70 MFL
LOW LOW 0.06 0.07 0.09 0.15
HI 0.07 0.08 0.12 0.22
HI LOW 0.14 0.15 0.17 0.23
HI 0.15 0.17 0.21 0.30
1 NFL - Million intermediate length (> 10 urn) fibers per length.
2 Assumptions regarding average concentrations for segments of the
population exposed below 7.1 MFL range from 0.0355 NFL to 2.25 MFL
under the low scenario, and from 0.07 MFL to 7.00 MFL under the high
scenario.
The number of persons exposed at various concentration levels could
not be estimated precisely in EPA’s occurrence analysis. As a result,
the estimates are presented as a range. It is estimated that between
16 million and 20 million persons are exposed to asbestos at
concentrations exceeding 0.071 MFL, and of these, between 0.73 and
1.2 million are exposed above 7.1 MFL (see Exhibit 5—4).
5—12
-------
6. REGULATORY FLEXIBILITY ANALYSIS AND
PAPERWORK REDUCTION ANALYSIS
6.1 Requlatory Flexibility Analysis
The Regulatory Flexibility Act (RFA), enacted on September
19, 1980, requires all executive agencies to explicitly consider
small entities in their regulatory design and implementation pro-
cess. The purpose of the RFA is to encourage regulatory agencies
to minimize any disproportionate burden that falls on small enti-
ties. The specific objectives of the RFA are:
o To increase agenciest awareness of their regulatory impact
on small entities;
o To compel agencies to explicitly analyze, explain, and
publish regulatory impacts on small entities; and
o To encourage agencies to provide regulatory relief to small
entities while still accomplishing their statutory
mandates.
These objectives are accomplished through the reqi.lireinents of
regulatory flexibility analyses for proposed regulations. If a
regulation does not have a “significant” impact on a “substantial”
number of small entities, then the regulatory flexibility analysis
will consist of a certification to that effect.
Prior to conducting a regulatory flexibility an lysis, a
regulatory agency such as EPA must define a small entity. The RFA
defines small entities a including small businesses, organ-
izations, and governments.’ A small business is defined as “any
business which is independently owned and operated and not dominant
in its field.” 3 A small organization is defined as “any non—profit
enterprise which is independently owned and operated and is not
dominant in its field.” Finally, small government entities are
defined as “those city, county, town, township, village, school
district, or special district governments serving a population of
less than 50,000 persons.”
1 1t should be noted that under the SDWA EPA’s Office of
Drinking Water employs a different definition of a “small entity”
for small water systems. The analyses presented in this section
are prepared only for compliance with the RFA.
2 Regulatory Flexibility Act of 1980, PL 96—354, Section
601(6)
USC, Section 632.
4 op. cit. , note # 2 supra , Sections 601(4) and 601(5).
6—1
-------
community water systems can be divided into three ownership
categories for the purpose of RFA analysis: 1) publicly owned; 2)
investor owned; and, 3) ancillary systems. Publicly owned systems
are those owned by governmental entities; investor owned systems
are privately owned; and ancillary systems are those small systems
that are ancillary to other enterprises such as mobile home parks
or hospitals. According to EPA, there are 26,424 publicly owned
community water systems. 5 Of this total, 98 percent serve fewer
than 50,000 persons.
Investor owned water systems are firms prima ily engaged in
production and distribution of water to consumers. According to
regulations published in accordance with the RFA, these companies
are considered to be sma .1 businesses if their annual receipts are
less than $3.5 million. To apply this standard to public water
systems, the Consumer Price Index for other utilities and public
services was used to normalize operating costs from 1984 to 1986
dollars. This procedure suggests that the upper limit for a small
water utility would be $3.8 million per year in 1986 dollars. The
EPA estimates that systems serving populations of 50,000 persons
generate revenues of approximately $3.8 million. 8 Revenues for
investor owned water systems serving 25,000 to 50,000 persons
averaged $3.5 million in 1986. For investor owned systems serving
50,000 to 75,000 persons, revenues in 1986 averaged approximately
$5.2 million. Therefore, public systems serving 50,000 persons or
fewer will be considered “small entities” for purposes of this
regulatory flexibility analysis.
