Drinking Water: Revisions to EPA's Cost Analysis for the Radon	 
Rule Would Improve Its Credibility and Usefulness (22-FEB-02,	 
GAO-02-333).							 
                                                                 
The Safe Drinking Water Act requires the Environmental Protection
Agency (EPA) to set a drinking water standard for radon. In a	 
proposed rule issued in November 1999, EPA presented a unique and
complex drinking water regulation for radon. GAO found that EPA's
analysis of the costs to implement the proposed radon rule has	 
several strengths. EPA's estimates of the typical costs for water
systems to buy and install radon removal technologies--a key	 
determinant of total national costs--are reasonable for 	 
estimating national compliance costs. Moreover, EPA used	 
recommendations from an expert panel to estimate the costs to	 
install and maintain radon removal equipment. EPA also developed 
a range of annual cost estimates, rather than a single estimate, 
to account for uncertainty about the extent to which the less	 
costly alternative standard will be adopted by states. EPA's	 
analysis of the national annual costs to comply with its proposed
radon drinking water rule has several limitations that, if	 
corrected, would likely increase EPA's best estimate of these	 
costs. EPA made two errors in estimating the various costs	 
associated with programs to reduce radon levels in indoor air	 
under the alternative standard--one that understated radon	 
testing and mitigation costs by $37 million and another that	 
overstated administrative costs by $31 million--resulting in a	 
combined understatement of costs by $6 million. In addition,	 
EPA's exclusion of "mixed" water systems, which use a mix of	 
groundwater and surface water sources, effectively understated	 
compliance costs by about $17 million.				 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-02-333 					        
    ACCNO:   A02733						        
  TITLE:     Drinking Water: Revisions to EPA's Cost Analysis for the 
Radon Rule Would Improve Its Credibility and Usefulness 	 
     DATE:   02/22/2002 
  SUBJECT:   Air pollution control				 
	     Cost control					 
	     Cost effectiveness analysis			 
	     Hazardous substances				 
	     Potable water					 
	     Proposed legislation				 
	     Radiation exposure hazards 			 
	     Water pollution control				 
	     Water quality					 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-02-333
     
United States General Accounting Office

GAO

Report to Congressional Committees

February 2002

DRINKING WATER

Revisions to EPA's Cost Analysis for the Radon Rule Would Improve Its
Credibility and Usefulness

GAO-02-333

Contents

Letter

Results in Brief
Background
EPA's Cost Analysis Has a Number of Strengths
Limitations of EPA's Cost Analysis Reduce Its Credibility and

Usefulness
Conclusions
Recommendations for Executive Action
Agency Comments
Scope and Methodology

1

3 4 9

15 27 28 28 29

Appendix I Adjustments to EPA's Best Estimates of Total National Costs to
Address Limitations in EPA's Economic Analysis

Appendix II GAO Contact and Staff Acknowledgments

United States General Accounting Office Washington, DC 20548

February 22, 2002

The Honorable James T. Walsh
Chairman
The Honorable Alan B. Mollohan
Ranking Minority Member
Subcommittee on VA, HUD, and Independent Agencies
Committee on Appropriations
House of Representatives

The Honorable Barbara A. Mikulski
Chairman
The Honorable Christopher S. Bond
Ranking Minority Member
Subcommittee on VA, HUD, and Independent Agencies
Committee on Appropriations
United States Senate

The National Academy of Sciences has estimated that about 20,000 lung
cancer deaths occur in the United States each year from exposures to
radon, a naturally occurring radioactive gas found in soil, rock formations,
and water. The vast majority of these deaths result from the inhalation of
radon that has been released to indoor air from soil beneath homes.
However, about 160 of these deaths are estimated to stem from inhalation
of radon that has evaporated from drinking water. The Safe Drinking
Water Act, as amended, mandates that the Environmental Protection
Agency (EPA) issue a regulation setting a drinking water standard for
radon. In setting the standard, EPA must, among other things, consider the
costs and benefits of control programs for radon from other sources, such
as air. While EPA does not directly regulate radon in indoor air,1 the Safe
Drinking Water Act authorizes EPA to set a drinking water standard that
allows states and water systems to offset high radon levels in drinking
water with reductions of radon levels in indoor air.

In a proposed rule issued in November 1999, EPA presented a unique and
complex drinking water regulation that used the framework for regulating
radon specified in the Safe Drinking Water Act. Under the proposed radon

1According to EPA, the agency does not have the statutory authority to
directly regulate radon in indoor air.

rule, states and water systems could, for the first time, choose one of two
different standards, or limits, for a drinking water contaminant. The first
standard reflects the typical regulatory approach under the Safe Drinking
Water Act-that is, it imposes a health-based limit on the level of radon in
drinking water and requires water systems to provide drinking water that
does not exceed the limit. The second standard, called the alternative
standard, allows considerably higher levels of radon in drinking water, but
may be used only when an EPA-approved program to reduce radon in indoor air
is also implemented. The alternative standard allows higher levels of radon
in drinking water because the associated health risks are offset by reduced
exposure to radon in indoor air. EPA believes that the most cost-effective
approach to reducing the health risks associated with radon in water is to
use the alternative standard, along with a program to reduce radon levels in
indoor air.

EPA estimated that the benefits of its proposed rule-consisting almost
entirely of reduced cancer deaths-would be worth $362 million annually. EPA
estimated that the annual cost of implementing the rule would range from $60
million to $408 million, with $121 million as EPA's best estimate.2 EPA's
cost estimates are designed to reflect the typical costs that water systems
would incur to monitor water and install treatment technologies, when
needed, to comply with the regulation. The estimates also reflect the
administrative costs that states and water systems would incur to implement
programs encouraging homeowners to reduce radon in indoor air and
homeowners' costs to do so. As of February 2002, EPA has not yet developed a
schedule for issuing the final rule.

Citing the significant financial implications of the 1999 proposed rule, the
conference report accompanying EPA's fiscal year 2001 appropriations act
directed us to report on the financial impacts of the proposed radon rule
and encouraged EPA to consider our findings prior to finalizing the rule. As
agreed with your offices, this report identifies the (1) strengths and (2)
limitations of EPA's cost analysis for the proposed rule. Because the scope
of our review was limited to EPA's cost analysis, we did not evaluate EPA's
analysis supporting the benefits the agency expects the proposed rule to
provide. EPA officials said the agency plans to revise its estimate of the
benefits in the final rule.

2EPA reported its estimates of the costs and benefits of the radon rule in
1997 dollars.

Results in Brief

EPA's analysis of the costs to implement the proposed radon rule has a
number of strengths. First, EPA's estimates of the typical costs for water
systems to purchase and install radon removal technologies-a key determinant
of total national costs-are reasonable for estimating national compliance
costs. Most of the radon removal technologies that EPA's cost analysis
assumes will be used most frequently are already being used by a number of
water systems across the country to remove radon and other contaminants, and
EPA's cost estimates are generally within the ranges of costs reported in
case studies on these technologies. Other strengths of the analysis include
EPA's (1) use of recommendations from an expert panel in estimating water
systems' costs for construction, engineering, and labor needed to install
and maintain radon removal equipment and (2) development of a range of
annual cost estimates, rather than a single estimate, to account for
uncertainty about the extent to which the less costly alternative standard
will be adopted by states.

EPA's analysis of the national annual costs of complying with its proposed
radon drinking water rule has several limitations that, if corrected, would
likely increase EPA's best estimate of these costs. Specifically, EPA

* made two errors in estimating the various costs associated with programs
to reduce radon levels in indoor air under the alternative standard-one that
understated radon testing and mitigation costs by about $37 million and
another that overstated administrative costs by about $31 million- resulting
in a combined understatement of costs by about $6 million; and

* excluded from its analysis "mixed" water systems, which use a mix of
groundwater and surface water sources, effectively understating compliance
costs by approximately $17 million.

Correcting these errors and including the mixed water systems alone would
increase EPA's estimate by about 20 percent, from about $121 million to
about $145 million (see app. I). Other aspects of EPA's analysis may have
further understated total compliance costs, although by an unknown amount.
For example, EPA assumed that more water systems than suggested by some
evidence would choose the less-costly alternative standard. In addition, EPA
may have insufficiently considered potential costs faced by water systems to
alleviate concerns about the use of water treatment technologies that would
vent radon near homes, schools, or other population centers. Further, EPA's
cost analysis reflects inadequate quality assurance, does not consistently
comply with EPA's guidelines calling for clarity and transparency in
presenting economic analyses, and includes a number of inaccuracies. For
example, the economic analysis contains inconsistent and inaccurate
information about an important cost

factor as well as unclear or conflicting information about the
responsibility for certain costs, and it omits assumptions underlying a key
cost determinant. EPA has taken steps to improve the agency's quality
assurance process, but it is not clear that these steps will be sufficient
to identify and correct deficiencies like those we found.

