GUIDELINES FOR REGULATORY

IMPACT ANALYSIS
2016

Office of the Assistant Secretary for Planning and Evaluation

U.S. Department of Health and Human Services
ACKNOWLEDGEMENTS
These Guidelines for Regulatory Impact Analysis were prepared for the U.S. Department of Health and Human
Services (HHS) Analytics Team, under the leadership of Amber Jessup (Office of the Assistant Secretary for
Planning and Evaluation). The primary authors were Lisa A. Robinson and James K. Hammitt (Harvard University
Center for Risk Analysis and Center for Health Decision Science) and Jennifer R. Baxter (Industrial Economics,
Incorporated, IEc). The work was performed between 2013 and 2016 under subcontract to IEc and Mathematica
Policy Research; Ms. Baxter was the IEc Project Leader. Dr. Hammitt’s work was also supported by an HHS
Intergovernmental Personnel Act agreement. The authors were assisted by IEc staff including Lindsay Ludwig,
who helped develop the initial drafts of several sections; Margaret Black, who provided additional editorial
support and helped draft the related primer; and Michael Welsh, who helped develop the index and glossary.
The draft Guidelines were independently peer reviewed by Joseph Aldy (Harvard Kennedy School) and David
Weimer (La Follette School of Public Affairs, University of Wisconsin-Madison). HHS staff also provided
substantial advice and comments, including Evell Barco, Laina Bush, Caroline Cochran, Daniel Converse, Walt
Francis, Sherry Glied, Scott Grosse, Kevin Haninger, Amber Jessup, Daniel Lawver, Nellie Lew, Clark Nardinelli,
John Rigg, Kakoli Roy, C'Reda Weeden, Daniel Wilmoth, Nancy Zhang, and David Zorn, as well as other Analytics
Team members and Centers for Disease Control and Prevention staff.
This guidance represents the current thinking of the Department of Health and Human Services (HHS) on the
conduct of regulatory impact analysis. It does not establish any requirements for any person and is not binding
on HHS, any HHS agencies or the public. You can use an alternative approach if it satisfies the requirements of
the applicable Executive Orders and regulations. To discuss an alternative approach, contact the Office of the
Assistant Secretary for Planning and Evaluation.

i
TABLE OF CONTENTS
Acknowledgements .....................................................................................................................................................i
Table of Contents ....................................................................................................................................................... ii
Acronyms ................................................................................................................................................................... iii
Chapter 1: Introduction ..............................................................................................................................................1
1.1 What Is Regulatory Impact Analysis? ................................................................................................................1
1.2 What Are the Benefits and Costs of Conducting an RIA? .................................................................................2
1.3 When Is an RIA Required? ................................................................................................................................2
1.4 What Are the Basic Components? ....................................................................................................................3
Chapter 2: Frame the Analysis....................................................................................................................................5
2.1 Explain the Need for Action and Identify Alternatives .....................................................................................5
2.2 Define the “Without Regulation” Baseline .......................................................................................................6
2.3 Describe the Consequences of Each Policy Alternative ....................................................................................7
2.4 Use Screening to Focus the Analysis .................................................................................................................8
Chapter 3: Assess Benefits ...................................................................................................................................... 10
3.1 Basic Concepts ............................................................................................................................................... 10
3.2 Valuing Mortality Risk Reductions ................................................................................................................. 13
3.3 Valuing Morbidity Risk Reductions ................................................................................................................ 18
Chapter 4: Assess Costs ........................................................................................................................................... 23
4.1 Basic Concepts and Approach ........................................................................................................................ 23
4.2 Assessing Compliance and Government Implementation Costs ................................................................... 25
4.3 Estimating Market-Level Impacts .................................................................................................................. 35
Chapter 5: Account for Timing ................................................................................................................................ 36
5.1 Basic Concepts and Approach ........................................................................................................................ 36
5.2 Adjusting for Inflation .................................................................................................................................... 37
5.3 Determining Present Values .......................................................................................................................... 38
5.4 Annualizing Impacts ....................................................................................................................................... 41
Chapter 6: Address Uncertainty and Nonquantifiable Effects ................................................................................ 43
6.1 Characterizing Uncertainty in Quantified Effects .......................................................................................... 43
6.2 Characterizing Nonquantified Effects ............................................................................................................ 47
Chapter 7: Conduct Distributional and Other Supplementary Analyses ................................................................. 52
7.1 Assess Distribution across Demographic Groups .......................................................................................... 52
7.2 Conduct Supplementary Analyses ................................................................................................................. 55
7.3 Address International Effects......................................................................................................................... 58
Chapter 8: Communicate the Approach and Results .............................................................................................. 60
8.1 Describe the Analysis and Results.................................................................................................................. 60
8.2 Provide Summary Tables and Figures ............................................................................................................ 61
Chapter 9: Conduct Retrospective Analysis ............................................................................................................ 64
9.1 Basic Concepts ............................................................................................................................................... 64
9.2 General Approach .......................................................................................................................................... 65
Appendix A: Agency Checklist: Regulatory Impact Analysis (OMB 2010) .............................................................. A-1
Appendix B: Consumer and Producer Surplus........................................................................................................ B-1
Appendix C: Methods for Estimating QALYs .......................................................................................................... C-1
References .............................................................................................................................................. References-1
Index ................................................................................................................................................................ Index-1
Glossary .......................................................................................................................................................Glossary-1

ii
Acronyms
ASPE Assistant Secretary for Planning and Evaluation
BLS U.S. Bureau of Labor Statistics
CBO Congressional Budget Office
CDC Centers for Disease Control and Prevention
CEA cost-effectiveness analysis
Census U.S. Census Bureau
CPI Consumer Price Index
CPS Current Population Survey
CRA Congressional Review Act
DOT U.S. Department of Transportation
ECEC Employer Costs for Employee Compensation
ECI Employer Cost Index
EQ-5D EuroQol-5 Dimensions
FDA Food and Drug Administration
FICA Federal Insurance Contributions Act
FRFA Final Regulatory Flexibility Analysis
G&A general and administrative
GAO Government Accountability Office
GDP gross domestic product
GSA U.S. General Services Administration
HHS U.S. Department of Health and Human Services
HRQL health-related quality of life
HUI Health Utilities Index
ICR Information Collection Request
IRFA Initial Regulatory Flexibility Analysis
IRS Internal Revenue Service
NCS National Compensation Survey
NHTSA National Traffic Highway Safety Administration
O&M operations and maintenance
OES Occupational Employment Statistics
OMB U.S. Office of Management and Budget
PRA Paperwork Reduction Act
QALY quality-adjusted life year
QCEW Quarterly Census of Employment and Wages
QWB Quality of Well-Being
RFA Regulatory Flexibility Act
RIA regulatory impact analysis
SBA Small Business Administration
SBREFA Small Business Regulatory Enforcement Fairness Act
SOP standard operating procedure
UMRA Unfunded Mandates Reform Act

iii
VSL value per statistical life
VSLY value per statistical life year
WTA willingness to accept compensation
WTP willingness to pay

iv
Chapter 1
Introduction
Executive Orders 12866 and 13563 (Clinton 1993, Obama 2011) call for a regulatory system that protects “public
health, welfare, safety, and our environment while promoting economic growth, innovation, competitiveness,
and job creation.” To achieve these goals, the Department of Health and Human Services (HHS) analyzes the
benefits, costs, and other impacts of significant proposed and final rulemakings, consistent with the
requirements of the executive orders.
In the HHS 2011 Plan for Retrospective Review of Existing Rules, the Assistant Secretary for Planning and
Evaluation (ASPE) was asked to establish an agency-wide Analytics Team to provide recommendations for
strengthening regulatory analysis, leveraging the existing expertise of economists and analysts from throughout
the Department’s operating divisions.1 The Analytics Team investigated current challenges and determined that
guidance was needed to address common difficulties and to ensure consistent treatment across agencies. To
meet that need, the Department developed these Guidelines for Regulatory Impact Analysis to assist its agencies
in conducting economic analyses that meet the goals of the executive orders. This chapter briefly introduces
related requirements and the contents of these Guidelines.

1.1 WHAT IS REGULATORY IMPACT ANALYSIS?

A regulatory impact analysis (RIA) reflects a well-established and widely-used approach for collecting,
organizing, and analyzing data on the impacts of policy options, to promote evidence-based decision-making. It
provides an objective, unbiased assessment that is an essential component of policy development, considering
both quantifiable and unquantifiable impacts. Along with information on legal requirements, general policy
goals, the distribution of the impacts, and other concerns, it forms the basis of the ultimate policy decision.
The RIA describes the effects of the regulation rather than
advocating a particular approach. The arguments supporting WHY PREPARE AN RIA?
the agency’s decision are provided separately in the preamble
to the Federal Register notice for the proposed and final RIAs provide objective information and
regulation. The core of the RIA is an assessment of the benefits analysis that is essential for evidence-based
and costs of regulatory and other policy options in comparison decision-making. They include a benefit-cost
to a “without regulation” (or “no action”) baseline. In addition, analysis as well as other analyses mandated
the RIA includes supplementary analyses that respond to by various statutes and executive orders.
various statutory and administrative requirements.
The RIA framework is described in general terms in Executive Orders 12866 and 13563 (Clinton 1993, Obama
2011).2 More specific guidance and oversight is provided by the Office of Information and Regulatory Affairs
within the U.S. Office of Management and Budget (OMB), which is part of the Executive Office of the President.
OMB reviews both the regulation and the supporting analysis prior to promulgation.3 Its primary analytic

1
We provide links to those documents that are freely available on the internet in the reference list. Where possible, we link to the webpage that features the
document rather than to the document itself, so that readers can check for updates.
2
These requirements apply only to the extent allowable by law.
3
Under the Congressional Review Act (CRA), agencies must also submit final rules and supporting analyses to the Government Accountability Office (GAO) for
congressional review prior to promulgation. This submission must indicate whether the rule is “major” as defined under the CRA (5 USC §804(2)): “‘major
rule’ means any rule that the Administrator of the Office of Information and Regulatory Affairs of the Office of Management and Budget finds has resulted in
or is likely to result in — (A) an annual effect on the economy of $100,000,000 or more; (B) a major increase in costs or prices for consumers, individual
industries, Federal, State, or local government agencies, or geographic regions; or (C) significant adverse effects on competition, employment, investment,
productivity, innovation, or on the ability of United States-based enterprises to compete with foreign-based enterprises in domestic and export markets. The
term does not include any rule promulgated under the Telecommunications Act of 1996 and the amendments made by that Act.” More information is
available on the GAO website ([Link]

1
guidance is provided in Circular A-4 (2003); it summarizes related requirements in a checklist for agencies
(2010), a compilation of frequently-asked questions (2011a), and a primer (2011b). The OMB checklist is
replicated in Appendix A of this document. Examples of RIAs completed by HHS and other agencies can be found
by searching [Link]; however, analysts should be aware that many of the HHS analyses were
completed prior to issuance of these Guidelines.4
In addition to the assessment of the benefits and costs, the RIA may include supplementary analyses that
address the following, as relevant.
 the distribution of the impacts;
 the Unfunded Mandates Reform Act;
 the Regulatory Flexibility Act and Small Business Regulatory Enforcement Fairness Act;
 Executive Order 13132, “Federalism;”
 Section 1102(b) of the Social Security Act, small rural hospitals; and,
 the Paperwork Reduction Act.
More information on these requirements, as well as on the conduct of the benefit-cost analysis, is provided in
the subsequent chapters of this guidance.

1.2 WHAT ARE THE BENEFITS AND COSTS OF CONDUCTING AN RIA?

The most important goals of the RIA are (1) to indicate whether Federal regulation is necessary and justified,
and, if so, (2) to identify the regulatory option that is most economically efficient, providing the largest net
benefits to society. A well-conducted RIA has numerous additional benefits. It develops the evidence to support
well-informed decision-making and supplies a record of the data, assumptions, and analyses considered –
providing a reasonable basis for rulemaking as required by the Administrative Procedures Act.
The RIA plays several other useful roles. For example, it:
 encourages comprehensive consideration of impacts;
 provides information on important regulatory outcomes expressed in physical and behavioral terms;
 estimates the economic value of the outcomes, based on the preferences of those who are affected;
 anticipates potential side effects, beneficial and adverse;
 supports consideration of non-quantifiable effects and uncertainty; and,
 aids decision-makers and stakeholders in clarifying areas of agreement and disagreement.
The costs of conducting RIAs include the need to devote staff and funding to preparing these assessments rather
than to other tasks. To ensure the efficient use of these resources, the analysis should be carefully tailored to
focus on providing the information that is most important for decision-making. Screening analysis, discussed in
the following chapter, is a useful tool for targeting efforts.

1.3 WHEN IS AN RIA REQUIRED?

An RIA is required for significant and economically significant regulatory actions as defined under Executive
Order 12866 (§3(d-f)) and Executive Order 13563. An economically significant regulatory action is one that:
 is likely to impose costs, benefits, or transfers of $100 million or more in any given year, or
 “adversely affect in a material way the economy, a sector of the economy, productivity, competition, jobs,
the environment, public health or safety, or State, local, or tribal governments or communities” (Clinton
1993, §3(f)(1)).

4
Many agencies also post their RIAs on their websites. For example, analyses completed by the Food and Drug Administration (FDA) can be found at:
[Link]

2
If a regulation is economically significant, then the analysis discussed in OMB Circular A-4 (and described in more
detail in these Guidelines) must be completed (Clinton 1993, §6(a)(3)(C)).
In addition, many other regulations are considered
“significant,” defined as those that: HOW DOES OMB INTERPRET THE $100 MILLION THRESHOLD?