There is some question, however, as to whether all investor
owned water utilities serving fewer than 50,000 persons qualify as
small businesses. Many of these utilities are not individually
owned, but are owned and controlled by large holding companies such
as American Water Works Service Co. and General Water Works. In
addition, every investor owned utility operates in a franchised
area and thus constitutes a monopoly. This raises the question of
whether domination in a limited geographic area is the same as
dominance in a field of enterprise. The Small Business
Administration considers dominance to mean on a national basis:
therefore, no individual water utility can be dominant in the
marketplace.
Most ancillary community water systems serve fewer than 3,300
persons (10 ancillary systems serve populations between 3,300 and
25,000). These could be considered small entities; however, the
main activity of the enterprise may be sufficiently large to
5 u.s. Environmental Protection Agency - Office of Drinking
Water. Survey of Community Water Systems . October, 1987.
4941.
749 Federal Register , No. 28, p. 5035, 1984.
8 o .cit., note # 5 supra , p. 42.
6—2
-------
disqualify some organizations as small entities. Howeve , at least
73 percent of ancillary systems are mobile home parks; it seems
unlikely that other revenue sources at these facilities would be
sufficient to offset the burden of regulations significantly.
Nevertheless, it is not possible to determine precisely how many
ancillary systems constitute small entities due to the lack of
data.
Purpose of the Requlation
The purpose of this regulation is to reduce public health
risks by limiting human exposure to inorganic chemical contaminants
in drinking water, and to comply with the provisions of the SDWA
Amendments. The SDWA authorizes EPA to set maximum contaminant
levels (MCLs) for those contaminants in drinking water having any
adverse effect on the health of persons. If the Administrator
finds that it is not technologically or economically feasible to
regulate contamination to a specific level, a treatment technique
requirement may be promulgated in lieu of establishing an MCL.
Moreover, the SDWA Amendments mandate that “. . .the Administrator
shall publish maximum contaminant level goals and national imary
drinking water regulations [ for 83 specific contaminantsJ.” The
inorganic chemicals evaluated in this document are included in that
list of 83 contaminants.
Number of Small Entities Affected
Because of the health risks associated with inorganic contain-
inants, the nature of their occurrence in public water supplies,
and the generally limited treatment currently in place, the revised
standards for these contaminants are likely to affect small water
systems.
EPA guidance on compliance with the RFA indicate that, in
general, a “substantial” number of small entities is more than 20
percent of the total. Of the estimated 199,390 public water
supplies serving fewer than 50,000 persons, 333 (0.2 percent) will
incur treatment and waste disposal costs under the revised bC
regulations (assuming an MCL equivalent to the MCLG). Therefore,
by the 20 percent rule, the proposed ICC regulations would not
affect a “substantial” number of small water utilities at the
preferred MCL alternatives. The incremental number of systems
affected for each contaminant is shown in Exhibit 6-1.
9 op_cit., note #5 supra , p. 9.
10 Safe Drinking Water Act Amendments of 1986, P.L. 99—339,
Section 1412 (b) (1)
6—3
-------
EXHIBIT 6—1
Estimated Number of Small Entities Affected by Proposed bC Rule
# of Small # of Small Net Number
Systems Systems of Small Percent
Affected at Affected at Systems of Small
Preferred MCL Interim Std. Affected Systems
Barium 0 19 0 0
Cadmium 380 188 192 0.1
Chromium 0 20 0 0
Mercury 0 0 0 0
Nitrate 917 917 0 0
Nitrite 0 0 0 0
Selenium 0 185 0 0
TOTAL 192 0.1
Economic Impacts on Small Entities
Under the RFA, annual costs of compliance are to be compared
to the existing cost of production. This comparison may not be
applicable for water supply since it is not clear what comprises
the cost of production in this industry. The cost of production
is reflected in the price of the product in most industries, but
as discussed in Chapter 3, pricing practices in the water industry
are highly variable and not necessarily reflective of the full cost
of production. Therefore, data from the 1986 Survey of Community
Water Systems on water utility operating expenses were used as a
proxy for cost of production. According to the 1986 survey, the
total annual operating expense for small systems was approximately
$14.7 billion.