Because of the limitations we identified, we are making several
recommendations aimed at improving EPA's cost analysis for the radon rule
and its economic analyses in general. In commenting on a draft of this
report, EPA officials from the Offices of Water; Air and Radiation; and
Policy, Economics, and Innovation generally agreed with our findings and
recommendations, providing some technical and editorial suggestions that we
have incorporated into the report, as appropriate. However, while agreeing
that the proposed rule contained the errors and other inaccuracies we
identified, the Office of Water did not agree that an effect of these errors
was reduced credibility of EPA's cost analysis. While our report identifies
a number of strengths of the cost analysis, these strengths do not offset or
negate the effects of the errors and inaccuracies we found. We continue to
believe that, collectively, the limitations we identified reduced both the
credibility and usefulness of the cost analysis for the radon rule.

Background The majority of the estimated cancer deaths in the United States
associated with radon are related to the inhalation of radon in outdoor or
indoor air. Specifically, the National Academy of Sciences has estimated
that about 700 deaths occur from inhaling radon in outdoor air and more than
19,000 deaths occur each year from inhaling radon in indoor air.3 Some radon
in indoor air is derived from the evaporation of drinking water, and the
academy estimated that about 160 of these 19,000 deaths are associated with
such releases. In addition, the academy estimated that consumption of
drinking water containing radon causes about 23 deaths from stomach cancer
each year. EPA relied on the academy's estimates of radon health risks in
developing the proposed rule on radon in drinking water.

EPA's proposed radon rule applies to water systems that include only
groundwater and those that include both ground and surface water (rivers

3National  Academy of Sciences, Risk Assessment of Radon  in Drinking Water,
Washington, D.C.: 1999.

and lakes), referred to as mixed systems.4 Radon is usually present in only
negligible amounts in surface water because this water is exposed to the air
and the radon in it will tend to be released to the air. Groundwater
originating in underground aquifers is not similarly exposed to air. As a
result, high levels of radon are sometimes found in groundwater that
collects and flows under the earth's surface. Radon levels in groundwater
vary across the country, with the highest levels in New England and the
Appalachian uplands of the Middle Atlantic and Southeastern states. There
are also isolated areas in the Rocky Mountains, California, Texas, and the
upper midwest where radon levels tend to be higher than the U.S. average.
According to EPA, at the state level, high levels of radon in drinking water
can occur in areas with low levels of radon in the soil (thus low levels in
indoor air) and vice versa.

1996 Safe Drinking Water Amendments Require EPA to Establish a Radon
Standard

Under the Safe Drinking Water Act, EPA sets health-based, legally
enforceable standards limiting the level of drinking water contaminants that
can adversely affect public health. In developing a standard, EPA
establishes a health-based goal at a level that causes no known or
anticipated adverse health effects and that allows an "adequate margin of
safety." If a contaminant, such as radon, is likely to cause cancer, EPA
generally sets the goal at zero. After setting the goal, EPA typically
issues a regulation establishing an enforceable standard, called a maximum
contaminant level, that is as close to the health-based goal as is feasible,
considering the available technology, treatment techniques, and costs. Under
the 1996 amendments, when proposing a standard, EPA is also required to
perform an economic analysis to determine whether the benefits of the
standard justify the costs. If the benefits do not appear to be justified,
EPA may adjust the standard to a level that "maximizes health risk reduction
benefits at a cost that is justified by the benefits."

The 1996 amendments also included a number of provisions specific to a radon
standard. First, the amendments required EPA to withdraw its 1991 proposed
rule on radon. Before issuing a new proposed rule on radon, EPA was required
to obtain from the National Academy of Sciences an

4The proposed rule excludes certain groundwater systems: (1) nontransient
noncommunity water systems are excluded on the basis that the more limited
exposure to radon from drinking water in the schools, hospitals, and
factories in this category results in lower health risks compared with
life-time exposures in homes and (2) transient noncommunity water systems
are excluded because most people who use such facilities (service stations,
campgrounds) do so only occasionally.

assessment of the health risks from radon in drinking water and arrange for
the academy to assess the health risk reduction benefits from various
measures to reduce radon levels in indoor air. In addition, EPA was also
required to (1) publish and seek public comment on its analyses of the costs
and health risk reduction benefits for standards being considered for radon
in drinking water and (2) respond to all significant public comments
received on the analyses in the preamble for the proposed rule. In setting a
radon standard for drinking water, EPA was also required, under the
amendments, to consider the costs and benefits of programs to reduce radon
exposures from other sources, such as indoor air.

The amendments specified that if the drinking water standard is more
stringent than necessary to reduce the concentration of radon in indoor air
from drinking water to a level equivalent to the national average
concentration of radon found in outdoor air, then EPA must also promulgate
an alternative standard (alternative maximum contaminant level). The
alternative standard must be set at a level that would result in a
concentration of radon in indoor air from drinking water equivalent to the
national average concentration of radon in outdoor air. The alternative
standard would allow more radon in drinking water than the more stringent
standard. To offset the higher level, the water systems that use the
alternative standard must be covered by an EPA-approved program to reduce
radon in indoor air. Such a program could be managed by either states or
water systems. To be approved, a program's expected health risk reduction
benefits must be equal to or greater than the health risk reduction benefits
that would result from compliance with the more stringent standard for radon
in drinking water. Finally, the 1996 amendments required EPA to propose a
new radon rule in 1999 and to promulgate it within 12 months. EPA issued the
proposed rule in 1999. As of February 2002, EPA has not yet developed a
schedule for issuing a final rule.

EPA Developed a Proposed Radon Rule and a Supporting Economic Analysis

In developing the proposed rule, EPA obtained and relied on assessments
conducted by the National Academy of Sciences and issued its economic
analysis for public comment in February 1999.5 When EPA issued the proposed
radon rule in November 1999, it set the health goal at zero, proposed a
standard of 300 picocuries per liter of water, and proposed an

5U.S. Environmental Protection Agency, Radon in Drinking Water Health Risk
Reduction and Cost Analysis (Washington, D.C.: Feb. 26, 1999).

alternative standard of 4,000 picocuries per liter to be used in conjunction
with state or water system programs that reduce radon in indoor air.6

In the economic analysis supporting the proposed radon rule, EPA indicated
that 46 percent of the 40,863 groundwater systems that would be subject to
the rule would, in the absence of any additional treatment, exceed the
standard of 300 picocuries. (The majority of these water systems serve 500
or fewer customers.7) Further, EPA estimated that if these systems had to
comply with the more stringent standard, the systems would incur costs of
about $2.5 billion to purchase and install radon treatment technologies, or
about $233 million each year for 20 years. EPA estimated that the total
annual cost of the rule would be about $408 million,8 as follows

* $233 million to purchase and install radon treatment technologies,

* $152 million to operate and maintain the treatment technologies,

* $14 million to monitor water for radon,

* $6 million to administer these activities, and

* $2.5 million for states to oversee the water systems.

However, EPA estimated that the vast majority of water systems would not
incur any water treatment costs because they would be subject to the
alternative standard of 4,000 picocuries. EPA estimated that only about 4
percent of water systems subject to the rule have radon levels in excess of
the alternative standard. EPA provided lower-bound, best, and upper-bound
cost estimates that reflect varying assumptions about the proportion of
states and local water systems that would choose the alternative standard
and thus implement programs to reduce radon levels in indoor air, as
follows:

* EPA's upper-bound estimate of $408 million annually assumed that all water
systems would be subject to the more stringent water standard.

* EPA's best estimate of $121 million annually assumed that about two-thirds
of water systems would be subject to the less stringent alternative
standard.

6A picocurie is one trillionth of a curie, a unit of radioactivity.

7Most  (96 percent) groundwater systems serve  10,000 or fewer customers; 67
percent of the systems serve 500 or fewer customers.

8Most of the estimated cost-$405 million-would be borne by water systems.