 “[c]reate a serious inconsistency or otherwise An RIA is required for economically significant regulations.
interfere with an action taken or planned by In defining “economically significant,” OMB (2011a) states,
another agency; “The $100 million threshold applies to the impact of the
proposed or final regulation in any one year, and it includes
 [m]aterially alter the budgetary impact of
benefits, costs or transfers.” The word “or” is important:
entitlements, grants, user fees, or loan programs
the categories are considered separately, not summed, so
or the rights and obligations of recipients
$100 million in any of the three categories -- annual
thereof; or
benefits, or costs, or transfers -- is sufficient. For example, a
 [r]aise novel legal or policy issues arising out of
regulation with $75 million in benefits, $60 million in costs,
legal mandates, the President’s priorities, or the
and $40 million in transfers is not economically significant.
principles set forth in this Executive order”
An RIA is also required for regulations deemed to be
(Clinton 1993, §3(f)(2-4)).
significant for other reasons and is an essential element of
For regulatory actions that are significant, but not good regulatory practice.
economically significant, Executive Order 12866
requires:
 “a reasonably detailed description of the need for the regulatory action and an explanation of how the
regulatory action will meet that need,” and
 “[a]n assessment of the potential costs and benefits of the regulatory action” (Clinton 1993, §6(a)(3)(B)).

Agencies may wish to complete RIAs for regulations that are not defined as significant to improve the
foundation for decision-making and to demonstrate the rationale and basis for the action.

1.4 WHAT ARE THE BASIC COMPONENTS?

The remaining chapters of this guidance are organized around the major components of an RIA, as illustrated in
Figure 1.1.
FIGURE 1.1. MAJOR RIA COMPONENTS

3
  The first three steps are discussed in Chapter 2: Frame the Analysis.

 Steps 4A and 4B are described in detail in Chapter 3: Assess Benefits and Chapter 4: Assess Costs.

 Topics that affect the assessment of both benefits and costs are considered in Chapter 5: Account for
Timing and Chapter 6: Address Uncertainty and Nonquantifiable Effects.
 The analyses under step 5 are discussed in Chapter 7: Conduct Distributional and Other Supplementary
Analyses.
 The presentation of the results is considered in Chapter 8: Communicate the Approach and Results.

 The Guidelines conclude by turning from the discussion of ex ante (prospective) analysis to ex post analysis
in Chapter 9: Conduct Retrospective Analysis.
Supplementary information is provided in the appendices.

4
Chapter 2
Frame the Analysis
Conducting an RIA involves first defining the problem to be addressed, identifying the policies to be assessed,
exploring their potential consequences, and developing the approach for subsequent analytic work. This chapter
describes these steps, focusing on the benefit-cost analysis that forms the core of the RIA. As introduced in
Chapter 1 and discussed in more detail in Chapter 7, an RIA includes several supplementary analyses, to which
the principles discussed in this chapter also apply. These analyses should be initiated in the early stages of the
regulatory development process, to inform both internal agency deliberations and discussions with other
stakeholders.
Benefit-cost analysis is a well-established systematic framework, based on economic welfare theory, for
assessing and comparing the positive and negative impacts of policy options. It addresses the question of
whether those affected by the policy, in the aggregate, value the benefits they receive more than the costs they
incur. The distribution of the impacts (who receives the benefits and who bears the costs) is assessed separately
(see Chapter 7).
The goal of the benefit-cost analysis is to indicate how limited resources can be best allocated to maximize net
social welfare. Welfare is based on individual preferences, and money is used as a convenient and practical
numeraire (or measuring rod) that describes the extent to which individuals are willing, as a society, to reduce
their consumption of other goods and services to achieve the policy outcomes.
Conducting a benefit-cost analysis is often useful and informative even if the resulting summary measure – net
benefits (benefits minus costs, which may be positive or negative) – is not used as a decision-making criterion or
is only one of many factors considered.5 The data and analysis provide a wealth of information on possible
impacts, including many that often were not anticipated or predicted, and this information has important
implications for regulatory design and implementation. The analysis should be descriptive, providing unbiased
and objective information.
Framing the analysis involves defining what will be assessed and developing the general analytic approach. This
chapter describes related activities, including explaining the need for the action and identifying the alternatives
to be addressed, specifying the baseline, and determining the consequences of the regulation. It concludes by
describing how screening analysis can be used to target analytic resources.

2.1 EXPLAIN THE NEED FOR ACTION AND IDENTIFY ALTERNATIVES

Consistent with OMB Circular A-4, agencies must first describe the market failure or other social purpose that
leads to the need for regulatory action. They must also describe why action at the Federal level, rather than at
the State or local level, is necessary or desirable. Agencies must indicate the significance of the regulation, based
on the definitions in Executive Order 12866 that are replicated in the previous chapter.

5
The normative basis for using benefit-cost analysis in decision-making begins with the Pareto principle, which states that a policy is desirable if it makes at
least one person better off and no one worse off. While attractive in theory, few policies meet this criterion: most will harm (or impose costs on) at least a
few people. To address this limitation, variations were developed by Nicholas Kaldor and John Hicks. These variations state that a policy is desirable if it
makes the winners better off by an amount large enough to compensate the losers, and, alternatively, that it should be rejected if the losers could
compensate the winners to not pursue the policy. These criteria do not demand that actual compensation take place. They imply that a policy for which costs
exceed benefits should not be adopted and, if more than one policy provides positive net benefits, the one with the largest net benefits should be adopted.
This principle is rarely applied strictly, as regulatory and other policy decisions are based on several considerations in addition to the results of the benefit-
cost analysis.

5
Agencies must also consider a range of regulatory
and non-regulatory alternatives, regardless of HOW MANY ALTERNATIVES MUST BE ANALYZED?
whether the statute or other authorities prescribe
Agencies must justify the need for regulatory action and
the option they can ultimately implement.6 OMB
consider a range of policy alternatives. These alternatives
Circular A-4 lists the types of alternatives that
must, at minimum, include at least one that is more
should be considered, not all of which will be
stringent and one that is less stringent than the preferred
applicable to a particular regulation:
option; additional options should also be assessed.
 different choices defined by statute;
 different compliance dates;
 different enforcement methods;
 different degrees of stringency;
 different requirements for different sized firms;
 different requirements for different geographic areas;
 performance standards rather than design standards;
 market-oriented approaches rather than direct controls; and,
 informational measures rather than regulation.7
Considering a wide-range of options both helps inform agency decision-making and encourages public
comment. The versions of the analysis published to support the proposed and the final rule must include, at the
very least, comprehensive analysis of one option that is more stringent and one that is less stringent than the
preferred option; in total, more than three options should be assessed. These options should represent diverse
approaches to meeting the policy goals and should be sufficiently distinct for the analysis to differentiate among
them. In some cases, the statute or other legal constraints, or issues of technical feasibility, will limit the types of
alternatives considered; this should be explicitly noted in the RIA. However, an option does not need to be
legally permissible to be assessed.
Prior to promulgation, the analysis conducted to support the regulatory development process should consider a
substantially broader array of options, which may be subject to varying degrees of assessment depending on
their feasibility and likely impacts. These additional options also should be discussed in the RIA documentation
to encourage public review and comment.
Selecting alternatives for assessment is an iterative process. As analysts gain a better understanding of the
benefits and costs of the options, the alternatives to be included in the final RIA are likely to be altered and
refined. Screening analysis, discussed later, can be used to eliminate many alternatives from detailed
consideration. The rationale for excluding and including alternatives, and the alternatives excluded, should be
explicitly discussed when documenting the analysis.

2.2 DEFINE THE “WITHOUT REGULATION” BASELINE

Each regulatory and non-regulatory alternative must be compared to a “no new regulatory action” baseline that
reflects expected future conditions.8 The analysis should, at minimum, compare conditions with and without the
policy once the policy is fully implemented. This may occur several years from the present, given the time
needed for notice and comment as well as implementation. In many cases, benefits and costs that accrue over
the transition period may be significant and should be assessed. In some cases, there may be a significant time
lag between when costs are incurred and when benefits accrue or vice-versa. In such cases, the analysis should
cover the full time period between when the impacts first occur and when benefits and costs are expected to

6
RIAs also aid the agency in identifying ways in which the statute can be improved. OMB Circular A-4 notes: “You should also discuss the statutory
requirements that affect the selection of regulatory approaches. If legal constraints prevent the selection of a regulatory action that best satisfies the
philosophy and principles of Executive Order 12866, you should identify these constraints and estimate their opportunity cost.” (OMB 2003, p. 17)
7
Alternatives that provide information and disclosure are discussed in more detail in Sunstein (2010a).
8
If the regulation is required by statute, the baseline should reflect the absence of the statutory requirement.

6
achieve equilibrium.9 The RIA should generally consider benefits and costs that accrue over a 10 to 20 year time
period, unless the program is expected to end sooner.
Analysts should explore likely trends rather than
simply assuming that current conditions will WHAT IS THE APPROPRIATE TIMEFRAME FOR THE ANALYSIS?
continue. These projections should address future In theory, the timeframe for the analysis should begin when
economic and health conditions as well as other regulated entities or others begin to change their behavior in
factors that may affect the regulatory environment. response to the regulation (which may occur before or after
Where future conditions are uncertain and changes the effective date of the regulation) and end when the
in baseline assumptions significantly affect the impacts of the regulation cease. However, it is generally
analytic results, analysts should consider modeling difficult to reasonably forecast effects far into the future.
more than one baseline or testing the sensitivity of OMB suggests that if the proposed regulation has no
their results to key assumptions. predetermined sunset provision, the agency should use its
Any difference between the baseline and a policy best judgment about the foreseeable future. “For most
alternative may have both positive and negative agencies, a standard time period of analysis is 10 to 20 years,
consequences, and both should be considered. and rarely exceeds 50 years” (OMB 2011a).
Conversely, neither the costs nor the benefits of
changes predicted in the absence of the regulation
should be attributed to the rule. For example, if a change in food handling procedures is expected under the
baseline, the associated costs would not be counted as costs of the regulation. Similarly, the benefits of that
change would have materialized in the baseline and cannot be attributed to the regulation.
When developing the baseline, analysts should also consider who has “standing;” i.e., whose benefits and costs
should be counted. OMB Circular A-4 (2003) indicates that the analysis should focus on U.S. residents and
citizens. At times, determining standing raises difficult issues, such as how to address the preferences of those
engaged in illegal activities. When such issues arise, the analysts should explicitly discuss their treatment in the
RIA documentation.
If a regulation is likely to have impacts outside of the United States, these impacts should be assessed separately
(see Chapter 7). A related issue is whether to assess only the immediate or direct impacts of the regulations, or
to also account for second-order or indirect effects, which may affect different groups of people. Screening
analysis is a useful tool for determining whether these less immediate effects are significant enough that they
should be considered.

2.3 DESCRIBE THE CONSEQUENCES OF EACH POLICY ALTERNATIVE

One of the most difficult steps in conducting regulatory analysis is predicting responses to the policy options,
given an evolving baseline, complex regulatory requirements, data gaps, and the diversity of the individuals and
organizations affected. Regulatory requirements typically lead to a series of consequences (events and
outcomes). It is important to distinguish between the initial requirement (e.g., hospitals must report certain
adverse drug reactions); subsequent events (e.g., hospital staff change their prescribing behavior); the ultimate
outcome (e.g., greater health improvements for some patients in comparison to the baseline); and its evaluation
(e.g., the monetary value of the behavioral changes and the health improvements). Evidence must be used to
establish the causal link between these events and outcomes. Analysts often find it useful to map these
relationships as a decision tree (Raiffa 1968) or as a logic model (Centers for Disease Control and Prevention
(CDC) 2007, Sundra et al. 2003, Wholey et al. 2010), which can be updated as more is learned about likely
impacts.

9
For meaningful comparison, benefits and costs should be measured over the same time period. If the nature of the impacts is such that assessing some over
longer periods than others provides important information, the time period over which only some impacts are assessed should be reported separately when
summarizing the analysis to avoid misleading comparisons.

7
Whether a particular consequence is classified as a benefit or cost does not affect the estimated net benefits, as
long as the sign is correct (i.e., positive benefits and negative costs increase net benefits; negative benefits and
positive costs decrease net benefits).10 However, for clear communication, analysts should follow a consistent
approach. Impacts categorized as benefits should relate to the intended outcome of the regulation (e.g.,
improved health); impacts categorized as costs should relate to the investment or inputs needed to achieve
those outcomes (e.g., safety expenditures by industry). In this case, any negative effects that relate to the
intended outcomes (e.g., through substitution of less safe drugs or less healthy foods for those that are
regulated) would be combined with the benefit estimates, while any offsetting savings from regulatory
compliance (e.g., increased efficiency from automation of previously manual tasks) would be combined with the
cost estimates.
Understanding these consequences is an iterative WHEN SHOULD AN IMPACT BE CLASSIFIED AS A
process, as each step in the analysis often COST VERSUS A BENEFIT?
provides new insights. Initially, analysts should
describe the possible outcomes (“fewer cases of Costs are the inputs needed to implement the regulation
cardiac and respiratory disease,” “higher (e.g., industry expenditures to improve safety); benefits
production costs”) in as much detail as possible. are the intended outcomes (e.g., health improvements).
What is ultimately assessed and quantified, and Counterbalancing effects, such as cost-savings (e.g., lower
the level of detail, will depend on the results of operating costs if the regulation allows industry to replace
the subsequent screening, as well as on what is older technology with more efficient equipment) or negative
learned in the course of the analysis. benefits (e.g., health risks of substituting less safe drugs or less
healthy foods) should be assigned to the same category as the
Analysts should comprehensively consider all effect they offset; i.e., as costs and benefits respectively.
potentially important consequences, including
both those that are intended and unintended
(positive or negative). They should also consider whether behavioral anomalies will lead to different outcomes
than expected under the rational actor model typically assumed in economics. For example, individuals may
respond to policies intended to increase safety by reducing their level of precaution, or such polices may lead to
changes in social norms that lead to healthier behaviors. Another example is hyperbolic discounting (or present
bias, sometimes described as self-control problems), which can lead individuals to engage in behavior (such as
eating too much, exercising too little, or continuing to smoke) that is contrary to their own self-described
preferences.11
Evaluating these consequences, through estimating their monetary value, is discussed in Chapters 3 and 4.
These monetary values should be estimated as accurately and comprehensively possible, given analytic goals
and time and resource constraints. Effects that cannot be quantified should be highlighted for consideration by
decision-makers, as described in Chapter 6. That chapter also discusses methods for addressing uncertainty in
the quantitative results.