The total annualized treatment and waste disposal costs for
removal of lOCs in affected systems serving fewer than 50,000
persons (using the market discount rate of 10 percent) are
approximately $14.6 million. As detailed in Chapter 4, the total
uncertainty associated with this estimate is between 20 and 28
percent, depending on the assumptions used regarding the correla-
tion of standard errors. Using 28 percent as an upper bound,
annualized incremental treatment and waste disposal costs for lOCs
may range from $6.8 to $22.4 million, with 95 percent confidence.
To examine the incremental impacts on small systems two
alternative methods were used. The first method, a macro-level
analysis, calculates aggregate incremental costs as a percent of
aggregate cost of production (operating expense) for all small
systems including both those systems affected and those not
6—4
-------
affected by the rule. Exhibit 6-2 shows the total cost impact to
range from 0.05 to 0.15 percent using the macro-level analysis.
EXHIBIT 6—2
Production Cost Increase
(Macro—Level Analysis)
Total Cost Increase Percent Increase
( S Millions) in Production Cost
Lower Bound 6.8 0.05
Best Estimate 14.6 0.10
Upper Bound 22.4 0.15
The second method for calculating incremental cost impacts,
a micro-level analysis, examines the impact on those systems within
each size category that will incur remediation costs under the
rule. The 1986 survey reports an average operating expense for
private systems serving fewer than 50,000 persons of $2.27 per one
thousand gallons. Using data from the same survey on the average
daily flows produced by systems of various sizes, the annual cost
figures cited above can be converted to a cents-per-thousand
gallons basis. This reflects the incremental additional cost of
production for individual systems. Exhibit 6-3 compares these
amounts to the present level of production costs.
Agency guidance regarding the RFA defines a percentage
increase in production cost of five percent or more as a signif i-
cant impact. The macro-level analysis indicates that the Inorgan-
ic Chemical Regulation would not exceed this threshold; however,
when examined at the micro-level it is very likely that the ICC
Rule exceeds this threshold according to the data shown in Exhibit
6—3.
RFA guid 1ines also call for an analysis of whether the com-
pliance costs as a percentage of sales for small entities are 10
percent or more higher than compliance costs as a percentage of
sales for large entities. There are two inherent features of the
water industry that would make such an analysis misleading.
In community water systems, the economies of scale are such
that virtually any SDWA compliance action that is undertaken in a
small entity is bound to cost more than 10 percent as much as it
would cost in a large water system. Thus, the analysis would not
be very informative.
With respect to non—community water systems, water supply is
often an ancillary activity undertaken in the course of providing
some other primary business service. For example, non-community
6—5
-------
water systems may be ski lodges, roadside restaurants, campgrounds,
and the like. The data on non-community water systems is not
adequate to perform an assessment of how many of them are engaged
in each of these different business activities. Thus, it is impos-
sible to analyze the impact in terms of the percentage increase in
costs as a percentage of sales.
E fIBIT 6-3
Production Cost Increase by System Size Category
(Micro-Level Analysis)
System Incremental Percentage
Size Production Cost Increase in
Category ( Cents/l000 Gal.) Production Cost
25 — 100 571 206
101 — 500 286 111
501 — 1,000 151 92
1,001 — 3,300 97 59
3,301 — 10K 78 56
10,001 — 25K 61 44
25,001 — 50K 52 63
All Size Categories:
High Bound 586 224
Best Estimate 381 146
Low Bound 176 66
RFA guidelines further require an analysis of the extent to
which capital costs of compliance represent a significant portion
of capital available to small entities and whether the requirements
of the regulation are likely to result in closures of small
entities. While it seems certain that some small entities will
encounter financial problems in achieving compliance, the data with
which to estimate the extent of such problems does not exist. The
SDWA has explicit exemption procedures, however, specifically
designed to accommodate the financial problems of small water
systems.