* EPA's lower-bound estimate of $60 million annually assumed that about 97
percent of water systems would be subject to the less stringent alternative
standard.

The benefits that EPA estimates would be provided by the proposed rule- $362
million annually-are the same under all the scenarios, differing only in the
extent to which the lives are saved because of water treatment versus
reductions of radon in indoor air. Considering these costs and benefits,
along with more subjective benefits and costs that EPA did not quantify,9
EPA determined that the costs of the proposed rule were justified by the
benefits. EPA stated that in making this determination, the agency also
considered that costs would be substantially less than $408 million annually
if most states implement indoor air programs, allowing water systems to
comply with the less stringent water standard. Regarding the benefits, we
note that EPA's estimates of future benefits in the proposed rule were not
discounted (reduced) to present value, although the cost of purchasing and
installing radon equipment was appropriately discounted. According to EPA
officials, EPA will discount the benefits in the final rule in response to
recommendations from the Science Advisory Board.

EPA's Guidelines for Preparing Economic Analyses highlight the importance of
economic analyses in making informed policy choices and specify criteria for
effective presentation of economic analyses, such as the cost analysis
supporting EPA's proposed radon rule.10 Primary criteria include clarity and
transparency of all aspects of the analyses and descriptions of all
important data sources, key assumptions, and their justifications. In
addition, EPA's Office of Water has a quality management plan to guide its
quality assurance and control activities, specifying which types of such
activities are necessary and the various procedures for conducting quality
reviews. The quality assurance elements include internal peer review,
external peer review, external agency review, and stakeholder meetings,
among other measures.

9These nonquantified benefits and costs include, among other items, customer
comfort from knowing that radon is being removed from their water and
customer anxiety about living near treatment plants that emit radon gas.

10EPA replaced its 1983 Guidelines for Performing Regulatory Impact Analyses
with Guidelines for Preparing Economic Analyses in September 2000. EPA was
using draft revised guidelines for economic analyses when the proposed radon
rule was issued. The criteria for effective presentation of economic
analyses were substantially the same in the 1999 draft guidelines and the
issued guidelines.

EPA's Cost Analysis Has a Number of Strengths

EPA's cost analysis has a number of strengths. First, and most importantly,
the estimates of the typical costs for a water system to remove radon are
reasonable for the purpose of estimating the rule's national costs. The
estimates of typical costs are for currently used technologies whose
effectiveness is generally known and for which published cost data are
available. Moreover, these estimates have been improved by input from a blue
ribbon panel of drinking water and cost experts and other stakeholders.
Further, while some concerned parties questioned several of EPA's
assumptions on water treatment issues-such as the extent to which water
systems would have to address certain water quality issues and how they
would be addressed-our review indicated that EPA generally had a reasonable
basis for its assumptions. Another strength of the analysis is that EPA
estimated a range of costs to account for uncertainty about the approach
that states will use to comply with the rule.

EPA's Estimates of Typical Costs for Water Treatment Are Reasonable for
Estimating National Costs

EPA's estimates of the typical costs for water systems to purchase, install,
and operate radon removal technologies-a key determinant of total national
costs to implement the proposed rule-are reasonable for use in estimating
national compliance costs. To estimate the typical costs for a system to
remove radon from drinking water, EPA generally assumed the use of aeration
technologies that have been commonly used by water systems to remove radon
and other contaminants, such as volatile organic compounds.11 EPA estimated
radon removal costs for eight size categories of water systems, ranging from
those that serve between 25 to 100 people to those serving between 100,000
and 1 million people. EPA's cost estimates for systems to purchase and
install treatment technologies ranged from about $45,000 for the smallest
systems to more than $6 million for the largest-or about $4,200 to $580,000
per year for 20 years. In addition, EPA estimated operations and maintenance
costs ranging from about $3,600 per year for the smallest systems to about
$440,000 for the largest.

Information on the cost of installing aeration technologies is available in
published case studies, which EPA used to cross-check its estimates. For
example, in 1998, the American Water Works Association published a guide for
water utilities to use for evaluating and selecting radon treatment

11 Aeration technologies  force air  through drinking  water and  strip away
contaminants, which are then vented into outdoor air.

technologies that includes construction cost information for 33 water
treatment sites, including 12 very small facilities, and performance
information for most of the sites. EPA's technology cost estimates are
generally within the ranges of costs identified in this and in other case
studies. Such data are not always available to inform regulatory cost
estimates. For example, in estimating costs for its arsenic rule, EPA had to
rely on more limited data because some of the technologies for removing
arsenic from drinking water are not commonly used.

Moreover, EPA's estimates of the typical costs for a water system to remove
radon have benefited from the recommendations of an expert panel of water
design and cost engineers from utilities, state and federal agencies,
consulting firms, and public utility regulatory commissions. Following the
reauthorization of the Safe Drinking Water Act, EPA convened the panel to
help improve the accuracy of the agency's cost estimates for all drinking
water regulations. EPA's cost analysis for the radon rule relies on the
panel's recommendations in estimating a water system's cost for
construction, engineering, and labor needed to install and maintain radon
removal equipment. For example, based on the panel's recommendations, EPA
increased its estimates of the labor costs to operate and maintain such
equipment to include not only base salaries but also fringe and other
benefits.12

In addition, EPA incorporated advice from other stakeholders in developing
its estimates. For example, in response to comments that its initial
estimates were not adequate, EPA increased its radon technology costs for
pumps and blowers needed to operate aeration equipment.

Although some stakeholders said that EPA did not increase its technology
cost estimates sufficiently, our analysis of the key issues they raised
indicates that EPA generally used reasonable assumptions in developing its
estimates, as the following examples show.

* Some parties commented that EPA did not include adequate costs for water
systems to remove iron and manganese from water. These parties said that
water systems would, in many cases, need to remove iron and manganese from
their water before it is aerated so as not to damage the

12EPA analysts had previously assumed labor rates for water treatment
professionals to be about $15 per hour. For the proposed rule, EPA adjusted
labor rates to add fringe and other benefits, resulting in hourly labor
costs ranging from $28 to $52, depending on the size of the water system.

aeration equipment. However, EPA assumed that water systems that need to
treat for iron and manganese would generally be able to add chemicals to
neutralize these elements, which is less expensive than removing them. Based
on estimates of the number of water systems with elevated levels of iron and
manganese, EPA included costs for 25 percent of small systems (systems
serving fewer than 10,000 people) and 15 percent of large systems (systems
serving more than 10,000 people). In addition, EPA assumed that systems with
levels of iron and manganese too high for chemical neutralization would
already be removing these elements because high levels of these elements
result in unacceptable discoloration of water. Because these removal costs
would not be incurred as a result of the radon rule, EPA's cost estimates do
not include them. We believe that EPA's assumptions are reasonable for the
purpose of estimating national costs.

* Some parties commented that EPA did not include adequate costs for water
systems to disinfect water that might be contaminated by microbes during
aeration. For example, a stakeholder said that the cost estimates may be
understated for clearwells-wells or tanks that are needed to hold water so
that it can be disinfected. This comment stemmed from two conflicting sets
of cost estimates for clearwells, one much higher than the other, which EPA
included in a supporting report on technology costs. EPA used the lower
estimates in its cost analysis. EPA officials told us that the higher
estimates were incorrect and were inadvertently included in the supporting
report and that the lower estimates-generated by EPA's cost model for
aeration technologies-were correct. The lower estimates are consistent with
EPA's guidance manual for disinfecting drinking water and incorporate best
engineering judgment.13 We note that national costs for clearwells may be
overstated because EPA included these costs for all systems that add radon
treatment. However, as case studies show, a number of systems (particularly
those that already disinfect their water) will be able to use existing
clearwells.

* A stakeholder commented that EPA did not include adequate costs to address
the increased corrosiveness of water resulting from aeration. We believe,
however, that EPA's addition of such costs for a small portion of water
systems is appropriate based on information from the National Academy of
Sciences and case studies indicating that aeration does not generally
increase the corrosiveness of water.

13U.S. Environmental Protection Agency, Guidance Manual for Compliance With
the Filtration and Disinfection Requirements for Public Water Systems Using
Surface Water Sources (Washington, D.C.: March 1991).