2.4 USE SCREENING TO FOC US THE ANALYSIS

Once the initial framing of the analysis is completed, as discussed above, the subsequent steps involve
determining how to best target future work, conducting the analysis, and reporting the results, as illustrated in
Figure 2.1. Analysts will need to follow a similar process to determine the types of supplemental analyses to be
conducted and the focus of that work. These processes are iterative; each step in the analysis will result in
another round of decisions about whether to address certain impacts in more detail or focus attention
elsewhere.

10
Whether an impact is counted as a benefit or cost will affect the ratio of benefits to costs. As noted in OMB Circular A-4, benefit-cost ratios (and cost-
benefit ratios) can be misleading (OMB 2003, p. 10) and generally should not be used as an indicator of economic efficiency. To avoid misunderstanding, such
ratios should not be reported unless accompanied by information clarifying their appropriate interpretation.
11
For more discussion of the valuation of the benefits of HHS policies that address habitual or addictive goods, see Cutler et al. (2015).

8
 F I G U R E 2 . 1 . A N A LY T I C S T E P S

Screening analysis is a useful tool for targeting subsequent work. Such analysis is typically based on easily
accessible data and simple assumptions; its goal is to provide preliminary information on the possible direction
and magnitude of the effects and to inform decisions about future work. For example, high-end values can be
used to determine whether various types of outcomes are likely to be significant even under extreme
assumptions. Depending on the results, this screening may be followed by more detailed assessment that
involves collecting additional data, refining the methods used, and possibly expanding the scope of the analysis
as discussed in the following chapters. The analysis should discuss non-quantified impacts along with the
quantitative results, and include assessment of uncertainties. The RIA should clearly document the results, as
well as discuss the data sources and analytic steps and the implications of uncertainties.
Because analytic resources are limited, the ideal regulatory analysis will not assess all policy options, nor
quantify all outcomes, with equal precision. In some cases, the cost of analyzing a particular policy option or
quantifying a specific outcome will be greater than the likely benefit of assessing it, given its importance for
decision-making.12 In other words, the analysis may not sufficiently improve the basis for decision-making to
pass an informal benefit-cost or value-of-information test. Conversely, options and outcomes that are important
for decision-making should receive substantial attention. “Importance” may depend on the likely magnitude of
the impacts; it may also depend on the need to respond to questions likely to be raised by decision-makers and
others.
The content and level of detail, and the length of the RIA (which may be very short or very long), are likely to
depend on the nature of the regulation, the characteristics of its benefits and costs, the populations affected,
and the data and other analytic resources available. It is not possible to design a “one size fits all” approach;
analysts need to exercise professional judgment in tailoring the analysis for an individual regulation. Generally,
conducting screening analysis and following a phased approach will help ensure that the work is carefully
focused and useful.

12
Such decisions include those related to assessing whether statutory change may be desirable; as noted earlier, the analysis need not be limited to
considering options allowed under current law.

9
Chapter 3
Assess Benefits
HHS regulations have many beneficial outcomes, including cost-savings as well as reduced health risks. As
introduced in Chapter 2, the distinction between benefits and costs is not always clear. Generally, impacts
categorized as benefits should relate to the intended outcomes of the regulation; i.e., the welfare improvements
that comprise its goals. Impacts categorized as costs should relate to the investment or inputs needed to achieve
those outcomes.
Methods for assessing both increases and decreases in costs are discussed in Chapter 4; this chapter addresses
changes in health risks.13 Such benefits are often the primary goal of HHS regulations and generally cannot be
valued using market measures.14 Calculating these benefits requires first estimating the change in risk associated
with each regulatory and non-regulatory option (in comparison to the baseline) then estimating its monetary
value. Below, we focus on valuation, first introducing basic concepts and methods, then describing specific
approaches for application in HHS analyses.

3.1 BASIC CONCEPTS

The starting point for valuation is an estimate of the impact of each regulatory option on specific health effects,
generally expressed as a change in the probability of illness or death for the average affected individual. The
monetary value of the benefit to the average individual can be calculated as the change in probability of the
illness or death multiplied by the value per statistical case, and summed across the affected population.
In practice, there is often little information on how the risk reduction or the value per statistical case varies
across individuals. It is common practice to aggregate the changes in risk over the affected population to
calculate the number of “statistical” cases averted by a regulation or other policy, and to multiply this by an
average value per statistical case.15 If, for example, a regulation would decrease the individual risk of a particular
illness or death by 10/20,000 annually throughout a population of 200,000, then 100 statistical cases would be
averted each year. The calculation is straightforward:
10/20,000 risk reduction x 200,000 individuals annually

= 100 statistical cases

Thus averting a statistical case or “saving” a statistical life is not the same as preventing an identifiable individual
from becoming ill or dying; rather, it is a sum of probabilities.16
The question for the regulatory analyst is thus how to best estimate the value of these risk changes. Because we
currently lack high quality, applicable studies that can be used to value the combined risk of illness and death,
we generally estimate the number of averted statistical cases of premature mortality and of morbidity
separately, then apply values to each and sum the results. The framework and methods for estimating these
values is described below.

13
Information on valuing other types of benefits, such as environmental improvements, is available in the U.S. Environmental Protection Agency’s Guidelines
for Preparing Economic Analysis (2014) and in Freeman et al. (2014).
14
The goal of some regulations is to provide cost-savings rather than risk reductions. In such cases, these savings would be categorized as benefits and the
methods described in the cost chapter would be used to value them. At times, whether to include an impact as a benefit or cost will be unclear, and analysts
will need to document this uncertainty in describing how the impact is categorized in the RIA.
15
This second approach yields the same result as the first, theoretically correct, approach if the risk reduction is the same across individuals, or the value per
statistical case is the same across individuals, or the risk reduction and value per statistical case are uncorrelated in the population.
16
In is typically impossible to identify ex ante whose illness or death will be prevented by a rule; in many cases, it is also impossible ex post.

10
3.1.1 ECONOMIC FO UND ATIO N
The approach for valuing mortality and morbidity risk reductions, as well as other policy impacts, is grounded in
four basic assumptions that underlie the standard economic model. The first is that each individual is the best
judge of his or her own welfare. This principle of consumer sovereignty means that benefit values should be
based on the preferences of those affected by a policy. Such framing allows analysts to provide decision-makers
with information on how those who would benefit are likely to value the improvement in their own health or
longevity.
The second is that individuals can be modeled as deriving utility (well-being) from the goods and services they
consume. If an individual chooses to buy a good or service, economists conventionally assume (consistent with
consumer sovereignty) that he or she values the good or service more than the other goods or services he or she
could have used that money to buy. Thus an individual's willingness to exchange money for different goods and
services can be used to measure the utility he or she receives from their consumption. The monetary value of a
risk reduction is appropriately measured by determining the change in wealth that has the same effect on utility
as the risk reduction.
The third is that estimates of individual willingness to pay (WTP) provide a conceptually appropriate measure of
value.17 WTP is the maximum amount of money an individual would voluntarily exchange to obtain an
improvement, given his or her budget constraints. It indicates the point at which the individual would be equally
satisfied with having the good and less money, or with spending the money on other things. In addition to
reflecting the trade-offs individuals make in everyday decisions related to spending on health and safety, this
framing mimics the actual trade-offs implicit in regulation. If we as a nation choose to spend more, for example,
on regulations that reduce food pathogen risks, we will have less to spend on other goods or services – including
other risk-reducing measures.
The fourth key assumption is that benefit values are determined
by the change in the amount by which aggregate WTP exceeds WHAT IS THE BASIS FOR VALUATION?
the market price, or “consumer surplus.” When WTP exceeds Benefits are valued based on the maximum
price, the individual benefits from the fact that he or she can amount of money an individual would
acquire the good or service for less than his or her willingness to willingly exchange for the improvement,
pay. If price exceeds WTP, the individual would not purchase the reducing his or her ability to purchase other
good or service. The difference between WTP and price can be things. This means that the value of mortality
aggregated across individuals to determine the consumer surplus and morbidity risk reductions is determined
associated with different price levels. Consumers generally by the affected individuals’ willingness to pay
benefit from price decreases, because WTP then exceeds price by for the change in their own risk.
a larger amount, and vice-versa. More information on this
concept is provided in Appendix B.18

3.1.2 VALUATION METH ODS

For goods such as mortality and morbidity risk reductions, prices do not exist because they are not directly
bought and sold in markets. Instead, we use the methods described below to estimate how much individuals
would be willing to pay for the risk reductions. We can then compare aggregated WTP for these risk reductions
(and other benefits) to the costs of a policy to determine the extent to which it is likely to yield net benefits.

17
Estimates of willingness to accept compensation (WTA); i.e., of the least amount of money an individual would accept to forgo an improvement, are also
consistent with this framework (see Robinson and Hammitt 2011, 2013). We refer to WTP throughout this discussion because it is more frequently studied. In
addition, regulations generally involve spending for improvements from the status quo, rather than compensation to forego an improvement, in which case
WTP is conceptually more appropriate.
18
A similar concept applies to producers, who earn a surplus when they can supply units of a good for less than the market price, as discussed in Appendix B.

11
For nonmarket outcomes, economists typically rely on revealed or stated preference studies to estimate WTP.19
Each has advantages and limitations: the choice of approach depends on the quality of the available research
and the extent to which it measures an outcome similar to the policy outcome.
Revealed preference studies rely on observed market behavior to estimate the value of related nonmarket
goods. For example, wage-risk (hedonic-wage) studies examine the compensation associated with jobs that
involve differing risks of death or nonfatal injury, using statistical methods to separate the effects of these risks
from the effects of other job and personal characteristics. While such methods have the advantage of relying on
actual behavior with real consequences, it may be difficult to find a market good that can be used to estimate
the value of a particular policy outcome.
Stated preference methods typically employ survey techniques to ask respondents about their WTP for the
outcome of concern. Such surveys may directly elicit WTP for a particular scenario, or may present respondents
with two or more scenarios involving different attributes and prices.20 In the latter case, estimates of WTP are
derived from the way in which respondents choose, rank, or rate alternatives. Stated preference methods are
attractive because researchers can tailor them to directly value the outcomes of concern; for example, the
survey can describe a particular type of illness from a particular type of exposure. A potential weakness is that
respondents do not directly experience the consequences of their decisions and may have limited incentives to
consider the questions carefully. Such surveys must be carefully designed and administered and satisfy various
tests for coherence to be considered reliable for use in regulatory analysis.
Analysts often must rely on existing studies when estimating parameters values as well as their
interrelationships, due to the substantial time and expense associated with conducting new primary research.
When used to value benefits, this approach is typically referred to as “benefit transfer,” and generally consists of
the five steps described in Figure 3.1. It requires careful review of the literature to identify high-quality studies
that are suitable for use in a particular context. “Quality” can be evaluated by considering the likely accuracy and
reliability of the data and methods used, referencing guidance on best practices.21 “Suitability” or “applicability”
involves considering the similarity of the risks and the populations affected.
In the subsequent sections, we describe how this framework should be used by HHS regulatory analysts to value
mortality and morbidity risk reductions. Numerous studies of the value of mortality risk reductions have been
conducted; morbidity risk reductions have received substantially less attention. In the latter case, because fewer
studies have been completed, analysts often rely on proxy measures.

19
Experimental methods (see, for example, Shogren 2005) and structural models that combine theoretical expectations with data from various sources (see,
for example, Smith et al. 2006) are less frequently applied but may be useful in some cases.
20
Although the terminology is not always used consistently, the first type of study is usually referred to as a contingent valuation survey; the second as a
choice experiment.
21
Some guidelines for determining study quality are provided in OMB (2003) and EPA (2014) as well as in the sources cited in this guidance. Because these
methods are continually evolving as additional research provides new insights, analysts should also consult recent articles and reports for updated guidance.

12
 FIGURE 3.1. BENEFIT TRANSFER FRAMEWORK

3.2 VALUING MORTALITY RISK REDUCTIONS

The approach for valuing mortality risk reductions is generally based on estimates of the value per statistical life
(VSL), from which a value per statistical life year (VSLY) is sometimes derived.22 We first introduce both concepts
then discuss recommendations for HHS analyses.

3.2.1 THE VALUE PER STATISTICAL LIFE AND THE VALUE PER STATIS TI CAL LIFE YEAR
As noted earlier, the starting point for valuation is typically an estimate of the individual risk change associated
with each regulatory option. Valuation also starts at the individual level, estimating what an individual would be
willing to pay for a defined change in his or her own risk, consistent with the principle of consumer sovereignty.
Values for mortality risk reduction reflect the rate of tradeoff between money and small changes in mortality
risk, referred to as the marginal rate of substitution between wealth and risk (Hammitt 2000). This value is
conventionally reported in dollars per statistical life (the VSL), and often estimated by dividing the value of a
small risk reduction by the size of the risk change.23 For example, if an individual is willing to pay $900 for a 1 in
10,000 reduction in his or her risk of dying in the current year, his or her VSL is calculated as:
$900 WTP ÷ 1/10,000 risk change
= $9.0 million VSL

22
Recommendations related to the use of the VSL or VSLY in HHS RIAs are discussed later in this chapter.
23
For the U.S. population, the annual likelihood of dying at each year of age increases from about 10/10,000 to about 100/10,000 between age 20 and age 65,
conditional on surviving to that age (Arias 2014).