The SDWA incorporates mechanisms for states to mitigate
economic impacts on small systems. At least for some systems in
the smallest size categories (i.e., less than 150 service connec-
tions), the most important mechanism may be the authority to exempt
systems from regulatory requirements on economic grounds while
alternative means of compliance are sought.
6—6
-------
6.2 Paperwork Reduction Analysis
Paperwork Reduction Act
Among the purposes of the Paperwork Reduction Act (PRA) 11 are
as follows:
o minimization of the federal paperwork burden for indivi-
duals, small businesses, state and local governments, and
other persons; and,
o minimization of the costs to the federal government of
collecting, maintaining, using, and disseminating infor-
mation.
Water utilities and state water supply agencies will be required
to maintain records on monitoring for lOCs and report results to
the EPA; this is likely to be the largest component of paperwork
associated with establishment of Federal bC regulations. The
Paperwork Reduction Act is intended to minimize the burden imposed
on utilities and states as they strive to protect the public health
by implementing the provisions of the SDWA.
Requirements of the Paperwork Reduction Act
EPA is required to submit to the Office of Management and
Budget COMB) proposed information collection requests. EPA also
must submit a copy of proposed rules containing an information
collection requirement. These proposed rules must be submitted no
later than publication of a notice of proposed rulemaking in the
Federal Register. When a final rule is published in the Federal
Register, EPA must explain how any information collection
requirements have been designed to be responsive to public com-
ments. 0MB determines the necessity, practicality, and utility of
the information being requested, and if approval of the request is
made, 0MB will issue a control order.
Under the Safe Drinking Water Act, EPA is authorized to
regulate contaminants in drinking water to protect the public
health. Inorganic contaminants are known to constitute a health
risk. To determine whether a specific water system exceeds an MCL
for bOCs, EPA must require water systems to collect arid analyze
samples and report results to the relevant primacy agent (i.e.,
either the applicable EPA regional office or the states). In the
case of inorganic contaminants, EPA, the states, water utilities,
and the public would use monitoring information for two purposes:
1) to determine the presence of contaminants which may affect human
health; and 2) to determine the reliability of a system to provide
waters free of inorganic contaminants. This monitoring data would
also allow appropriate action plans and removal decisions to be
made by affected utilities.
11 Public Law 96—511; 94 STAT 2812
6—7
-------
Number of Systems Affected/Respondent Burden
A detailed discussion of the nuin.ber of water systems affected
by monitoring and paperwork requirements associated with the
proposed rules is provided in the Information Collection Request
Documents.
6—8
-------
7. SUMMARY OF COSTS, BENEFITS, AND UNCERTAINTY
This chapter summarizes the aggregate impacts estimated to
result from the proposed revisions to drinking water regulations
for lOCs. Due to the structure of the water supply industry (as
outlined in Exhibit 5-1), such gross comparisons of aggregate
impacts produce a deceiving picture of the true relationship
between costs and benefits at the system level. Aggregate com-
parisons average together an excess of positivebenefjts in larger
systems with an excess of negative benefits in small and very small
systems. For this reason, aggregate analyses should be viewed only
as roughly illustrative of the magnitude of the problem at the
national level.
7.1 Total Incremental National Costs
Total incremental national costs of proposed IOC regulations
are summarized in Exhibits 7-1 and 7-2 at the most stringent MCL
alternatives and the proposed MCL alternatives, respectively.
Total national costs include treatment, waste disposal, monitoring,
and impacts on state regulatory programs associated with
implementation.
Exhibit 7—1 shows that regulating all lOCs at the most
stringent MCL alternatives may affect as many as 115,800 water
systems, excluding projected mercury occurrence. Total incremental
capital costs of this regulatory alternative would be approximately
$41 billion. Annualized costs for treatment and waste disposal,
which would include both amortized capital and annual expenditures
for O&M, would be between $7.3 and $8.4 billion, depending on the
discount rate applied. Incremental monitoring costs would be an
additional $4 million per year.
At the most stringent MCL alternatives, incremental impacts
would result for all contaminants because the standards would be
lowered from their current levels and many additional water systems
would be affected. However, these estimates may represent an over-
estimation of impacts due to co-occurrence of lOCs at such low
concentrations, and the simultaneous removal of multiple con-
taminants which would generally occur during treatment.