* A stakeholder commented that EPA might have understated treatment costs
for large water systems because it underestimated the number of treatment
sites at these systems. Underestimating the number of treatment sites that
potentially need to have radon treatment technologies would understate costs
because costs increase as the number of sites needing radon treatment
increases. In its cost analysis, EPA estimated that there were an average of
13.1 treatment sites for groundwater systems serving between 100,000 and
1,000,000 customers.14 In contrast, the stakeholder commented that its
survey of water systems serving over 100,000 customers indicated the correct
number of average treatment sites would be 23.8. We believe, however, that
EPA's estimate was adequately supported, based on our review of EPA's
assumptions and data as well as the information provided by the stakeholder.
For example, the stakeholder's estimate of 23.8 sites included systems
serving more than 1,000,000 customers and therefore was not directly
comparable to EPA's estimate. When, as part of our review, the stakeholder
provided us with an estimate that was consistent with EPA's size categories,
the estimate was 14.7-a number reasonably close to EPA's estimate of 13.1.
Further, our limited review of the stakeholder's data indicates that such
data would need to be evaluated for accuracy and representativeness.
Specifically, in reviewing information about 4 of the 102 survey
respondents, we found that an entity identified as having 650 sites-an
atypically high number15- was incorrectly classified as a single system.
This entity represents a number of affiliated water districts of varying
sizes in different locations, each with its own rate structure. The
stakeholder acknowledged that these water districts should have been
analyzed as individual water systems, not aggregated into one system.16

In finding that EPA's technology cost estimates are appropriate for
estimating national costs, we recognize that some systems would incur higher
costs than EPA estimates and others would incur lower costs. Costs for
individual water systems-even those of similar size-would vary

14This is the largest category of groundwater systems for which EPA included
costs. According to EPA, its review of the two groundwater systems that
serve more than 1 million customers indicated that these systems would not
have to treat for radon.

15According to the stakeholder's survey data, most large systems have an
average of 11 sites.

16Because this system serves more than 1,000,000 customers, its inclusion
distorted the stakeholder's estimate of 23.8 sites per system but did not
distort its subsequent estimate of 14.7 sites.

depending on factors such as the technologies selected and the site-specific
conditions, including water quality and management skills. EPA's estimates
should not be interpreted as being representative of the individual
decisions that thousands of water systems will make on the basis of their
unique circumstances.

EPA Supported Its Assumption about States' Compliance with the Rule

In developing its best estimate of total national costs for the rule, EPA
assumed that 50 percent of states would implement EPA-approved indoor air
programs that have expected health risk reduction benefits equal to or
greater than the benefits that would result from compliance with the more
stringent drinking water standard. As a result, the vast majority of local
water systems in those states would not have to mitigate radon levels in
drinking water because only those with radon levels in excess of 4,000
picocuries (the alternative standard) would have to reduce the radon in the
water.17 EPA's assumption was supported by the results of a survey of state
indoor air and drinking water officials conducted by the American Water
Works Association.18 Fifty percent of the survey respondents indicated that
they probably would adopt the indoor air option, and 9 percent indicated
that they definitely would.

EPA's assumption about the decisions states would make is also supported by
extensive discussions between EPA and state representatives and recognizes
the potential difficulties that may dissuade some states from implementing
the option. Most states already conduct indoor air programs for radon. To
comply with the proposed rule, states would have to upgrade their existing
programs by setting quantitative risk reduction goals, reporting on progress
toward those goals, and allowing for extensive public participation in
developing the indoor air programs. According to EPA, program officials from
many states believe they would be able to comply with these requirements
without too much difficulty. Furthermore, officials from many states believe
that it makes good public health sense to mitigate radon levels through
indoor air, rather than water, because indoor air risks are much higher than
drinking water risks and indoor air mitigation is more cost-effective.
However, other states are not inclined to use the indoor air option to
comply with the rule because of concerns about potential public relations
problems that could arise

17EPA estimates that 1,776 systems nationwide have radon levels above 4,000
picocuries.

18American Water Works Association, State Response to the Proposed
Multimedia Mitigation Program Option, Washington, D.C.: 2000.

because of different protections people will receive, depending upon where
they live. Specifically, the executive director of the Association of State
Drinking Water Administrators told us that states are greatly concerned
about explaining to the public that the same level of radon in drinking
water is safe in some states but unsafe in others, depending on whether the
state adopted the indoor air option. She stated that trading the drinking
water risks of the community for the indoor air risks of relatively few
residents would cause substantial public confusion. Similarly, the National
Academy of Sciences stated that concerns regarding the equity of risk
trading "might ultimately constitute the deciding factor" in whether the air
mitigation option is undertaken. A program official from one state told us
that her state would likely not adopt the air mitigation option partially
because the unequal treatment of radon risks could lead to court cases.

Nonetheless, according to EPA officials, their estimate that 50 percent of
the states would upgrade their indoor air programs to comply with the radon
drinking water rule may be conservative. These officials told us that after
several workshops EPA held with the states in late 2000, nearly two-thirds
of the state program managers indicated that they would likely adopt the
indoor air option. However, an EPA official acknowledged that some state
program managers that favor the program may not be the ones making this
decision, and therefore some of these "likely" states may not adopt the
program. This recognition corroborates a potential impediment that program
officials from some states had discussed with us-that is, adopting the
indoor air option could be subject to the approval of the legislature or the
governor. As a result, the decisions of the state program managers may be
subject to political processes, the outcomes of which cannot be certain.
Overall, we believe that EPA had a reasonable and supported basis for its
assumption, reflected in its best estimate of costs, that 50 percent of
states would choose the indoor air option, as the assumption reflects the
potential disincentives that could prevent some states from participating.

EPA's Estimates Include a Range of Costs to Account for Uncertainty about
How States Would Choose to Comply with the Rule

Another strength of EPA's cost analysis for the proposed radon rule is that
it provided a range of national annual compliance costs to address the
uncertainty about how states would choose to comply with the rule. EPA's
designation of a range helps decisionmakers and the public understand how
costs could vary depending on how the rule is implemented, and it is
consistent with EPA's guidelines for preparing economic analyses, as well as
with guidance issued by the Office of Management and Budget in 2000. Because
it is generally less costly to reduce radon risks in indoor air than

Limitations of EPA's Cost Analysis Reduce Its Credibility and Usefulness

in drinking water, the estimated costs of compliance with the radon rule are
lower when more states are assumed to implement indoor air programs.
Reflecting a range of potential responses, EPA's cost estimates range from
$60 million to $408 million per year.19 EPA designated $121 million as its
best, or most likely, estimate of annual costs.

EPA's analysis of the national annual cost of complying with its proposed
radon drinking water rule has several limitations that, if corrected, would
likely increase EPA's best estimate of national costs. Specifically, EPA
made two errors in estimating the annual costs associated with programs to
reduce radon levels in indoor air under the alternative standard-one that
understated radon testing and treatment costs by about $37 million and
another that overstated administrative costs by about $31 million.
Correcting the two errors would increase EPA's best estimate of national
annual costs for the proposed rule by about $6 million. In addition, EPA
excluded from its analysis mixed water systems-those that get their water
from a mix of both groundwater and surface water sources-which effectively
understates compliance costs by approximately $17 million. Correcting these
errors and including the mixed water systems alone would increase EPA's
estimate by about 20 percent, from about $121 million to about $145 million.
Other aspects of EPA's analysis may have underestimated total compliance
costs, although by an unknown amount. Specifically, EPA

* assumed that more water systems than suggested by some evidence would
choose the less-costly alternative standard and

* insufficiently considered potential costs faced by water systems that
remove radon from water using aeration facilities-which vent the radon to
the outside air-located near homes, schools, or other population centers.

Further, EPA's cost analysis reflects a lack of quality assurance in certain
important respects. For example, the analysis contains inconsistent and
inaccurate information about an important cost factor and unclear or
conflicting information about the responsibility for certain costs, and
omits the assumptions underlying a key cost determinant. EPA has taken steps
to improve its quality assurance process, but it is not clear that they

                   19EPA's estimates are in 1997 dollars.

will  be  sufficient to  identify  and  correct deficiencies  like those  we
identified.

EPA Miscalculated Costs for the Rule's Indoor Air Option

In estimating costs for reducing radon levels in indoor air under the
alternative water standard, EPA made two errors-one that understated the
costs of testing and treating indoor air for radon by about $37 million and
another that overstated administrative costs to implement and oversee the
programs by about $31 million. Correcting the two errors would increase
EPA's best estimate of national annual costs for the proposed rule by about
$6 million. (See app. I.)