13
The key parameter is the individual’s WTP for the 1 in 10,000 risk reduction (i.e., the $900); it is expressed as the
VSL (i.e., the $9.0 million) largely for convenience.24 The value of a statistical life is not the value of saving an
individual’s life with certainty.
In principle, WTP should change nearly in proportion to the change in risk, as long as the risk change is small
enough that WTP does not substantially limit other spending. Thus a single VSL can be used to value a range
of small risk changes.25 In other words, if we decrease the risk change in the above equation by a factor of 10,
to 1/100,000, we assume that WTP will also decrease by a factor of 10, so the VSL will still be $9.0 million
(= $90 WTP ÷ 1/100,000 risk change).
The VSLY is a related concept. In contrast to the VSL, which is the rate at which the individual substitutes money
for reductions in current mortality risk (within the current year or other short time period), the VSLY is the rate
at which he or she substitutes money for gains in life expectancy. A reduction in current mortality risk implies a
corresponding increase in life expectancy and hence a corresponding gain in life years.26
Under the VSLY approach, a reduction in mortality risk is typically valued by calculating the corresponding gain in
life expectancy and multiplying it by a VSLY. (Generally, future life years are first discounted to account for time
preferences; discounting is discussed in more detail in Chapter 5.) As does WTP more generally, both the VSLY
and the VSL vary depending on the characteristics of the individual and of the risk, and may increase, decrease,
or remain the same depending on the age (and remaining life expectancy) of the affected individual. However,
few primary research studies directly estimate the VSLY; it is typically instead derived from a VSL estimate using
simple assumptions.

3.2.2 LITERATURE REV IEW

HHS commissioned a review of the VSL literature to identify values that are suitable for use in its regulatory
analyses (Robinson and Hammitt 2016). The review had two goals: (1) to identify studies that meet evolving
criteria for “best practices” for VSL research; and (2) to tailor the estimates used by HHS to the types of risks it
regulates.
The criteria for that review were derived from several reports and articles that describe best practices for
valuing mortality risk reductions in regulatory analyses (OMB 2003, EPA 2010, Kling et al. 2011, Cropper,
Hammitt, and Robinson 2011, and U.S. Department of Transportation (DOT) 2015a). The criteria are listed in
Figure 3.2 and discussed in more detail by Robinson and Hammitt (2016) as well as in these source documents.

24
The VSL is at times described as aggregating individual WTP across a population; i.e., if each individual is willing to pay $900 for a 1 in 10,000 risk, and the
population included 10,000 such individuals, then the value per statistical case would be $9 million ($900 * 10,000 individuals). This definition can be
misleading, however, because WTP for a similar risk reduction is likely to vary across individuals.
25
Many VSL studies consider risks in the range of 1/10,000 or 1/100,000. While applying the resulting VSL to smaller risk changes is appropriate, care must be
taken in cases where the risk change is substantially larger. As the risk change increases, WTP will be increasingly limited by income, reducing the VSL (see
Alolayan et al. 2015 for more discussion).
26
Because death can be postponed but not prevented, reducing the risk of dying at one time necessarily increases the risk of dying at some later time.
Similarly, reducing the chance of dying from one cause necessarily increases the risk of dying from some other cause. For example, if a policy were to reduce
the chance of dying this year from 5 percent to 2 percent, then the chance of dying in a future year would increase from 95 percent to 98 percent. In
general, a regulation may reduce individuals' hazard function (the chance of dying at specific dates or ages conditional on being alive). This shift in the
hazard can be expressed as a reduction in the expected number of deaths in a specified time period (less than one for an individual) or as an increase in the
expected number of years lived; the individual's WTP for the shift in the hazard can be expressed as a VSL or a VSLY by dividing WTP by the expected change
in deaths or years lived (see Hammitt 2007).

14
 FIGURE 3.2. SELECTION CRITERI A FOR VSL STUDIES

General Criteria
1. Be publicly available.
2. Be written in English.
3. Provide estimates for the general U.S. population.

Criteria for Revealed Preference Studies

4. Use hedonic methods that address the trade-off between wages and job-related risks.
5. Control for potentially confounding factors, such as nonfatal injury risk as well as both industry and occupation.
6. Rely on high quality risk data, equal or superior to the Census of Fatal Occupational Injuries.

Criteria for Stated Preference Studies

7. Elicit values for private risk reductions that accrue to the respondent.
8. Express the risk change as a probability.
9. Estimate willingness to pay, not willingness to accept compensation.
10. Provide evidence of validity, including sensitivity of willingness to pay to changes in risk magnitude.

The review yielded six revealed preference studies that meet the selection criteria, all of which consider the
trade-off between wages and occupational risks, as well as one meta-analysis of these studies.27 Of the stated
preference studies, three met the selection criteria.28 These latter studies consider fatal risks associated with
motor vehicle accidents, ingesting pesticide residues on food, and unspecified causes. One considers only fatal
injuries, the other two also address illness-related fatalities.
When adjusted for inflation and real income growth, the VSLs highlighted in these studies range from $4.4
million to $14.2 million, with a mid-point of $9.3 million (2014 dollars and income levels).29 Applying these
results in HHS analyses requires additional adjustments, as described below.

3.2.3 RECOMMENDED VALUES

The range of VSL estimates that result from this review form the WHAT VSL SHOULD BE APPLIED
basis for HHS’ approach for valuing mortality risk reductions; HHS IN HHS ANALYSES?
anticipates periodically updating these estimates to reflect the For analyses conducted in 2014 dollars, risk
results from new research. This section discusses issues related to reductions that occur in 2016 should be
adapting these values for application in different regulatory valued using a central VSL estimate of $9.6
contexts and in different years. The approach it discusses should million. Analysts should test the sensitivity of
be applied in all HHS RIAs to provide a common reference case their results to values of $4.5 million and
that is comparable across analyses. However, analysts may also $14.6 million. The text describes how to
report results using alternative estimates or assumptions, if well- adjust these values for other years.
justified given the characteristics of the policy and the available
research.30
We expect the VSL to vary depending on individual characteristics such as age and health status, and on risk
characteristics such as whether death occurs immediately or after an extended illness. However, the effects of
many of these characteristics have not been well-studied, and the results of the available research are often
inconsistent. Thus the same population-average VSL should be applied in all RIAs, accompanied by discussion of

27
The six wage-risk studies are: Viscusi (2004), Kniesner and Viscusi (2005), Hersch and Viscusi (2010), Lee and Taylor (2013), Scotton (2013), and Viscusi
(2013). The meta-analysis of wage-risk studies is Viscusi (2015).
28
The three stated preference studies are Corso, Hammitt, and Graham (2001), Hammitt and Haninger (2010), and Cameron and DeShazo (2013).
29
The estimates reported in Robinson and Hammitt (2016) ranged from $4.2 million to $13.7 million, with a mid-point of $9.0 million (2013 dollars and income
levels). They have been updated to 2014 dollars and income levels in these Guidelines, using the approach described below.
30
See Chapter 6 for more discussion of the analysis of uncertainties.

15
uncertainties.31 The values cited above should be adjusted for inflation and real income growth as well as for
latency or cessation lag if relevant.32 Sensitivity analysis also should be conducted in cases where the individuals
affected are predominantly very young or very old. Each of these adjustments is discussed below.
The first set of adjustments is needed to reflect the time that has elapsed since the VSL studies were conducted,
and involve addressing both inflation and changes in real income. The process for inflating values to reflect
economy-wide price levels as of a common dollar year is discussed in Chapter 5.33 Adjusting for real income
growth is a separate step that requires two inputs: an estimate of the change in population-wide real income
per person, and an estimate of the extent to which WTP is expected to change in response to the income
change. The latter is generally expressed as the percentage change in the VSL associated with a one percent
change in real income; i.e., the income elasticity.
Although both economic theory and numerous empirical studies suggest that the VSL increases as real income
increases, the rate of increase is uncertain (Hammitt and Robinson 2011, DOT 2015a). Some research suggests
that a one percent change in income leads to less than a one percent change in the VSL (e.g., Viscusi and Aldy
2003), and other research suggests that it leads to more than a one percent change (e.g., Kniesner, Viscusi and
Ziliak 2010, Viscusi 2015). Given this uncertainty, HHS analysts should apply an income elasticity of 1.0 in their
analyses.34
Once the VSL has been inflated to the common dollar year used in the analysis, the formula for adjusting for real
income growth (assuming a constant rate of income growth and a constant income elasticity) is:
elasticity * (y – x)
VSL(year y) = VSL(year x) * (1+real income growth rate)

Because no single source provides data on both actual and projected changes in real income, analysts will need
to use different sources depending on the time period. More specifically, analysts should use Current Population
Survey (CPS) data to adjust for past income growth, and Congressional Budget Office (CBO) data to adjust for
future income growth; both focus on earnings, consistent with the measures generally used in the VSL studies.35
The most recent CBO report (2015, p. 112) projects real earnings growth at 1.4 percent per year for 2015
through 2040.
Table 3.1 provides an example of the adjustments to these values over a 10-year period using the data sources
identified above, including a VSL income elasticity of 1.0 and real income growth of 1.4 percent per year. (Note
that while values should be inflated only to the common dollar year used in the analysis (2014 in this example),
the adjustment for real income growth is needed for each subsequent year that the analysis covers.) As
indicated by the table, if the analysis is conducted in 2014 dollars, mortality risk reductions that accrue in 2016
would be valued using a central VSL estimate of $9.6 million. At minimum, analysts should test the sensitivity of
their results to the values at the low and high ends of the range; i.e., $4.5 million and $14.6 million.

31
This implies there should be no adjustment for morbidity prior to death for fatal cases. If regulation of a hazard (such as a foodborne pathogen) can prevent
both fatal and nonfatal illness, the expected reduction in fatal cases should be valued using the VSL and the expected reduction in nonfatal cases should be
valued using appropriate estimates of WTP or monetized QALYs as discussed in Section 3.3.
32
Latency is the time between when an individual is exposed to a hazard and when the adverse effect results; cessation lag is the time between when an
individual’s exposure to a hazard ends (or is reduced) and when his or her risk of adverse effect declines. These time periods are not necessarily equal.
33
The calculations in the text use the Consumer Price Index (CPI, available at [Link] to adjust for inflation. As discussed in Chapter 5,
analysts may use the Gross Domestic Product implicit price deflator instead of the CPI.
34
If changing the income elasticity estimate is likely to substantially change the analytic results, analysts should explore the effects of applying alternative
elasticities.
35
More specifically, for income growth in prior years, analysts should use CPS data on the annual median usual weekly earnings of employed wage and salary
workers, for fulltime workers (usual working hours over 35), reported on an average per capita basis in constant dollars, which are available at
[Link] For income growth in future years, analysts should use the estimates in the CBO Long-Term Budget Outlook. The 2015
CBO report is available at [Link] See Chapter 4 of these Guidelines for more discussion of the use of median versus mean
values.

16
 TA B L E 3 . 1 . V S L E S T I M AT E S B Y Y E A R ( 2 0 1 4 D O L L A R S )

YEAR LOW VSL ESTIMATE CENTRAL VSL ESTIMATE HIGH VSL ESTIMATE

2014 $4.4 million $9.3 million $14.2 million

2015 $4.4 million $9.5 million $14.4 million
2016 $4.5 million $9.6 million $14.6 million
2017 $4.5 million $9.7 million $14.8 million
2018 $4.6 million $9.9 million $15.0 million
2019 $4.7 million $10.0 million $15.2 million
2020 $4.7 million $10.1 million $15.4 million
2021 $4.8 million $10.3 million $15.6 million
2022 $4.9 million $10.4 million $15.9 million
2023 $4.9 million $10.6 million $16.1 million
Note: See text for discussion of assumptions and calculations.

Thus if a regulation reduced the number of statistical cases of premature mortality by 75 in 2016, applying the
central VSL estimate would result in benefits of $720.0 million (75*$9.6 million), with a low of $337.5 million
(75*$4.5 million) and a high of $1,095.0 million (75*$14.6 million).
In some cases, analysts may also need to adjust the VSL estimates to reflect a lag or delay between when
exposure to a hazard is reduced and when the risk change occurs. If the risk reduction is expected to occur in the
same year that the regulatory costs are incurred, then the VSL for that year should be applied. If the risk change
occurs later, then the VSL should be applied at the time when the risk change occurs, rather than in the year in
which the associated regulatory costs are incurred.36 In other words, using the values above, if both the costs
and the risk reductions occur in 2016, than $9.6 million would be used as the central VSL estimate. If instead the
costs are incurred in 2016 but the risk reduction does not occur until 2018, then the central estimate would be
$9.9 million, which would be discounted back to 2016 for comparison with the costs incurred in that year, using
the same discount rate as applied elsewhere in the analysis. Recommended rates, as well as the mechanics of
discounting, are discussed in Chapter 5.
Finally, some regulations may predominantly affect the very young or very old, rather than those of all ages. In
these cases, the age distribution of those affected is likely to differ significantly from the age of those included in
the VSL studies that underlie the approach discussed above, which often address individuals between the ages
of 18 and 65 (with some exceptions). There is substantial uncertainty regarding how VSL varies with age (see, for
example, Aldy and Viscusi 2007, Krupnick 2007, and Hammitt 2007; Robinson and Hammitt 2016 discuss related
theory and empirical research in more detail).
If a regulation largely affects the very young or the very old, analysts should at minimum provide a supplemental
sensitivity analysis based on estimates of the expected value of future quality-adjusted life years (QALYs).37 In
other words, regulations that primarily affect young children or the elderly should include two calculations: a
primary benefit estimate based on the VSL recommendations in this section, and a sensitivity analysis based on
monetized QALY estimates, which are discussed in detail in the next section. In this sensitivity analysis, the value
per QALY is multiplied by the present value of the expected life year gain.38
In addition to the uncertainties represented by the ranges of values and adjustments discussed above, analysts
should provide a qualitative discussion of the other limitations of this approach. The major limitations include
differences in the types of risks addressed in the underlying studies and those addressed by the particular policy,

36
As noted earlier, this delay is described as the “cessation lag” when it refers to risk reductions rather than risk increases.
37
Analysts may also explore the effects of alternative assumptions regarding the relationship between the VSL and age or life expectancy, if clearly explained
and well-justified. Many analyses have used a VSLY estimate rather than an estimated value per QALY to explore these effects.
38
Such sensitivity analysis would not noticeably change the results if the age distribution of those affected by the regulation is similar to the U.S. age
distribution, as long as the value per QALY (or VSLY) is calculated from a population-average VSL.