Exhibit 7-2 shows incremental impacts of regulations at the
preferred MCL alternatives. For barium, chromium, mercury, ni-
trate, and selenium, the revised standards would be equal to or
greater than the existing standards and therefore no incremental
impacts for treatment, waste disposal or exposure reduction would
be attributable to the rule for these substances. Also, it is
assumed that the costs associated with treatment to address as-
bestos contamination would be subsumed as systems complied with
the lead and surface water treatment rules. Finally, it is assumed
7—1
-------
EXHIBIT 7-1
Estimated Impact of Regulating lOCs at Most Stringent MCL Alternatives 1 ’ 2 ’ 3
Contaminant
I Benefits, IMonitoring’t treatment Costs
Mutter of Kecbct;on in I
Sysieme Poçaaiation Arraial Capital 0tH
tug/I) I Affected I Eaposed I t aijyr) (S m l (S m/yr)
Waste Costs
Capital DIM
c i a) tS 0/yr)
Total Costs I Aswajalized Costs ft yr)
Capital DIM I
tie) (Salyi) a3x ass a x
Estimated tepacta at Host Stringent MCIs
tat imated lepaci of Current MCLI Promulgated Under MiPOsis
19 16,800 I
188 168,711 I
20 111,884
I 3,034,994
2,527 I 1,309,354 I
185 I 699,122 I
5,340,865 I
2,939 2,305,871 I
I S i 5 0
I 59 4 I 31 2 I
19 2 8 I
I 553 62 99 ‘ I
I 469 67 I 26.4 12 I
I 45 5 I 3 0
2.0 I 1,130 141 I 410 24 I
sAl 577 311 15 I
10 1
70 6I
27
652 71 I
735
48 s
1.540 165
888 94,
Estimated Met incremental lepacts at Host sir ingent i4CLs
water systems serving less than 50,000 people for alt contaminants
include alt public water supplies serving fewer than 50,000 people.
over 20 years.
Bariin
Catiiaa
I Chrmeii,n
Mercury
Mliraie/Mltrlte
I Selenii.as
iota!
Total teas Mercury
I°°
Oi l
10 I
02 I
100
SI
6,7 13 I
760 I
1,323
1,378 I
108,227 p
1,713 I
8 .621 ,581 I
682,525
4,821,091 I
5 ,763,242 I
76,753,205 I
10,287,008 p
2,090
196
I 765
1,206
22,335
p 409
—2
206 I
18 I
78 I
137
3,456 I
120,114 p 106,998,652
118,736 p 101,225,410
2,032
127
572
234
12,536
672
6 1 27,001 3,942 I 16,173
BA 25,795 3,805 I 15 ,939
137 4,122 343
7 I 323 251
46 1,337 124 I
18 1,440 155
60 1 34,871 4,057
42 I 1,081 891
851 41,174 4,793
833 I 4i,734 4,638
Mama 1,000
Ca kaa 10
Chromlias 50
Mercury 2 I
I Nitrate/Nitr Ite 10,000
teieni.n 10 p
iota!
Total teas Mercury
Bar ia n
I Cadaitsa
I Chromiaaa
Mercury
I Miiraie/ itriie
Salenti.aa
total
total less Mercury
I 6,694 p
I 572 1
I i,303 I
I 933 I
105,700
I 1,528
8,604,781 I
511,814 I
4,779,207 I
2,728,248
75,441,85 1 I
9,587,886 p
I
iS?