EPA's first error resulted in an underestimate of the annual costs for
reducing radon levels in indoor air that the agency expects will be borne by
individual households. EPA multiplied its assumed cost of testing and
treating for radon in indoor air per life saved ($700,000)20 by the expected
number of lives saved per year,21 and then amortized the result over 20
years at a discount rate of 7 percent. However, EPA should not have
amortized the result because it represents costs that would be incurred
every year. For example, in its best estimate, EPA assumed that 59 lives22
would be saved every year at a total cost of $41 million (59 lives times
$700,000 per life). EPA then amortized the $41 million over 20 years,
resulting in an estimated annual cost of $3.9 million, even though the $41
million cost would be incurred each year. EPA officials acknowledged that
they should not have amortized the $41 million cost and told us that they
intend to correct this error in EPA's economic analysis for the final rule.

20This assumption is based directly on an estimate described in EPA's
Technical Support Document for the 1992 Citizen's Guide to Radon (May 1992).
EPA did not adjust the estimate, which was reported in 1991 dollars, for
inflation because it believed that testing and treatment costs had not
increased since 1991.

21EPA's estimate of the number of lives saved per year assumes that the
indoor air programs adopted in conjunction with the less stringent drinking
water standard would save the same number of lives as compliance with the
more stringent standard. This assumption is based on the requirements of the
1996 amendments to the Safe Drinking Water Act.

22EPA estimated that if all systems were required to comply with the
alternative standard of 300 picocuries, 62 lives would be saved each year.
However, in the case of EPA's best estimate, about 95 percent of systems
either would be in states with indoor air programs or would implement their
own indoor air programs, so indoor air programs would be expected to save 59
lives (95 percent of the 62 lives that would be saved assuming that all
systems were required to comply with the more stringent standard of 300
picocuries).

The second error occurred because EPA inadvertently included costs to
administer and oversee indoor air programs of water systems that will not
implement such programs. EPA's estimate did not reflect the fact that
regardless of whether the states choose to implement indoor air programs,
water systems with radon levels below 300 picocuries would be required only
to monitor the radon level in their water. They would not be required to
treat their water, nor would they be required to implement their own indoor
air programs. The estimate also did not reflect that water systems with
radon levels above 4,000 picocuries are unlikely to implement indoor air
programs because they would be required to treat their water regardless of
whether they implement these programs. As a result, EPA's best estimate of
national annual costs included $53 million in costs associated with
oversight and administration of about 18,400 water systems' indoor air
programs, inadvertently including costs for about 10,800 water systems. We
estimate that correcting this error would reduce costs associated with
oversight and administration of water systems' indoor air programs to about
$22 million. EPA plans to correct both errors in its economic analysis for
the final rule.

EPA Excluded Certain Systems That Would Be Subject to the Rule

EPA's cost analysis excluded mixed water systems, which get their water from
a combination of groundwater and surface water sources, even though these
systems would be subject to the radon rule. EPA officials told us that they
did not include costs for mixed systems in the economic analysis because of
data limitations and because their preliminary analysis indicated that
including the mixed water systems in its analysis would not have a
significant effect on the total annual cost of complying with the rule.
However, EPA did include mixed systems in its economic analysis for the
January 2001 arsenic rule. An EPA official told us that there does not seem
to be a strong technical basis for handling mixed systems differently in the
two rules. In January 2002, EPA officials told us that the agency would
consider including costs for these systems in the economic analysis for the
final rule. We estimate that including mixed systems in EPA's best estimate
would further increase total annual costs by about $17 million

(see app.  I).23 Including  these systems would also  increase the estimated
benefits of the proposed rule.24

EPA May Have Underestimated Compliance Costs for Some Water Systems

A key factor in EPA's cost estimate is the extent to which programs to
mitigate radon levels in indoor air would supplant the more costly approach
of mitigating radon levels through water treatment. While EPA's assumption
about the number of states that would adopt indoor air programs is well
supported, we found that its assumption about the number of local water
systems that would do so is not and appears to be overly optimistic. A
decrease in the estimated number of systems choosing the less expensive
approach would increase the total annual cost of compliance.

In the 50 percent of states where EPA did not assume selection of the indoor
air option, EPA assumed that 90 percent of local water systems would elect
the alternative standard and establish their own indoor air programs.
According to EPA officials, the assumption was based solely on the premise
that water systems would choose the least costly approach to mitigating
radon risks. EPA officials acknowledged that they did not collect any data
on the extent to which water systems would establish air programs.

This data limitation reduces the credibility of EPA's optimistic assumption
in light of questions that have been raised about the likelihood of small
water systems adopting air programs. For example, EPA's assumption conflicts
with the opinions of the National Academy of Sciences and state and industry
associations. Specifically, the National Academy of Sciences reported in
1999 that "non-economic considerations" could play a large role in a local
water system's decision about whether to use an indoor air program to meet
the rule's requirements. According to the academy's report, experience with
certain provisions of the Clean Water Act indicates

23Our estimate of the costs for mixed systems is based on an estimate that
EPA developed and we reviewed. The estimate includes 1,074 mixed water
systems receiving more than 50 percent of their water from groundwater
sources. It does not include the costs for three systems serving more than
1,000,000 customers that receive some of their water from groundwater
sources and that EPA believes would incur costs to comply with the radon
rule. We did not develop an estimate for these systems, which should also be
included in EPA's cost estimate for the final radon rule.

24Of the limitations we found, the exclusion of mixed systems is the only
one that has an effect on EPA's estimates of the benefits of the rule.

that small entities have had difficulties dealing with complex federal
program requirements. While EPA may be correct in its assertion that the
requirements for indoor air programs are not as complex as the requirements
cited by the academy, small water systems may have limitations or concerns
that could lead them to choose compliance with the more stringent standard.
For example, the executive director of the Association of State Drinking
Water Administrators told us that local water companies, especially the
small ones, will not want to be involved in public meetings, goal setting,
and program monitoring and reporting- activities required under the proposed
rule's indoor air option. The National Association of Water Companies, a
trade association representing the nation's privately owned drinking water
utilities, expressed similar doubts in its comments on the proposed rule:
"We believe that the prospect of water systems implementing local (indoor
air) programs in the absence of state programs is unrealistic... Tracking
new home construction and remedial venting of existing homes is far removed
from the chartered objectives of community water systems, not to mention the
expectations of water ratepayers." In our view, these are reasonable
concerns.

The academy also raised concerns that the indoor air option may not be
practical for some local water systems that have elevated radon in their
water but not in their customers' indoor air. EPA acknowledges that elevated
radon in drinking water and in indoor air may not occur in the same
geographic area. Program officials from several states concurred that the
indoor air option may be problematic for some local water systems for this
reason. One of these officials also said that using the indoor air option
would not work nearly as well at the local level as it would at the state
level. The official explained that because states have a larger geographic
area than local water systems, states would have a much better chance to
offset one area's elevated radon in drinking water by mitigating another
area's indoor air radon.

Considering the evidence indicating a fair amount of uncertainty about the
extent to which local water systems, in the absence of a state program,
would choose the indoor air option to comply with the rule, EPA's assumption
that 90 percent of systems would do so appears overly optimistic. As
discussed previously, the uncertainty about adopting the indoor air option
is particularly strong for small water systems-the majority of systems
subject to the rule. Adjusting EPA's assumptions to reflect less optimistic
scenarios would increase the total national cost estimate for the proposed
rule. For example, assuming that 75 percent, instead of 90 percent, of local
water systems would choose the indoor air

option would increase national annual costs by $23 million; assuming 50
percent of systems would choose the option increases the estimated national
annual cost by $61 million (see app. I). In January 2002, EPA officials told
us that in finalizing the economic analysis for the rule, they plan to
include a range of costs based on different assumptions about how many water
systems will adopt indoor air programs. EPA's planned use of varying
assumptions about the choices of water systems would appropriately reflect
the uncertainty associated with the responses of water systems. (As
discussed previously, EPA's proposed rule already included a range of costs
to reflect uncertainty about how many states would adopt indoor air
programs.)

EPA May Have Underestimated the Costs to Address the Risks from Radon
Emitted during Aeration

EPA's cost estimates may not adequately account for the additional costs to
address the health risks from radon that would be emitted into outdoor air
as it is removed from drinking water through aeration. These risks may be of
particular concern where water treatment facilities that remove radon in
water and vent it into the air would be located close to homes and schools
and other population centers. In these cases, water systems may face public
relations problems due to residents' concerns. Credible information on the
estimated risks from such emissions could help water systems address such
concerns.