17
as well as in the population affected. Thus this approach may over- or understate the value of mortality risk
reductions. Where the analytic conclusions are particularly sensitive to the approach used to value mortality
risks, analysts may also wish to conduct breakeven analysis to identify the VSL at which the costs would be equal
to the benefits, as discussed in Chapter 6.

3.3 VALUING MORBIDITY RISK REDUCTIONS

Valuing morbidity risk reductions is more complicated than valuing mortality risk reductions for two reasons.
First, morbidity risks are more diverse, differing in duration and severity as well as in the attributes of health
that are affected (e.g., physical or cognitive functioning). Second, high quality WTP estimates are not available
for many morbidity risks, requiring the use of proxy measures.
Thus, as discussed below, HHS analysts should first review the HOW SHOULD MORBIDITY RISK REDUCTIONS
literature to determine whether WTP estimates of reasonable BE VALUED?
quality are available for risks similar to those addressed by the
Analysts should first review the literature to
regulation, applying the benefit transfer framework described
determine whether suitable WTP estimates
previously.39
of reasonable quality are available. If not,
If such estimates are not available, analysts should instead apply they should use monetized QALYs as a proxy,
values that combine estimates of the resulting QALY gain with following the approach described in this
estimates of the monetary value per QALY. Cost-savings that are section.
not reflected in the QALY measure may be added to these values,
including those that accrue to third parties (such as savings in
insured medical costs). Because of the diversity of the health effects and the gaps in the research literature, the
discussion that follows focuses on the approach analysts should follow to develop estimates, rather than
recommending values for particular health conditions.

3.3.1 QUALI TY-ADJUSTED LIFE YEARS

The QALY is a nonmonetary measure that integrates the duration and severity of illness. QALYs are widely used
to rank and prioritize public health programs, analyze the cost-effectiveness of health policy and medical
treatment decisions, and compare health status across individuals or population groups. In these contexts,
QALYs are generally not assigned a monetary value, but monetization is needed to apply these estimates in
regulatory analysis.40
QALYs are derived by multiplying the amount of time an individual spends in a health state by a measure of the
health-related quality of life (HRQL) associated with that state. HRQL is estimated using a scale anchored at zero
and one, where one corresponds to full health and zero corresponds to a state that is as bad as dead (values
cannot be greater than one but may be less than zero for states that are judged to be worse than dead). In
principle, the HRQL associated with a health state may vary among individuals, but in practice a common value is
used for each health state. Expected QALYs are then calculated by weighting the HRQL experienced in each
future year of life by the probability of living in that year (i.e., by the survival curve).41 In addition, future QALYs
are usually discounted using the same rates as applied to monetary values. Appendix C provides more
information on the estimation of QALYs.
Once HRQL is determined for a particular health state and multiplied by the duration of that state, the resulting
QALYs can be summed across the health states (e.g., acute and chronic phases) associated with a particular
illness, and across the illnesses associated with a particular hazard. For example, for foodborne illness, QALYs
39
Due to the lack of a reasonably recent and comprehensive review of this research, analysts will need to search bibliographic databases, such as EconLit
([Link] and EVRI ([Link] to identify potentially applicable studies and
conduct a criteria-driven review that follows the benefit transfer framework introduced above.
40
In cost-effectiveness analysis, valuation is implicit, because monetary thresholds are needed for comparison to the cost-effectiveness ratio to determine
whether an intervention is worth implementing. In addition, valuation is implicit in any policy decision that results, which involves choosing to fund a
particular invention rather than using the money for other goods or services.
41
For the U.S. population, survival curves are updated annually by the CDC; see [Link]

18
can be summed across cases of acute gastrointestinal illness, including those that do and do not require
hospitalization, as well as more severe effects. For regulatory analysis, health status with the regulation or other
policy must be compared to health status in the absence of the regulation, which is likely to be less than full
health. In particular, health status generally deteriorates with age, so that average HRQL for older individuals is
generally less than 1.0 (see Hamner et al. 2006). Some regulations may also target individuals with pre-existing
conditions or lifestyle characteristics that will not be ameliorated by the regulation.
An example of these calculations is provided in Figure 3.3. For simplicity, in this example we do not discount
future impacts; however, as discussed in Chapter 5, discounting should be used to reflect time preferences when
similar calculations are performed in regulatory analyses.
F I G U R E 3 . 3 . E X A M P L E O F Q A LY C A L C U L AT I O N S

 Assume that, in the absence of the policy, the average individual affected will experience health-related quality of
life of 0.7 throughout their estimated remaining life span of 20 years.
 With the policy, assume that the average individual affected will instead experience health-related quality of life
of 0.9 over the same time period
 The QALY gain attributable to the policy is the difference between 20 years with a health status of 0.9 (18 QALYs)
and 20 years with a health status of 0.7 (14 QALYs), which equals 4.0 QALYs, prior to discounting.

The research base for estimating QALYs is extensive, including numerous primary research studies as well as
population databases that collect HRQL data for a wide range of conditions. Thus regulatory analysts can
generally rely on existing research to estimate the QALY gains associated with reducing the risks of various types
of morbidity.42 Estimates from many previously completed studies can be found in the Tufts Cost Effectiveness
Analysis (CEA) Registry (described in Thorat et al. 2012), using the benefit transfer process discussed earlier to
assess their quality and applicability. However, this database does not include studies that estimate QALYs or
HRQL without comparison to costs, so analysts should search the research literature to identify other potentially
applicable studies.
Another option is to rely on population-wide surveys. Some large national surveys (such as the U.S. Medical
Expenditure Panel Survey or MEPS) have at times included one or more of the generic HRQL indices, such as the
EQ-5D which is described in more detail in Appendix C. These HRQL estimates can then be multiplied by duration
estimates from research on the health state of concern. Relying on such surveys can be particularly useful for
regulatory analysis, because they provide consistently-derived estimates across a wide range of outcomes and
enable analysts to control statistically for the effects of other factors (such as age and co-morbidities) on HRQL.43
For some health effects, however, these surveys may not include enough cases to reliably estimate HRQL.
A 2006 Institute of Medicine report provides more detailed discussion of these measures and their application in
regulatory analysis, recommending factors that should be considered in selecting among the available sources of
HRQL and QALY estimates. In particular, to the extent possible, QALY estimates should satisfy the criteria listed
in Figure 3.4.

42
In those rare cases where suitable estimates are unavailable, analysts may need to rely on expert judgment to estimate the QALY gains associated with the
regulation. Analysts should apply the EQ-5D index with U.S. weights when implementing this approach.
43
For example, EQ-5D scores for a large number of health conditions based on MEPS are provided in Sullivan and Ghushchyan (2006). This article, and a
calculator that allows users to retrieve EQ-5D scores by International Classification of Disease code and demographic characteristics, is available online at
[Link]

19
 F I G U R E 3 . 4 . S E L E C T I O N C R I T E R I A F O R Q A LY E S T I M AT E S

1) QALY estimates should be based on research that addresses the risks and populations affected by the regulation.
2) The description of the effects of the health state on quality of life should be based on information from those who have
experienced the condition (such as patients).
3) The preference weights placed on the health states should be based on a survey representative of the general U.S.
population.
4) The “without new regulation” baseline (with the condition) should be compared to a realistic estimate of “with-
regulation” health status, which takes into account factors (such as age and co-morbidities unrelated to the regulated
44
hazard) that may lead those affected to be in less than perfect health once the regulation is implemented.
5) The implications of related uncertainties should be discussed and addressed quantitatively if significant.

Developing approaches for measuring QALYs and testing their implementation is an active area of research.
There continue to be diverse opinions on many technical issues such as the dimensions of health that should be
considered, the types of survey questions that should be used to explore these dimensions, the elicitation of
preferences, and the statistical analysis of the results (Lipscomb et al. 2009). Thus the approaches described
above continue to evolve, and new options are under development.

3.3.2 THE VALUE OF A QALY

To use QALY estimates to value morbidity risk reductions in regulatory analysis, they must be assigned a
monetary value. One approach would be to rely on emerging research that explicitly considers individual WTP
per QALY (e.g., Haninger and Hammitt 2011); HHS is currently exploring this research to determine whether it is
possible to develop a function that reflects how the value varies depending on factors such as the severity and
duration of the effect.
In the absence of such a function, analysts often assume that the value per QALY is a constant, frequently
applying a VSLY estimate, calculated by dividing the VSL by the discounted expected number of life years
remaining. A preferable approach is to calculate a constant value based on expected QALYs rather than
expected life years. Future QALYs are generally less than future life years because health tends to deteriorate
with age. Dividing the VSL by future QALYs yields an average value per QALY larger than the VSLY (see Hirth et al.
2000).
For analyses conducted in 2014 dollars, HHS analysts should estimate the value of a QALY based on the VSLs
reported in Table 3.1. For analyses that use a different dollar year, the VSL estimates will first need to be
adjusted to reflect inflation and real income growth, as discussed earlier.
Based on data reported in the underlying VSL studies, analysts should assume that the average individual in
these studies is 40 years of age. Table 3.2 reports the value of a QALY that results when health-related quality of
life in each subsequent year is estimated using the U.S. EQ-5D results reported in Hamner et al. (2006) and the
conditional likelihood of survival for each year of age is based on the population-averages in Arias (2014).45 The
value of future years should be discounted at the same rates as used elsewhere in the analysis. The table
provides the results of these calculations for risk reductions that occur in 2016, in 2014 dollars.46

44
This point is of particular importance in regulatory analysis, which is intended to realistically reflect the health of the affected population without and with
the policy. In the absence of regulation, this population may suffer from a variety of health conditions, some of which will not be affected by the policy
change. For example, a food safety regulation that targets the risk of gastrointestinal illness is not likely to affect air pollution-related respiratory effects. In
addition, health status generally declines with age. Thus the average health of the affected population is likely to be less than perfect health (i.e., less than
HRQL = 1.0) even after the regulation becomes effective.
45
Arias (2014) provides life tables for 2009. Updated values may be used when available; see: [Link]
46
Many previous analyses value QALYs using a constant VSLY rather than the value per QALY presented here. As noted earlier, both are calculated from a VSL
based on the average life expectancy of the individuals studied. The difference is that the resulting VSLY estimate implicitly averages over future health,
while the value per QALY estimate takes into account the expected decline in health status associated with aging.

20
 TA B L E 3 . 2 . VA L U E P E R Q A LY I N 2 0 1 6
(2014 DOLLARS)

VALUE PER QALY

VSL
3% DISCOUNT RATE 7% DISCOUNT RATE

$4.5 million $230,000 $380,000

$9.6 million $490,000 $820,000
$14.6 million $750,000 $1,200,000

For example, if a regulation leads to a 0.2 QALY gain per affected individual on average in 2016, then applying
the central VSL estimate (using a 3 percent discount rate) from Table 3.2, the value of that gain would be
$98,000 (0.2*$490,000). If the gain accrues to 75 members of the population, than the total value of the risk
reduction would be $7,350,000 (75*$98,000).
Estimates of the averted costs of illness may be added to estimates of WTP or monetized QALYs, as long as the
same cost-savings are not counted elsewhere in the analysis (see Chapter 4 for more discussion of medical
costs).47 These cost estimates should always be reported as a separate line item in the RIA, so that their
treatment is clear. If a WTP study is used for valuation, the analysts should review the study to ensure that the
costs are not already captured in the WTP estimates. Typically, WTP studies may capture out-of-pocket costs
and lost earnings, and possibly informal care provided by household members, but do not include costs paid by
third parties, such as medical expenses paid by insurance. If a regulation reduces these costs, the savings can be
added to the WTP estimate. Any cost-savings included in the analysis of regulatory costs should not also be
added to the benefit estimates.
If monetized QALYs are used for valuation, the extent to which costs are included is highly uncertain given that
the measure does not directly reflect monetary consequences. Occasionally, studies that estimate HRQL instruct
respondents to assume their medical costs and lost income will be offset by insurance. In the absence of more
specific information, analysts may add medical costs paid by third parties to the monetized QALYs, but should
not add estimates of lost productivity or income to avoid potential double-counting.
Estimates based on QALYs monetized using a constant value are likely to be less accurate than approaches based
on direct estimation of WTP, but may provide a reasonable proxy when WTP estimates are unavailable. The
limitations of this approach relate in part to the characteristics of the QALY measure and in part to the approach
used for valuation, and should be discussed when documenting the analysis.
The construction of the QALY assumes that how individuals value health states is independent of the duration of
the state, the age at which it is experienced, the individual’s remaining life expectancy, and his or her wealth and
income (Hammitt 2002, 2013, Institute of Medicine 2006). Moreover, QALYs do not explicitly account for the
changes in wealth or income that result from changes in health, nor for how individuals are willing to trade off
spending on particular risk reductions versus spending on other goods and services.

47
For some rules, whether medical costs should be counted as a “cost” or “benefit” will be uncertain, and analysts will need to be clear about how these costs
are treated when documenting the analysis. Generally, if changes in medical costs are part of the implementation of the requirements (i.e., a policy input),
then they should be counted on the cost-side of the equation. If they are one of the policy outcomes, then they should be included in the benefit calculation.

21
In addition, relying on a constant value per QALY does not reflect the likely variation in value due to factors such
as duration and severity.48 More research is needed to develop a valuation function for QALYs that better
approximates individual WTP for risk reductions.
Given the above discussion, HHS analysts should first consult the WTP research to determine whether suitable
estimates are available for the morbidity risk reductions of concern. If not, they may use monetized QALYs as a
proxy, recognizing that we are uncertain whether the resulting values under- or overstate individual WTP for the
risk reduction. Regardless of whether WTP or monetized QALY estimates are applied, analysts should document
any concerns about the quality or applicability of the selected studies.