14 I
96
746
76 I
564
I
655
75
135
I
21,866
3,389 p
12,2 72
366
42
116,750 101,657,78 7 p
p 1 15 ,79? p 98,929,539 I
4 1 25 ,871 3,801 15 ,763
a x 25,218 3,726 I 15,628
137 I 4,112 342
I 253 19 1
I 1,310 i
I 788 841
589 I 34,158 3,978
42 I 1,033 84 I
827 41,654 4.628 I
818 40,846 4,544
1 Includes Cofrninity and non-transient non-coimsinity
except Nitrate/Nitrite. Nitrate/Nitrite estimates
2 Costs expressed in 1986 Dot tars and are annualized
618
36
209
137
6,273
153
7,426
7,289
672 7301
39-
226 245 I
147 158 I
6,717 7,200 p
167 182
7,969 8,558
7,822 8,400
3 At most stringent MCI alternatives there may be significant co-occurrence of lOCs.
4 Routine cotrpliance monitoring costs include estimates for all contaminants; in addition, monitoring for unregulated contaminants
is estimated to result in a one time cost of $0.2 million.
-------
EXHIBIT 7-2
Estimated Impact of Regulating lOCs at Proposed MCL Alternatives
I I 4 Benefits: IMonitorin?I Treatment Costs Waste Costs 4 Total Costs 4 lAnnuatized Costs (S m/yr)
I Number oflReduction liii I I I I I
MCI Systems Population Annual I Capital 0&M Capital O&M Capital O&M I
Contaminant 1 (ug/l) I Affected I Exposed (S m/yr) (S m) (S m/yr) (S m) (S m/yr) I (S m) (S m/yr) a 3% a sz a 7%
Estimated Impact of Proposed MCI5
Cadmium 5 I 380 I 341,125 4.1 79 8 64 4 143 12 - - - - - -
I I I
Estimated Impact of Current MCL5 Promulgated Under NIPDWR
Ca d mium 10 168,711 2.0 3 9 4 I 31 2 I 70 6 I - - - - - -
Estimated Net Incremental Impacts of Proposed MCI5
Cadmium I 192 I 172,414 I 2.1 I 40 4 I 33 2 I 73 6 I 11 12 13 I
I I
1 The preferred MCL5 for barium, chromium, mercury, nitrate, and selenium are greater than or equal to the interim MCI5, hence
the incremental impacts of removing these contaminants are assisiied to be zero. Nitrite occurrence is assumed to be negtigible,
therefore no benefits or treatment and waste disposal costs would be incurred for this contaminant Benefits and treatment and waste
disposal costs for asbestos are subsumed by other drinking water regulations.
2 Include coomunity and non-transient non-coeininity water systems serving Less than 50,000 people.
3 Routine coepliance monitoring costs include estimates for all contaminants; in addition, monitoring for unregulated contaminants
is estimated to result in a one time cost of $0.2 million.
4 Costs are expressed in 1986 Dollars and are annualized over 20 years
-------
that nitrite occurrence is negligible and that no treatment or
waste disposal costs would be incurred under the proposal.
It is estimated that 192 water systems will incur treatment
costs for cadmium. Capital costs associated with the preferred MCL
alternatives are $73 million and the total annualized treatment and
waste disposal costs will amount to $11 to $13 million depending
on the discount rate applied. Monitoring costs will be
approximately $2 million per year.
An estimate of the total costs to state programs for the
entire Phase II regulatory package (i.e., including both lOCs and
synthetic organic chemicals) is included in the draft Regulatory
Impact Analysis for Synthetic Organic Chemicals.
7.2 Total Incremental National Benefits
Incremental national benefits at the most stringent MCL
alternatives and at the preferred MCL alternatives are summarized
in Exhibits 7-1 and 7—2, respectively.
Due to the lack of dose/response data for sub—lethal health
effects, it was necessary to present aggregate benefits estimates
for non-carcinogens as a reduction in population exposed. At the
most stringent MCL alternatives, it is estimated that approximat-
ely 102 million persons may be provided such a reduction in
exposure. This may over—estimate the true reduction in exposure,
however, due to the probability of co-occurrence at these concen-
trations.
At the proposed MCL alternatives, the total incremental
benefits will include a reduction in exposure to lOCs of approx-
imately 170,000 persons, and avoidance of six cases of cancer per
year due to arsenic exposure.
1 While treatment and waste disposal costs will be borne only
by systems which violate the standard for arsenic or cadmium,
monitoring costs will be borne by all systems, and include costs
for all contaminants in the proposed rule.
7—4
------- |