However, we found that EPA's analysis of these health risks has some
limitations that tend to underestimate the risks and reduce its credibility.
The limitations stem from the use of outdated health risk data and the
outdated air quality model EPA used to develop its estimate of risk from
emissions for the proposed radon rule. Specifically, EPA's estimates of the
risks from radon emitted during aeration inadvertently did not incorporate
updated information from the National Academy of Sciences. EPA officials
said that updated information from the academy indicated that the health
risk from exposure to radon was about 2.5 times higher than its previous
estimate. While EPA incorporated this revision in its estimates of the risks
from radon in drinking water, it erred by not doing so for its estimates of
the risks associated with radon emissions from aeration. Furthermore, EPA
used the deficient health risk data in an outdated 1988 model that the
agency acknowledged has substantial limitations, even though the agency had
newer models available. Specifically, EPA's documentation of the 1988 air
quality model states that the resulting estimates of human health risks
associated with radon emitted during aeration are (1) "preliminary in nature
and should be used with caution," and (2) "do not account for the additive
impact of emissions from plants located close to one another."

A 1999 study commissioned by a water district in California suggests that
EPA's 1988 model may understate these health risks. Specifically, the study
includes a comparison of the estimated health risks associated with using
aeration technologies to remove radon from the district's water first using
the 1988 model, and then using an updated EPA model. In this site-specific
analysis, the risk estimate developed from the updated model was five times
higher than the estimate developed using EPA's 1988 model.

In discussing this issue, EPA officials told us that they may update the
final rule's estimates of the health risks from radon emissions by
incorporating the academy's updated risk information and using updated air
quality models. However, the officials said that taking these steps would
not substantially change the overall risk estimates for exposures to radon
emissions from water treatment shown in the proposed rule. They also said
that the health risks associated with emissions of radon from treatment
plants would still be negligible compared to the risks of radon in water.
While the officials agreed that higher risk estimates could lead to higher
national costs to implement the rule, they believe the cost increases would
be insignificant. However, without an updated risk estimate using current
data and models, water systems may have difficulty addressing concerns their
customers may raise about the risks-actual or perceived- of radon treatment.
These concerns could increase costs if, for example, customers demand more
expensive technologies to reduce risks associated with treating water for
radon.25 Along these lines, we note that in commenting on the proposed rule,
the American Water Works Association pointed out that "the perception of
risk is often as important as the actual risks when siting any industrial
process, including water treatment systems. The mitigation of such concerns
of the citizenry can result in substantially increased costs...."

EPA's Cost Analysis Indicates Inadequate Quality Assurance

EPA's cost analysis for the proposed radon rule does not consistently comply
with EPA guidance calling for clarity and transparency of all aspects of the
analyses and inclusion of all important data sources, key assumptions, and
their justifications. These presentation deficiencies, as well as analytical
errors, occurred despite the agency's quality assurance

25One alternative technology, granular activated carbon (GAC), is much more
costly than aeration for all but the very smallest water systems. This
technology does not vent radon into the air during treatment but collects
the radon in filters.

Documents Were Not Uniformly Clear, Accurate, and Consistent

process. As a result, the credibility of the analysis was reduced and the
ability of affected parties to provide informed comments was hampered.

EPA's guidelines for preparing economic analyses state that EPA should
strive for maximum clarity and transparency of all aspects of the
assessments and clearly describe all important data sources, key
assumptions, and their justifications. The guidelines also say that the
presentation should highlight the key elements that dominate modeling
frameworks and its results and address uncertainties by identifying ranges
for inputs and results. We found a number of instances in which EPA did not
comply with its guidance. For example

* EPA's proposed rule and the accompanying regulatory impact analysis
contain inconsistent and inaccurate information about an important cost
factor-the number of sites at which each water system would have to monitor
radon concentrations and potentially install treatment technologies. In both
documents, EPA states that it assumed treatment would occur at each well
with a radon level higher than the applicable drinking water standard.
Tables presenting numbers of wells accompany these statements. However, EPA
actually assumed that testing and treatment would occur at sites known as
"entry points" where water from multiple wells is often combined. Because
there are many fewer entry points than wells, assuming treatment at each
entry point instead of each well leads to a much lower national cost
estimate. Comments on the proposed rule show that affected parties were
confused about how EPA could have arrived at its national cost estimate by
assuming treatment at each well, as EPA incorrectly indicated it had done.

* EPA's documents are not clear about whether implementing an indoor air
program in lieu of treating radon in drinking water is a choice or a
requirement for small local water systems-because the documents contain
conflicting statements. Both the proposed rule and the economic analysis
state that small water systems "must" implement an indoor air program if
there is no state indoor air program that meets the rule's requirements.
However, both documents also state that small water systems may "choose" to
either implement an indoor air program or comply with the more stringent
water standard. In fact, under the proposed rule, implementing an indoor air
program would be a choice- not a requirement-for small local water systems,
just as it is for large water systems. These conflicting statements confused
some affected parties about the proposed rule's requirements and the
associated costs.

* Both the proposed rule and the economic analysis lack transparency and
clarity about who would incur the costs to test and, if necessary, treat

indoor air for radon. The documents incorrectly indicate that state and
community water systems would pay to test indoor air and reduce radon levels
in homes under the proposed rule's alternative standard. However, EPA
officials told us that, in fact, households would be expected to bear most
of these costs and that EPA intends to clarify this in the final rule. By
misstating who would bear a substantial portion of the costs of the proposed
rule, EPA did not disclose that the success of this rule depends on the
ability of states and water systems to persuade thousands of households to
spend a total of about $41 million each year to reduce their health risks
from exposure to radon in indoor air. As discussed previously, EPA also
erred in estimating these annual costs in the proposed rule, reporting them
as $3.9 million.

* EPA's documents do not disclose the agency's assumptions regarding how
many systems would need to remove 50 percent, 80 percent, or 99 percent of
the radon in their water under the proposed rule. Because costs increase
with the level of radon removed, the number of systems assumed to fall into
each of the removal categories is a key determinant of the total national
cost of the proposed rule. EPA's omission of these assumptions prevented
stakeholders from readily assessing the reasonableness of EPA's cost
estimate.

* EPA's proposed rule is unclear about whether EPA accounted for the
additional costs to reduce the risks from radon that would be emitted into
outdoor air as it is removed from drinking water through aeration. For
example, in the proposed rule, EPA first states that its cost estimates do
not include the additional costs associated with reducing the risks from
such radon emissions. Yet later in the proposed rule, EPA states that its
cost estimates do include these additional costs. EPA's confusing
presentation contributed to affected parties' concerns that EPA's estimates
had not accounted for any of these additional costs. In fact, EPA did
include some costs for reducing the risks from radon that would be emitted
through aeration,26 as shown by documentation that EPA provided to us.
However, as noted previously, EPA may have understated these costs because
it underestimated the health risks associated with radon emissions generated
by aeration equipment.

26Specifically, EPA's analysis assumed that (1) no systems serving
populations smaller than 3,301 would face additional costs to address
emissions from water treatment because EPA believes that these small systems
are generally located in rural areas where emissions would not be a concern
and (2) 15 percent of systems that serve populations larger than 3,300 and
that install aeration treatment would incur additional costs to address
emissions, at an average annual cost of $39,000 per system. This cost
represents a 35-percent increase to these systems' average costs for
treating and monitoring their water for radon.

EPA's Quality Assurance Process Did Not Identify Analytical Errors and
Documentation Flaws in the Cost Analysis

We also found that stakeholders questioned certain cost estimates, in part,
because of the lack of clarity and transparency about cost elements in the
documents supporting the rule. For example, as discussed previously, EPA
cited two sets of cost estimates for clearwells. In addition, regarding the
costs for treating iron and manganese, EPA said in its proposed rule that it
included some costs for this task, but said in its regulatory impact
analysis that it excluded them.