48
Given these concerns, an expert panel recommended against assigning monetary values to QALYs in regulatory analysis (Institute of Medicine, 2006);
however, OMB has not amended Circular A-4 to adopt this recommendation. It continues to suggest that the use of monetized QALYs is acceptable as long as
analysts acknowledge the limitations of the approach.

22
Chapter 4
Assess Costs
HHS regulations may impose costs on individuals, industries, other organizations (both for-profit and nonprofit),
and government entities. In some cases, costs may be offset by savings; for example if a regulation reduces or
streamlines existing requirements by replacing paper with electronic recordkeeping and reporting.49 This
chapter begins by describing some basic concepts that are particularly important when estimating costs. It then
describes approaches for estimating the most common types of costs in more detail.

4.1 BASIC CONCEPTS AND APPROACH

Below, we describe economic concepts that are of particular importance when estimating costs: opportunity
costs, transfers, and producer surplus. A discussion of the general approach to the cost analysis follows.

4.1.1 ECONOMIC FO UND ATIO N

Three fundamental notions from economic theory are of particular importance in assessing costs. The first is
that economists measure costs by the value of forgone opportunities. In other words, costs are incurred when
resources are used for one purpose and hence cannot be used for another purpose. The opportunity costs are
the value of the benefits that could have been provided by devoting the resources to their best alternative use.
This interpretation differs from the concept of accounting costs (i.e., actual expenses plus depreciation of capital
equipment). It is consistent with the concept of WTP, as discussed in Chapter 3.
The second is the distinction between resource costs and
transfers. Transfers are monetary payments between OPPORTUNITY COST VERSUS ACCOUNTING COST
persons or groups that do not affect the total resources
Opportunity costs are easy to confuse with
available to society.50 They are a benefit to recipients and
accounting costs. Some may argue that a proposed
a cost to payers, with zero net effect. For example, some
regulation will not have any “costs” because
types of taxes, fees, and surcharges can be categorized as
regulated entities will simply re-allocate existing
transfer payments. Such transfers often can be ignored in
resources to comply with the regulation; no new
benefit-cost analysis, as long as they do not lead to
expenditures are incurred.
behavioral changes that significantly affect the
calculation of net benefits. However, transfers should be However, if resources are shifted for compliance
included in the distributional analysis, as discussed in purposes, other productive uses of those resources
Chapter 7. are forgone. If labor is shifted to compliance from
production, for example, the opportunity cost is the
Where the imposition of transfer payments affects
value of forgone production.
behavior, associated impacts should be taken into
account in the benefit-cost analysis. For example,
reductions in government payments to hospitals would
often be viewed as a transfer. However, the affected hospitals may accept fewer patients or use less expensive
treatments, in turn affecting health outcomes. This change in health should be addressed in the benefit-cost
analysis, if significant. Similarly, taxes can also change behavior; for example, taxes on wages provide a
disincentive for working and higher taxes may lead more people to stay out of the labor force.51 In addition,
transfers involve transaction costs that may be significant in some cases. When identifying the costs to be

49
As discussed in Chapter 2, analysts should decide whether to report offsetting cost savings as negative costs or positive benefits depending on whether these
savings relate to the inputs needed to achieve regulatory goals, or the outcomes associated with those goals.
50
Because RIAs focus on the effects on the U.S. population, transfers from the United States to other nations, and from other nations to the United States,
should be included in the benefit-cost analysis.
51
HHS regulations rarely, if ever, affect tax rates. If such rates are affected, analysts may wish to consult Boardman et al. (2011) and other resources on
estimating the associated deadweight loss, typically referenced as the marginal excess tax burden.

23
quantified, analysts should consider the potential for significant net losses or gains nationally resulting from the
imposition of transfer payments.52
The third fundamental notion is the difference between compliance costs and changes in producer and
consumer surplus. As introduced in Chapter 3 and discussed in Appendix B, consumer surplus is the benefit that
consumers receive when they are able to purchase products for less than they are willing to pay; producer
surplus is the difference between the revenue producers receive and their cost of production. When a
regulation increases production costs, the market price is likely to increase, inducing consumers to reduce their
consumption and producers to reduce production. The cost of the regulation includes both the direct
compliance costs and the “deadweight loss” associated with the reduction in output. However, regulation often
has negligible impact on prices, in which case the deadweight loss will be quite small and compliance costs will
be a reasonable approximation of total costs. We return to this issue later in this chapter, when discussing the
use of partial and general equilibrium models.

4.1.2 GENERAL APPROA CH

Social cost is the sum of the resource costs incurred as a result of implementing the regulation. These costs may
include costs incurred by regulated entities in the form of resources (labor, material, equipment) used to comply
with the regulation, valued by their opportunity costs. Social cost may also include costs incurred by
governments to implement and enforce the regulation. Other effects, such as consumer decisions to replace the
regulated product with a substitute, may also occur in response to the compliance costs.
In principle, analysts could develop a model that includes all the interactions between regulated entities,
consumers, and related markets to capture the total social cost of a regulation. However, such analysis is usually
impractical given data, time, and resource constraints. Furthermore, most regulations are likely to have
negligible impacts on price, in which case such complex modeling is not necessary to understand key impacts.
During the framing and screening process (see Chapter 2), analysts should determine the cost categories of
interest and the modeling techniques to be applied, recognizing that this is an iterative process. Changes in the
approach may be needed as more is learned about the potential impact of the policy options. Nonquantified
costs, as well as the reason for not quantifying them, should be reported as well (see Chapter 6).
In most cases, the analysis focuses on estimating the incremental compliance costs incurred by the regulated
entities, assuming full compliance with the regulation, and government costs.53 Compliance costs include the
resources used by the regulated entities to comply with the regulation. These costs often account for the largest
proportion of social costs and are an important input into the supplemental analyses discussed in Chapter 7. The
analysis should also include costs incurred by the government. Such costs generally involve guiding and
monitoring implementation of the regulation, as well as providing information and training as needed. In some
cases, the government may have an ongoing operational role; for example, it may provide services in addition to
those provided by the regulated community. Government costs also include enforcing the regulation through
activities such as inspections and reporting requirements. When significant, these government costs should be
quantified.
If compliance costs are significant on a per entity or industry basis, they may result in other impacts. If these
additional effects are sufficiently large, they should be quantified. For example:
 Compliance costs may result in substituting behaviors. If industry or consumers shift to alternative products,
or if industry develops new products to replace the regulated products, analysts should consider the net
effect on society.

52
As noted in Chapter 1, an RIA is required for regulations resulting in significant transfers, because of the additional costs or benefits that may result.
53
Analysts should consider the uncertainty associated with an assumption of full compliance and provide analysis of alternative assumptions, as appropriate.

24
  Compliance costs may result in changes in available services, which could result in additional, and possibly
non-pecuniary, costs (e.g., time losses associated with needing to find new doctors or traveling farther for
treatment). Such costs should also be taken into account.54
Finally, care should be taken to identify transfer payments as discussed earlier. For example, proposed
regulations may require the payment of fees to HHS agencies for processing paperwork or adjudicating claims.
The fees may be set to recover the HHS labor and other costs associated with administering the program. If the
opportunity cost to HHS of administering the requirement is already captured in the analysis, the fees represent
a transfer payment that should not be counted. However, if the HHS opportunity costs are not separately
calculated, then the fees paid by regulated entities might be a reasonable proxy for these opportunity costs and
should be included as a social cost.

4.2 ASSESSING COMPLIANCE AND GOVERNMENT IMPLEMENTATION

COSTS
Chapter 2 discusses the screening process used to identify key cost categories. Typical categories include
administrative costs (including time, materials, and travel), capital and operations costs, and medical costs. As
discussed above, government implementation costs should also be considered.

4.2.1 ADMINISTRATIVE COSTS

Most regulations impose administrative costs on regulated entities or the implementing government agency.
Related activities may include, for example, reviewing the new regulations, developing protocols for compliance,
collecting and reporting data, and training staff on implementation. The following sections describe how to
quantify and monetize the components of these costs, including estimating the amount of time required for
administrative tasks, valuing this time in monetary terms, and locating data on administrative expenditures.
Amount of time required: Estimating the amount of time needed to comply with administrative requirements is
relatively straightforward.55 Usually, time should be measured in terms of “hours” so that the quantity can be
easily combined with information on the value of time, which is generally measured in terms of hourly
compensation (see below).56 Analysts may obtain estimates of the number of hours needed to review the
requirements, fill out forms, transmit data, or complete other similar tasks using surveys, information
interviews, past analysis, or Information Collection Requests (see Chapter 7). Who is undertaking these activities
is also important, as it affects the monetary value of time. Finally, care must be taken to ensure that the hours
estimates reflect the net effect of the regulation. For example, the regulation may require that workers
discontinue some activities (e.g., completing paper forms) and replace them with others (e.g., maintaining
records electronically). The time saved by discontinuing activities will offset time spent on the new activities to
some extent.
Table 4.1 lists the types of administrative tasks that may result from new regulations and provides suggestions
for quantifying the amount of time associated with each task, noting associated costs (such as travel or
materials) that should also be addressed. Time spent complying with a regulation may vary by establishment
type or size. Thus, analysts should explore differences across key groups. In addition to providing more accurate
cost estimates, this information is used in the supporting analyses that address impacts on entities of differing
sizes and types (see Chapter 7).

54
For a detailed discussion of the identification and assessment of secondary effects, see Chapter 5 of Boardman et al. (2011).
55
The same general approach can be used when regulations affect other types of time use.
56
For rules requiring substantial amounts of labor, such as the hiring of additional, full-time employees, analysts might instead estimate the number of new
employees needed and annual salaries using the data sources identified in the next section.

25
 TA B L E 4 . 1 . T Y P I C A L A D M I N I S T R AT I V E TA S K S

ADMINISTRATIVE TASK EXAMPLES OF SOURCES OR METHODS USED FOR QUANTIFICATION

Regulation and Guidance Review: All  Interview representatives of the affected community to obtain estimates of the
regulated entities, including those who incur amount of time required to review regulations, including time spent by legal or
no other compliance costs, will require time to technical experts.
read and interpret the regulation.  Review prior agency analyses for relevant data or conduct other literature reviews.
 Assume reviewers read at the average adult reading speed (approximately 200 to
250 words per minute) and allow time for both review and interpretation.
Development or Revision of Standard  Interview the affected community to obtain estimates of the amount of time
Operating Procedures (SOPs): The affected required to review and revise SOPs.
entities may need to devise a compliance plan  Review prior agency analyses for relevant data or conduct other literature reviews.
that may require them to change their SOPs.
SOPs are “detailed, written instructions to
achieve uniformity of the performance of a
specific function” (International Conference
on Harmonization).
Training: Once new SOPs are established,  Interview the affected community to obtain estimates of the amount of time
entities will spend time training staff on how required for training.
to implement the regulation.  Review prior agency analyses for relevant data or conduct other literature reviews.
 Note that training may also require travel costs.
Sampling and Testing: A regulation may  Contact third party vendors to obtain estimates of sampling or testing costs. In
require entities to sample or test materials. such cases, the cost per test likely includes both labor and materials.
 Some entities may be able to conduct the testing more cheaply using in-house
staff. Interviews with the affected community may provide data on the amount of
time required. In these cases, material costs should be added.
Record Keeping and Reporting: Some rules  Interview the affected community to obtain estimates of the amount of time
require entities to perform additional record required for record keeping and reporting.
keeping to track training, inspections, and  Review prior agency analyses for relevant data or conduct other literature reviews.
infractions. In addition, entities may be  Note that record keeping may also result in substantial storage costs.
required to submit recurring reports to the
regulating agency.

Valuing time: Once the amount of time needed has been calculated, analysts multiply these hours by a per hour
value of time. This value will vary depending on the characteristics of the activities, the preferences of those
affected, the duration of the activities, and other factors. As in other components of the analysis, the approach
to valuation requires comparing the value of time use with and without the regulation, to calculate the
opportunity costs the regulation imposes.
In RIAs, as in other types of analysis, time use is often valued based on simplifying assumptions that allow
analysts to use readily accessible data on compensation. We introduce the default assumptions for HHS analyses
below, then discuss their implications in more detail. If the characteristics of the regulation and the available
data justify a different approach, the rationale for the approach should be included in the RIA along with the
detailed calculations.
The value of time is an active area of research, and HHS is now working on a project that will further explore
these values. 57 This new work will investigate the extent to which the default values discussed in this section
appropriately measure different types of time use, including the suitable treatment of overhead costs. In the
interim, analysts should apply the default values described below. Regardless of the approach used, the
assumptions and related uncertainties should be addressed as discussed in Chapter 6.

57
See Boardman (2011) and DOT (2015b) for more discussion of the related literature.

26
The starting point for valuing changes in time use involves distinguishing between paid and unpaid time; i.e.,
between market production and nonmarket activities including leisure, household tasks, and volunteer work.
The first default assumption is that regulatory activities undertaken by paid employees will displace other paid
work tasks, while activities undertaken during non-work time will replace other unpaid activities. In other words,
the work-related administrative requirements likely to be imposed by HHS regulations (e.g., completing
additional reports) would not require that the affected individuals spend more time at work and less time on
leisure activities; instead they would spend less time on other tasks associated with their current occupation and
the employer might rearrange work assignments. If new employees are hired, this approach assumes that the
activities required for regulatory compliance would replace the activities they pursued in their previous job;
those hired would be transitioning between similar jobs rather than moving from unemployment to
employment.
The second default assumption is that average or median estimates appropriately measure the value of a
marginal unit of time. In reality, this marginal value will likely vary based on a variety of factors, such as the
amount of time (e.g., a few hours versus days or months) or the types of time use affected. However, marginal
estimates generally are not readily available; most easily accessible data sources provide averages or medians.
The third default assumption is that the value of activities conducted during paid work time can be best
approximated by the cost of labor to the employer. The standard economic model assumes that employers are
willing to incur labor costs equal to the value of workers’ marginal product. Conceptually, this amount
represents the value of what the employee would have otherwise produced in the absence of the regulation.
Thus the opportunity cost of paid work time can be approximated based on the employer costs, including pay,
benefits, taxes, and associated overhead.
The fourth default assumption is that the opportunity cost of unpaid time can be best approximated by post-tax
wages. Consistent with the standard economic model, this approach assumes that individuals decide whether to
engage in paid work depending on whether the incremental income exceeds the value they place on unpaid
time, a decision generally described as the labor-leisure trade-off. Taxes and benefits are usually excluded from
this calculation, assuming that individuals focus on their take-home pay in making related decisions. In other
words, an additional hour of paid work is valued differently by the employee than the employer, because many
costs to the employer are not received by employees (e.g., income and payroll taxes) or may not be visible to
the employees (e.g., benefits) and hence are unlikely to be taken into account in their decision-making. As is the
case throughout the analysis, particularly if the value placed on time use significantly influences the analytic
conclusions, then the approaches discussed in Chapter 6 should be used to assess the implications of related
uncertainties.58
Table 4.2 summarizes the assumptions used to develop default estimates of the value per hour of time. Below,
we discuss these assumptions in greater detail.

58
Estimating the value of time for individuals who are not active in the labor market, such as children or seniors, is particularly challenging. As discussed later
in this chapter, analysts should apply the same approach for valuing non-work time to all individuals.

27
 TA B L E 4 . 2 . C O N S T R U C T I N G D E FA U LT E S T I M AT E S O F T H E VA L U E O F T I M E

CONTEXT COSTS INCLUDED IN HOURLY VALUE DATA SOURCES AND KEY ASSUMPTIONS

Employees undertaking administrative  Pre-tax wages  OES or NCS ECEC data on wages
tasks while working
 Benefits:  Industry-specific data as available,
- Paid time off or assume benefits plus other
- Health benefits overhead costs equal 100 percent
- Retirement benefits of pre-tax wages (i.e., for a fully-
- Other legally required benefits loaded wage rate, multiply pre-tax
- Payroll taxes wages by a factor of “2”)
 Other overhead costs:
- General and administrative (G&A)
- Fixed overhead
- Insurance
- Accounting profit
Individuals undertaking administrative  Post-tax wages  OES or NCS ECEC data on wages
tasks on their own time  Adjust wage estimates using data
on household income before and
after taxes collected in the CPS
Acronyms:
CPS – Current Population Survey, available at [Link] (U.S. Census Bureau)
ECEC – Employer Costs for Employee Compensation, available at [Link] (U.S. Bureau of Labor
Statistics)
NCS – National Compensation Survey, available at [Link] (U.S. Bureau of Labor Statistics)
OES – Occupational Employment Statistics, available at [Link] (U.S. Bureau of Labor Statistics)

On-the-job activities: For paid work-related activities, the opportunity cost, or value, of a unit of time devoted
to regulatory compliance equals the marginal value of the product that would have otherwise been produced in
the absence of the regulation. Another way to frame this concept is to ask, what benefit to the economy would
the employee have produced with an additional hour of time?59 Data on this value are not readily available;
however, market data on employee compensation provide a reasonable proxy.
From an employer’s perspective, when making a decision about whether to hire more help, the company will
think about the entire cost of the new employee, including wages, fringe benefits, and other overhead costs
needed to support the employee in accomplishing the work. Thus, to estimate the value of an hour of time,
analysts should sum the costs of these items. Wages (pre-tax) generally include base pay, cost-of-living
allowances, guaranteed pay, hazardous-duty pay, incentive pay (commissions, bonuses), and tips.60 Fringe
benefits generally include paid time off, health benefits, retirement benefits, other legally required benefits
(e.g., worker’s compensation) and payroll taxes.61 The combination of wages and fringe benefits is often referred
to as the employer costs of compensation.
In addition, employers incur other costs that support labor and are not directly relatable to the production of
goods and services. These costs are generally referred to as “overhead costs,” and may include the following
categories:

59
Analysts should generally assume that time losses associated with HHS regulations only affect the quantity of hours worked, not the price of goods and
services produced with that time. Thus, the analysis focuses on the marginal value of production associated with an hour of time. If a proposed regulation
will result in time reallocations that are sufficiently large to affect the price of goods and services produced, more complex analysis may be required.
60
Throughout the remainder of this chapter, the term “wages” is used generally to refer collectively to all of these categories.
61
Conceptually, payroll taxes are included in this calculation because analysts need a full accounting of the total cost paid by employers for their employees’
time. As explained above, this cost is used as a proxy for the value of the employees’ production.

28
  General and administrative (G&A) costs, such as human resources, payroll, accounting, sales personnel,
executive salaries, legal fees, office supplies, equipment, communications, administrative buildings, office
space, travel, subscriptions, and other items related to administrative activities that support operating
(production) labor;
 Fixed costs, such as building services (safety, general
USING MEDIAN VS MEAN WAGE DATA
engineering, general plant maintenance, janitorial, cafeteria);
 Insurance costs, such as liability, property, and travel; and Whether the median or mean (i.e., average)
 Accounting profit, which reflects the opportunity cost of is the best central tendency estimate of
62
equity capital. compensation depends on the extent to
which the distribution is highly skewed for
Thus combining data on wages, benefits, and other overhead
workers in the occupations of concern. When
costs approximates the value of time from the employer’s
considering the overall population, the
perspective.63
average is significantly greater than the
Two data sources published by the U.S. Bureau of Labor Statistics median because of the small number of
(BLS) provide national information on hourly wages by industry people who are very highly compensated.
sector (see bottom of Table 4.2 for hyperlinks to data sources).64 Thus, if only a fraction of the U.S population
The Occupational Employment Statistics (OES) are generated from is affected by a regulation, the best estimate
a semiannual mail survey that covers a broad number of may be the median (which is the center of
establishments across the United States.65 The National the income distribution), rather than the
Compensation Survey (NCS) is an in-person survey of a subset of mean (which is closer to the upper tail of the
establishments and provides information on quarterly changes in distribution). However, if the entire
employer costs (the Employer Cost Index, or ECI) and cost levels population is affected, applying the mean
(Employer Costs for Employee Compensation, or ECEC).66,67 may be appropriate. Analysts should consider
the specific characteristics of the rule when
Both surveys use statistical methods to collect nationally
selecting the most appropriate measure.
representative samples. The OES survey is larger, covering a
greater range of occupations and geographic areas, and provides
estimates of median, as well as mean, wages.68 In contrast, the
NCS program samples fewer establishments, but conducts the survey in-person and collects more detailed
information on occupations within an establishment. In addition to reporting wage and salary information (pre-
tax, average only), the NCS provides data on other compensation, including benefits (paid leave, insurance,
retirement). Generally, OES is the preferred source for national estimates of hourly wages given its broader
geographic coverage. The ECEC is useful for identifying compensation rates for specific categories of employees
(e.g., managers).

62
“Accounting profit” is a different concept from “economic profit.” Firms require the investment of capital to operate and must provide a reasonable return
on that investment, or the capital would be put to other uses. Accounting profit is a measure of the return on this capital investment. Economic profit, in
contrast, equals sales revenue minus all costs, including the cost of equity capital. Under perfect competition, long-run economic profits are zero. See U.S.
Environmental Protection Agency’s Science Advisory Board (2007) for more discussion.
63
Publicly-available estimates of overhead costs may include fringe benefits; thus whether to separately include benefits will depend on the data sources used
in the analysis. The overhead rate discussed in this section is intended to be inclusive of fringe benefits; therefore it is applied to an estimate of wages that
does not include benefits.
64
We focus on data sources providing hourly wage data, as opposed to weekly, annual, or household estimates, to avoid the need for additional assumptions
about the number of hours worked and/or the number of employed workers in a household. If data on annual salaries is required, additional sources, such as
the U.S. Bureau of Labor Statistics’ Quarterly Census of Employment and Wages (QCEW, available at [Link] may also be used.
65
OES excludes farm establishments and self-employed persons.
66
See [Link] pages on “OES Frequently Asked Questions” for a comparison of the OES and NRC.
67
NCS excludes federal government employees.
68
Ideally, analysts would use estimates of the marginal wage rate (i.e., the increment paid for the last hour worked) rather than the average cost across all
hours worked. However, average or median values are generally used due to the lack of data on marginal rates.

29
Obtaining data on other overhead costs is challenging. Overhead costs vary greatly across industries and firm
sizes. In addition, the precise cost elements assigned as “indirect” or “overhead” costs, as opposed to direct
costs or employee wages, are subject to some interpretation at the firm level.
No readily available, national data exist on overhead rates by industry or sector. Data available in the ECEC
suggest that benefits average 46 percent of wages and salaries.69 Because this figure excludes overhead costs
other than benefits, it represents a likely lower bound on the overhead rate. In the private sector, analysts often
use a “rule of thumb” assumption that total overhead costs (benefits plus other overhead) equal 150 percent of
wages. As a an interim default, while HHS conducts more research, analysts should assume overhead costs
(including benefits) are equal to 100 percent of pre-tax wages (roughly the midpoint between 46 and 150
percent), and they should test the sensitivity of their results to alternative assumptions. Figure 4.1 provides an
example of this calculation.
F I G U R E 4 . 1 . S A M P L E C A L C U L AT I O N O F T H E VA L U E O F T I M E S P E N T O N A PA I D A D M I N I S T R AT I V E TA S K

Assume a proposed rule will result in five additional hours each year of administrative work for occupational therapy
assistants (OES Occupation Code 31-2011). The opportunity cost of the time spent undertaking these activities would
be calculated as follows:

Mean wages for occupational therapy assistants

(a)
(national estimate) (OES, May 2014): $27.53 per hour
Overhead cost per direct labor hour: 100 percent
Hours spent per employer: 5 hours
Opportunity Cost Per Employee: $27.53 * 2 * 5 = $275.30
(a)
Because all occupational therapy assistants employed throughout the United States will assume additional
administrative tasks in response to the regulation, use of the mean is appropriate in this example.

Unpaid activities: HHS regulations may also impose administrative burdens on individuals (e.g., filling out
additional paperwork for health care reimbursements) unassociated with their job. Unlike individuals employed
in the labor market, those engaged in nonmarket labor activities are not compensated. As a result, the rationale
for selecting a rate for valuing time spent performing such activities is less straightforward than for market
labor.
As discussed earlier, economists often assume that the marginal value of an hour of uncompensated activity is
equal to marginal compensation received. In other words, the opportunity costs of not working equal the value
of the compensation the individual would have received if he or she chose to work. This value is generally
estimated based on the post-tax wage an individual would have received for market work. This interpretation
applies both to people employed in the labor force, who (in principle) could adjust their working hours and
compensation, as well as to those out of the labor force, who presumably have chosen not to work because they
value their time more highly than the rate at which they would be compensated.70
To estimate the hourly value of unpaid administrative tasks, analysts should apply the post-tax wage rate. This
rate can be obtained by adjusting the pre-tax wage rates reported in the OES or NCS to remove taxes, which
vary as a percentage of wages over time and across locations.71

69
See [Link] Table 1. Civilian workers, by major occupational and industry group (September 2015). Total benefits
account for 31.4 percent of total compensation (wages and salaries plus benefits). To calculate the size of total benefits relative to wages and salaries, apply
the following equation: 31.4/(100 – 31.4) = 45.8 percent.
70
Analysts should also apply this approach when valuing time costs incurred by children or seniors (e.g., time spent at additional medical appointments),
noting related uncertainties.
71
National estimates of the Federal and State income taxes paid as a percentage of pre-tax income are difficult to obtain. The Internal Revenue Service (IRS)
provides detailed reports of total Federal income tax collected relative to adjusted gross income; however, these data exclude State taxes (see, for example,
“Individual Income Tax Rates and Tax Shares” at [Link]

30
To estimate the tax rate, including both Federal and state taxes, analysts should use data on household income
before and after taxes collected in the CPS, a joint effort by the U.S. Census Bureau (Census) and BLS. The CPS
collects data from a nationally-representative sample of 60,000 households on a monthly basis.72 The Census
maintains a tool called the “CPS Table Creator,” which allows analysts to create customized data tables.73 It
provides both mean and median income; as with wage rates, which central tendency estimate analysts should
use will depend on the specific characteristics of the rule.74 Figure 4.2 provides an example calculation of the
value of time spent on unpaid administrative tasks.
For both paid and unpaid work time, the representativeness of the wage and tax rate estimates is likely to be
uncertain. Where plausible alternative estimates exist, analysts should test the sensitivity of their results to
these assumptions (see Chapter 6), particularly if the alternative estimates significantly affect the analytic
conclusions.
Materials: Materials used to complete administrative activities may include office supplies or other items.
Generally, analysts should obtain cost estimates from readily-available office supply catalogs or websites and
courier services (e.g., the U.S. Postal Service, Federal Express, United Parcel Service, DHL). In addition, the rule
may generate a need for records storage, either electronically or on paper. If a substantial amount of data must
be stored, analysts should consider the costs of electronic file storage and backup, rent for additional storage
space, or the cost of filing cabinets or boxes.
Travel: Administrative costs may include travel, particularly where the new rule creates a need for meetings or
training activities. The U.S. General Services Administration (GSA) provides per diem travel rates for lodging and
meals.75 For air travel, plane fares can be obtained using internet travel search engines. For travel by car, the
Internal Revenue Service (IRS) publishes reimbursement rates for mileage.76 The mileage rate can be applied to
estimates of miles traveled obtained from internet websites providing travel directions. Analysts should also
include travel time, as discussed earlier.77

72
Household tax rates are appropriate because ideally individuals should make decisions based on the tax rates they actually pay.
73
To estimate mean or median household income before taxes, under “Data Options” select the relevant calendar year and get a count of “Persons-All.” Next,
“Define Your Table” by selecting “Household Income – Alternative” as a row variable. Under the “Statistics” section, in the subsection called “Additional
numeric variable statistics” choose “Household Income-Alternative” and “Mean” or “Median.” In the “Income Definition” section, select “Customize your own
income definition” and then select “1. Earnings (wages, salaries, and self-employment income)” and “19. Federal Earned Income Credit.” For household
income after taxes, follow the same steps and add the following additional selections in the customized income definition: “20. Federal Income Taxes after
refundable credits except EIC,” “21. State income taxes after all refundable credits,” and “22. Payroll taxes (FICA and other mandatory deductions).” For
2014 (select 2015 as the most recent year of data), median pre-tax household income ($53,000) minus post-tax income ($44,599) and divided by median pre-
tax income results in a median tax rate of 16 percent. (To access the CPS Table Creator, see [Link]
74
As with wage rates, ideally, analysts would use estimates of the marginal tax rate (i.e., the tax rate applied to the last dollar of income earned) to make this
adjustment, rather than the average tax rate paid for all income. While data on the distribution of marginal tax rates paid by the U.S. tax filers are available
from the IRS, they only include Federal taxes; excluding State or other taxes. Thus, analysts should use the CPS data, even though it provides mean or
median, rather than marginal rates, because it includes both Federal and State taxes.
75
See GSA’s website “Per Diem Rates Look-up” at [Link]
76
See IRS’s website “Standard Mileage Rates” at [Link]
77
See DOT (2015b) for recommended adjustment factors for the hourly estimates of value of time spent traveling, for different types of travel (available at:
[Link] That document provides a detailed discussion of the theoretical and empirical
basis for these adjustment factors.