Some of the flaws we identified in the cost analysis for the radon rule are
similar to those that we previously identified in EPA's economic analyses
for other rules. For example, in 1997 we reported that in several of the
analyses we reviewed, EPA did not describe certain key assumptions used to
estimate costs and benefits.27

The analytical errors and documentation flaws that we identified in EPA's
proposed rule and cost analysis were not detected or corrected by the
agency's quality assurance process. EPA officials said that the Office of
Water has a quality management plan that guides its quality assurance and
control activities, specifying which types of such activities are necessary
and the various procedures for conducting quality reviews. The quality
assurance elements include internal peer review, external peer review,
external agency review, and stakeholder meetings, among other measures.
According to EPA officials, the primary quality assurance elements that EPA
relied on for the proposed radon rule were

* the National Academy of Sciences' assessments of the health risks from
radon;

* recommendations from an expert panel on the costs water systems would
incur for construction, engineering, and labor related to installing radon
water treatment equipment;

* recommendations from EPA's National Drinking Water Advisory Council on
issues related to the analysis of costs and benefits of drinking water
regulations in general;

* recommendations from EPA's Science Advisory Board valuing the benefits of
cancer cases avoided in environmental regulations in general;

* comments received at three national meetings with stakeholders; and

27U.S. General Accounting Office, Air Pollution: Information Contained in
EPA's Regulatory Impact Analyses Can be Made Clearer, GAO/RCED-97-38
(Washington, D.C.: Apr. 14, 1997).

* meetings with the American Water Works Association to examine technical
components of the rulemaking.

In addition, EPA published its health risk reduction and cost analysis for
public comment more than 6 months prior to issuing a proposed rule on radon.
The expert groups and the public comments provided EPA with valuable
information that it used to improve key components of the proposed rule
during its development. However, these reviews did not provide a detailed,
comprehensive review of the completed cost analysis supporting the proposed
radon rule.

EPA's economic analyses do not undergo external peer review nor have they
typically undergone formal internal peer review by experts outside of the
program offices that prepared them. We have previously stated that important
economic analyses supporting regulations should receive peer review-the
critical evaluation of scientific and technical work products by independent
experts-to enhance the quality, credibility, and acceptability of both the
analyses and the associated agency decisions.28 Experts in economic analysis
have also noted the importance of peer review. For example, a diverse panel
of renowned economists recommended in a 1996 paper29 that peer review of
economic analyses be used for regulations with potentially large economic
impacts. In addition, the Presidential/Congressional Commission on Risk
Assessment and Risk Management reported that agencies did not give enough
attention to the quality and interpretation of economic analyses and
recommended that these analyses receive adequate peer review.

EPA has recognized the need to improve the quality of its economic analyses,
and in August 2001, the EPA Administrator approved the implementation of the
recommendations of an agency work group to, among other things, require
internal review of EPA's major rules and the economic analyses supporting
them. As part of this effort, EPA's National Center for Environmental
Economics has begun to develop a process to systematically review economic
analyses for the agency's major rules. The

28U.S. General Accounting Office, Regulatory Reform: Comments on S. 981-The
Regulatory Improvement Act of 1997, GAO/T-GGD/RCED-97-250
(Washington, D.C.: Sept. 12, 1997).

29Arrow, Kenneth J. et al., Benefit-Cost Analysis in Environmental, Health,
and Safety
Regulation: A Statement of Principles (the American Enterprise Institute,
the Annapolis
Center, and Resources for the Future, 1996).

center has developed a draft "Economic Regulatory Review Summary and
Critique," dated December 12, 2001, that includes a lengthy checklist the
center may use to review key data, assumptions, and modeling techniques used
in the analyses and the transparency and clarity of the economic analyses.
The center's reviews of the economic analyses prepared by EPA's program
offices could provide the agency with meaningful internal peer review of its
economic analyses.

Implementation of the regulatory work group's recommendation that the
economic analyses supporting major rules undergo internal peer review has
the potential to improve the quality of EPA's rules and eliminate some of
the errors and other limitations we identified in the proposed radon rule.
EPA's National Center for Environmental Economics has already conducted
several internal peer reviews of economic analyses supporting major rules as
case studies, and EPA has found that such reviews can produce meaningful
results. For example, the center's director told us that one of the peer
reviews served as a forum for airing differences of opinion among program
office and legal staff on whether and how to account for pre-existing
subsidies. As a result of discussions of this issue during the internal peer
review, the proper accounting method was selected. If the subsidies had not
been properly recognized and accounted for, the costs of the rule would have
been understated by about $700 million.

The draft checklist that the center is developing to peer review economic
analyses contains many questions divided into nine sections: regulation
description, baseline, benefits, costs, economic impact analysis, equity
assessment, discounting, sensitivity analysis, and summary and critique of
the entire economic analysis. The sections on costs and the summary and
critique of the economic analysis include the following:

* Was the proper modeling approach used to assess the economic costs?

* Were relevant and high quality data sources used?

* Did the analysis address all significant economic costs?

* Are all of the data sources and assumptions clearly described?

* Is the analysis generally clear and transparent?

The questions in the draft are reasonable and specifically address some of
the problems we identified with the proposed radon rule, such as the lack of
clarity and transparency. However, the questions do not incorporate basic
quality assurance checks for accuracy and consistency that could better
ensure that the agency's economic analyses do not contain errors such as
inappropriately amortized costs. The errors that EPA made in conducting and
presenting its economic analysis supporting the radon rule

Conclusions

could, in most cases, have been easily avoided with basic quality assurance
checks for accuracy and consistency.

While EPA's proposed radon rule was issued before the agency started to
implement its new regulatory review process, the director of EPA's National
Center for Environmental Economics told us that the drinking water rule on
radon will be subject to the review process before the rule is finalized.

In developing its proposed rule on radon in drinking water, EPA sought and
was generally responsive to advice from experts and stakeholders, which
strengthened important aspects of the cost analysis supporting the proposed
rule. However, because of the limitations we identified in EPA's cost
analysis, the agency did not provide policymakers and stakeholders with
complete and reliable estimates of the expected compliance costs of the
proposed rule and who would bear them. Identifying the regulatory costs that
water systems are expected to incur is particularly important in light of
the anticipated financial demands on water systems to enhance security and
comply with other pending drinking water regulations. It is also important
to accurately estimate the costs that households would have to incur-on a
voluntary basis-to remove radon from the indoor air in their homes to reduce
radon health risks as anticipated by the rule. The limitations in EPA's cost
analysis and presentation also hampered the ability of interested parties
and the public to provide informed comments to EPA. Whether addressing these
limitations would change EPA's conclusion that the rule is economically
justified is not known given that EPA will also be revising its estimate of
the benefits in the final rule, for example, to respond to recommendations
from the Science Advisory Board that estimates of benefits be discounted to
present value.

EPA appears to be moving in the right direction by requiring internal peer
reviews of the economic analyses supporting its major rules and starting to
develop standard procedures for these reviews. The internal peer reviews-if
properly and routinely conducted-should improve the credibility and
usefulness of the agency's economic analyses and improve its regulatory
actions overall. While we continue to believe that some economic analyses
may also warrant external peer review, there are signs that an internal peer
review process could produce meaningful results at EPA. In our view, the
agency's efforts to establish standard procedures for the reviews of
economic analyses could help the agency ensure that its reviews are thorough
and consistent. Yet it is not clear whether EPA's review procedures, as
presently drafted, would be sufficiently rigorous

and detailed to identify some of the errors we identified, such as the
accounting error that incorrectly amortized an annual cost over a 20-year
period. However, EPA still has the opportunity to build in such procedures.

Recommendations for To improve the credibility and usefulness of its
economic analysis for the final drinking water rule on radon, we recommend
that the administrator,

Executive Action

*

*

* *

*

Agency Comments

EPA, require the Office of Water to

correct its cost estimates for testing for and treating radon in indoor air
and disclose that homeowners are expected to bear these costs,
correct its estimates of states' and water systems' costs for administration
of indoor air programs,
include mixed water systems in its economic analysis,
revise its economic analysis to include less optimistic assumptions about
how many water systems will use indoor air programs to comply with the
rule, and
revise its estimate of the risks from radon emitted during water treatment
by incorporating the National Academy of Sciences' increased estimate of
these risks, and by using the agency's current air quality models, and
assess the extent to which the revised risk estimate would change costs.

To better ensure the quality of economic analyses for the radon rule and
other major rules prepared by EPA, we also recommend that the
administrator, EPA, require the agency to expeditiously implement
standard procedures for conducting internal peer reviews of its economic
analyses. These procedures should include quality assurance measures to
identify errors in calculations; check the reasonableness of assumptions
and methodologies; and ensure that the documentation of the analyses is
clear, transparent, accurate, and complete.

We provided EPA with a draft of this report for its review and comment.
In response, officials from the Offices of Water; Air and Radiation; and
Policy, Economics, and Innovation generally agreed with our findings and
recommendations. The officials provided some technical and editorial
suggestions that we have incorporated into the report, as appropriate.
However, while agreeing that the proposed rule contained the errors and
other inaccuracies we identified, the Office of Water did not agree that an
effect of these errors was reduced credibility of EPA's cost analysis. While
our report identifies a number of strengths of EPA's cost analysis, these
strengths do not offset or negate the effects of the errors and

Scope and Methodology

misstatements we found. We continue to believe that, collectively, the
limitations we identified reduced both the credibility and usefulness of the
cost analysis for the radon rule. For example, we believe that the
credibility of EPA's estimates was reduced by an analytic error and a
presentation error that EPA made related to an important cost component-the
cost of testing and treating indoor air for radon. The analytic error
involved EPA estimating annual costs of $4 million, when the correct
estimate is about $41 million. The presentation error involved EPA
indicating that states and local water systems would bear these costs, when
such costs would actually be borne largely by individual households. We
believe that correcting these and other errors would, in fact, improve the
credibility and usefulness of the analysis to policymakers and stakeholders.

To assess the strengths and limitations of EPA's cost estimate for the
November 1999 proposed radon rule, we reviewed the two primary EPA documents
describing the agency's cost analysis; namely, the proposed rule and the
agency's economic analysis supporting the rule (the regulatory impact
analysis). In addition, we reviewed key EPA documents that the agency used
to support its economic analysis, including

Technologies and Costs for the Removal of Radon from Drinking Water;
Methods, Occurrence, and Monitoring Document for Radon in Drinking Water;
Technical Support Document for the 1992 Citizen's Guide to Radon; and the
October 1999 supporting statement for information collection request for
radon. We reviewed case studies identifying the costs associated with
installing and maintaining aeration equipment, including Critical Assessment
of Radon Removal Systems for Drinking Water Supplies published by the
American Water Works Association Research Foundation and the American Water
Works Association in 1998. In addition, we met with EPA officials
responsible for the proposed rule and the economic analysis to obtain
information about key assumptions and methodologies. These officials
provided us with internal documents, such as costing models and spreadsheets
that supported the analysis, which we also reviewed. We reviewed stakeholder
comments made on the proposed rule, including those by the American Water
Works Association, the National Rural Water Association, and the Association
of State Drinking Water Administrators. We met with representatives of the
American Water Works Association and interviewed water treatment
professionals who have experience with radon removal.

Because the scope of our review was limited to assessing EPA's cost
estimate, we reported on, but did not evaluate, EPA's estimates of the

expected benefits of the proposed rule. As also agreed with your offices, we
reviewed the assumptions EPA used in its cost models to generate its water
treatment estimates, but we did not validate the costing models or the data
EPA used in developing its cost estimates. For several of the analytical
limitations we identified, we developed estimates of the change in EPA's
estimated costs if EPA were to correct its analysis (see app. I). We
developed these estimates based on information from EPA's primary documents
and additional internal documents that we obtained from EPA officials. We
did not have a basis to estimate increased costs that water systems might
incur if EPA updated its estimate of the health risks from radon that would
be emitted into outdoor air as it is removed from drinking water through
aeration. We conducted our work from May 2001 through January 2002 in
accordance with generally accepted government auditing standards.

We will send copies of this report to the administrator, EPA, and make
copies available to others who request them. If you or your staff have
questions about this report, please call me on (202) 512-3841.

John B. Stephenson Director, Natural Resources and Environment

Appendix I: Adjustments to EPA's Best Estimates of Total National Costs to
Address Limitations in EPA's Economic Analysis

                            Dollars in millions

                              Cost components

EPA's best estimate

EPA's best estimate corrected for two accounting

errors

 EPA's best estimate corrected for two accounting errors and including mixed
                                                                   systemsa

EPA's best estimate corrected for two accounting

errors; including mixed systems;

and adjusting overly optimistic assumption from

90% of systems to 75% of systems

EPA's best estimate corrected for two accounting

errors; including mixed systems;

and adjusting overly optimistic assumption from

90% of systems to 50% of systems Water Indoor air

             Capital                 $23.1     $23.1     $26.0      $44.6         $75.6
          Operations and             15.5       15.5      17.4      29.6
           maintenance
            Monitoring               14.1       14.1      15.9      15.9
              System                  6.1       6.1       6.3        6.3
          administration
       State administration           2.5       2.5       2.5        2.5
             Subtotal                $61.3     $61.3     $68.1      $98.8        $150.1

Testing and 3.9 41.2 51.3 47.2 treatmentb

                  System                      45.1       18.6        19.1       15.9
            administration of
           systems' indoor air
                 programs

State oversight of 7.8 3.3 3.4 2.9EURsystems' indoor airEURprogramsEUR

                 State administration of states' indoor air

2.9 2.9 2.9 2.9

c

programs

                   Subtotal $59.8 $66.1 $76.7 $68.9 $56.0

               Total costs $121.1 $127.4 $144.8 $167.8 $206.1

Note: Line items may not sum to totals due to rounding.

aOur estimate of the costs for mixed systems is based on an estimate that
EPA developed and we reviewed. The estimate includes 1,074 mixed water
systems receiving more than 50 percent of their water from groundwater
sources. However, it does not include the costs for three systems serving
more than 1,000,000 customers that receive some of their water from
groundwater sources and that EPA believes would incur costs to comply with
the radon rule. We did not develop an estimate for these costs, which should
also be included in EPA's cost estimate for the final radon rule.

bEPA assumes that most of these costs will be borne by individual
homeowners.

cEPA's best estimate assumes 25 states will have indoor air programs, but
its estimate of state administration costs is based on costs for 23 states.
According to EPA, the estimate in the final rule will include costs for 25
states.

Source: GAO's analysis of EPA data.

Appendix II: GAO Contact and Staff Acknowledgments

GAO Contact Christine Fishkin (202) 512-6895

Staff Other key contributors to this report include David Goldstein, Timothy
Guinane, Patricia Manthe, Cynthia Norris, and Amy Webbink.

Acknowledgments

GAO's Mission

Obtaining Copies of GAO Reports and Testimony

The General Accounting Office, the investigative arm of Congress, exists to
support Congress in meeting its constitutional responsibilities and to help
improve the performance and accountability of the federal government for the
American people. GAO examines the use of public funds; evaluates federal
programs and policies; and provides analyses, recommendations, and other
assistance to help Congress make informed oversight, policy, and funding
decisions. GAO's commitment to good government is reflected in its core
values of accountability, integrity, and reliability.

The fastest and easiest way to obtain copies of GAO documents is through the
Internet. GAO's Web site (www.gao.gov) contains abstracts and full-text
files of current reports and testimony and an expanding archive of older
products. The Web site features a search engine to help you locate documents
using key words and phrases. You can print these documents in their
entirety, including charts and other graphics.

Each day, GAO issues a list of newly released reports, testimony, and
correspondence. GAO posts this list, known as "Today's Reports," on its Web
site daily. The list contains links to the full-text document files. To have
GAO e-mail this list to you every afternoon, go to www.gao.gov and select
"Subscribe to daily e-mail alert for newly released products" under the GAO
Reports heading.

Order by Mail or Phone The first copy of each printed report is free.
Additional copies are $2 each. A check or money order should be made out to
the Superintendent of Documents. GAO also accepts VISA and Mastercard.
Orders for 100 or more copies mailed to a single address are discounted 25
percent. Orders should be sent to:

U.S. General Accounting Office P.O. Box 37050 Washington, D.C. 20013

To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

Visit GAO's Document GAO Building

Distribution  Center  Room 1100,  700 4th  Street, NW  (corner of 4th  and G
Streets, NW) Washington, D.C. 20013

To Report Fraud, Contact: Web site: www.gao.gov/fraudnet/fraudnet.htm,

Waste, and Abuse in E-mail: [email protected], or

Federal Programs 1-800-424-5454 or (202) 512-7470 (automated answering
system).

Jeff   Nelligan,  Managing   Director,   [email protected]   (202)  512-4800

Public Affairs U.S.  General Accounting Office, 441 G. Street NW, Room 7149,
Washington, D.C. 20548
*** End of document. ***