31
 F I G U R E 4 . 2 . S A M P L E C A L C U L AT I O N O F T H E VA L U E O F T I M E S P E N T
O N A N U N PA I D A D M I N I S T R AT I V E TA S K

Assume a proposed rule will result in five additional hours each year of administrative work for a subset of affected
individuals (working and non-working adults, children, and seniors) in the United States. The opportunity cost of the
time spent undertaking these activities would be calculated as follows:

Median wages, all occupations

(a)
(OES, May 2014) = $17.09 per hour
Median household tax rate
(a)
(CPS, 2014 data) = 16 percent
Hours spent per individual = 5 hours
Opportunity cost per individual = $17.09 * [1-0.16] * 5 = $71.78
(a)
In this instance, the distribution of income among the subset of the population subject to the regulation may not be
representative of the U.S. income distribution. Therefore, the median may represent the best central tendency
estimate of wage and household tax rates for affected individuals.

4.2.2 CAPI TAL AND OP ERATIONS AND M AINTEN ANCE COSTS

Regulated entities may also need to purchase and operate new equipment to comply with regulatory
requirements. For example, they may need to purchase new computers and software, change equipment or
maintenance schedules at a production facility, or adopt other new technology. In this section, we describe
methods for estimating such costs.
Equipment and other capital components: Capital costs generally refer to the reallocation of resources needed
to purchase and operate additional equipment or other inputs that are not immediately consumed in the
production process. 78 Typical capital costs may include, for example: purchasing computers and software to
support administrative tasks; or installing or retrofitting new equipment associated with the production of food,
drugs, or other goods. Some regulations may lead to capital expenditures to acquire buildings or land.
Generally, analysts use market data to estimate the price of purchasing and installing such equipment. These
data may be obtained through interviews, literature reviews, review of online merchandise catalogues, or other
sources. In some cases, the cost of the equipment may include installation costs and it will not be necessary to
separately estimate the costs of associated labor. Otherwise, labor costs should be estimated in terms of the
fees paid to licensed installers or, if the work is completed in-house, using the approach for valuing paid time
described above. Information describing the useful life of the equipment is also necessary to determine whether
the equipment must be replaced during the time period of the analysis. Finally, a side cost often associated with
installation is the temporary shutdown of operations (i.e., forgone revenues net of avoided variable operations
and maintenance (O&M) costs). In many cases these costs are minimized by installing or retrofitting equipment
during regular downtimes (e.g., for maintenance).
Operations and maintenance costs: O&M costs include the annual costs of labor, utilities, and other resources
required to operate and maintain capital equipment, as well as other expenditures that do not involve the
purchase of a capital asset. Typical O&M costs include labor costs (discussed earlier); electricity and other
utilities; replacement parts; raw materials and other inputs to production. O&M costs may be variable, in that

78
Note that capital costs described in this section should not be confused with the fixed overhead component of the overhead rate used to estimate the value
of time (see Table 4.2). In the latter case, overhead costs are used as a proxy to estimate the value of time. This section, in contrast, describes the valuation
of additional equipment or other goods that may be necessary to implement a proposed regulation. Where an entity purchases new equipment (e.g., hard
drives) to store compliance information and shifts staff resources (without hiring additional staff) to undertake administrative tasks, time cost and capital
cost should be summed (they are not duplicative).

32
they fluctuate with production levels, or fixed, where the costs are not tied to production levels.79 Again,
analysts generally use market data to estimate such costs.
For both capital and O&M costs, analysts must be careful to estimate incremental costs. For example, if a firm
needs to purchase new and improved equipment to replace current machinery (or the machinery they would
purchase during their next scheduled turnover), the incremental costs of the rule include only the costs above
and beyond those associated with the equipment the firm would have otherwise purchased. Therefore, data are
required on the cost and useful life of both the existing equipment and the newer technology needed to comply
with the regulation.80

4.2.3 MEDICAL COSTS

Medical costs may be relevant to either the benefit or cost calculations depending on the characteristics of the
regulation. As noted in Chapter 2, costs are generally the inputs and benefits are the outputs or outcomes of a
policy. Thus if increases or decreases in medical costs are part of the implementation of the requirements (i.e.,
an input), they should be counted on the cost-side of the equation. If they are part of the intended outcome,
then they should be included in the benefit calculation, taking care to avoid double-counting with other benefit
measures. For some rules, whether medical costs or savings should be counted as a “cost” or “benefit” will be
uncertain, and analysts will need to discuss where they are counted in documenting the analysis. Medical costs
generally should be presented as a separate line item in the calculations so their treatment is clear.
The appropriate calculation of medical costs in benefit-cost analysis is an area where more work is needed,
because of the substantial distortions introduced by regulation of the health care sector and the effects of
government and private insurance reimbursement policies. These distortions drive a wedge between market
prices and opportunity costs, which make estimation difficult. Comparison across studies suggests that different
approaches can lead to noticeably different results (e.g., Bloom et al. 2001; Akobundu et al. 2006; Larg and Moss
2011), but there is no established set of recommended best practices. In addition, much of the available data
were developed to support reimbursement decisions and are not necessarily appropriate for estimating
opportunity costs. HHS is now undertaking a project to further explore this issue; in the interim analysts should
follow the general approach described below and discuss associated uncertainties.
When benefits consist of mortality and morbidity risk reductions, as discussed in Chapter 3, only some types of
costs should be added to the estimates of individual WTP used for valuation. More specifically, the value per
statistical life (VSL) estimates used to value mortality risk reductions, and the WTP estimates used to value
morbidity risk reductions (including the estimates of monetized QALYs) used as proxies when suitable WTP
estimates are not available), may include costs borne by the affected individuals. They presumably reflect the
effect of the risk reductions on the activities the individual undertakes (including the allocation of both work and
non-work time), and may also reflect out-of-pocket costs. Hence to avoid double-counting, savings in medical
costs are generally not added to these benefit values. The one exception is when the costs in the absence of the
regulation would be borne by third parties, in which case any savings in resource costs (excluding transfers) may
added.
When a regulation imposes costs on the health care sector, for example by establishing or changing
requirements for treatment, then medical costs may be included in the cost analysis. As a simple illustration,
assume that a regulation requires monitoring the health of all workers exposed to contaminants while cleaning
up after a natural disaster. The costs of the regulation would include the incremental cost of the medical

79
For example, variable costs, such as raw materials used as inputs to production, will rise or fall with production levels. Fixed costs, such as rent or utilities,
do not vary with production levels in the near-term.
80
Analysts should consider whether compliance costs may decrease over time as regulated entities gain experience with the new regulation. A significant body
of literature related to the operation and management of industrial processes suggests that the per-unit cost of producing or using a given technology
declines as experience with that technology increases over time (see Baloff 1971, Dutton and Thomas 1984, and Epple et al. 1991). For a review of the
literature measuring the “learning rate” for different industries and technologies, see Auerswald et al. (2000).

33
monitoring. The health benefits that result from earlier detection and treatment (than in the absence of such
monitoring) would be valued using the approaches discussed in Chapter 3.
Analysts must also consider whether the cost assessment requires prevalence-based or incidence-based per
case estimates. The former typically reflect the average costs of all cases in a given year, and may be appropriate
for short-lived effects, such as acute health conditions or time-limited monitoring and treatment programs (e.g.,
in the immediate wake of a natural disaster). Incidence-based estimates instead track or model the lifetime
costs per case, and are desirable when the regulation affects the incidence of chronic conditions or longer-term
monitoring and treatment programs. In these cases, costs are likely to fluctuate over time, and extrapolating
lifetime costs from prevalence-based estimates may understate or overstate actual costs. Incidence-based
estimates that consider the entire, multi-year progression of the disease may be preferable. The appropriate
measure will depend on the data available as well as the nature of the health effect.
Analysts will need to review the existing literature for recent studies of the specific health effects and types of
costs needed for a particular regulatory analysis. Akobundu et al. (2006) and Larg and Moss (2011) provide
useful overviews of the characteristics and limitations of different measurement approaches applied by
researchers. Lund et al. (2009) provide a comprehensive inventory of relevant data sources. In addition, analysts
should consider contacting health economists who focus on the conditions of interest, such as technical experts
at the Centers for Disease Control and Prevention, academic institutions, or nonprofit research organizations.
The series of articles included in Yabroff et al. (2009) also provide useful information.
In summary, estimating medical costs requires substantial professional judgment; the appropriate approach will
depend on the characteristics of the practices affected by the regulation as well as the available data. Analysts
are encouraged to work with subject matter experts if they are unfamiliar with methods used to estimate
medical costs or with the particular health effect of interest.

4.2.4 GOVERNM ENT IMP LEMENTATION COSTS

Government entities may also incur costs, either as an implementing or regulated entity. For example, a
regulation may impose new review, reporting, and record keeping requirements on State or local government
entities responsible for recording vital statistics, such as births and deaths. In this example, the HHS may incur
implementation costs related to developing guidance, conducting training, and increasing enforcement.
Likewise, State and local governments may incur compliance costs to train staff, adjust their electronic
databases and reporting systems, and alter how they store information.
If the government is involved in implementing the regulation, the costs to the agency represent an opportunity
cost of the regulation, as do similar costs imposed on industry, even if no new staff are hired. The effort
undertaken to implement the regulation would otherwise be spent on other productive tasks. Thus, these costs
should be counted in the analysis, using the methods discussed above. Information included in internal budget
estimates, such as full-time equivalent labor needed for the program or requests for capital expenditures, are
useful sources of data for these cost estimates.
If the government is the subject of the regulation, estimating related costs also follows the same approaches
described elsewhere in this chapter. If grants or other funding are provided by HHS to support implementation
of the regulation by industry or others, these funds are transfers and should not be included as costs, assuming
they have no behavioral impacts that could affect the estimates of national net benefits. However, the amount
of the funding may serve as a proxy estimate of the compliance costs imposed by the rule, to the extent that
related costs are fully covered.

34
4.3 ESTIMATING MARKET-LEVEL IMPACTS
The preceding sections assume that the proposed regulation will not significantly affect the quantity of goods
produced (e.g., a new regulation resulting in increased costs to electronically store and transmit data related to
certain medical procedures will not affect the quantity of procedures performed). When the regulation is
anticipated to affect the quantity of goods produced, a more precise estimate of social costs would involve
estimating changes in consumer and producer surplus (see Appendix B) using partial or general equilibrium
models.
Where a single market or a small number of unconnected
markets are affected, partial equilibrium analysis provides a WHEN SHOULD ANALYSTS USE PARTIAL OR
useful tool for estimating welfare changes. Analysts use GENERAL EQUILIBRIUM MODELS?
information about the quantity and price of goods produced Analysts should consider employing partial or
without the regulation, compliance costs, and elasticities of general equilibrium models when changes in
supply and demand to estimate the equilibrium output with the consumer and producer surplus are likely to
regulation and net changes in consumer and producer surplus significantly affect the analytic conclusions.
(see Appendix B and Boardman et al. 2011).81 In addition to For example, such effects might be important
providing a more precise estimate of welfare changes, a partial if large sectors of the U.S. economy are
equilibrium model also provides insight into who bears the cost affected or if impacts are likely to be
of the regulation. Such information may be important if analysts measurable at a national scale (e.g., relative
anticipate the regulation will have significant distributional to U.S. gross domestic product, GDP). In most
impacts (see Chapter 7).82 cases, estimating compliance costs is a
Where multiple, interconnected markets are affected, or sufficient proxy for changes in surplus.
substantial international effects are anticipated, analysts might
consider using computable general equilibrium analysis to
estimate impacts. Such modeling may also be useful when a
regulation is part of a larger suite of regulations that may have economy-wide, interactive effects. These models
measure shifts in production and consumption resulting from compliance costs. In addition, they estimate how
shifts in quantity or price in one market affect related markets. General equilibrium models are complex and
generally require a significant amount of data to capture the effects of a regulation (see Berck and Hoffman
2002 and Lofgren et al. 2002). Such analysis requires working with a pre-existing model developed by one of
several academic, government, or other institutions.83

81
Because compliance costs serve as the basis for changes in the supply curve, analysts should not combine separate estimates of total compliance costs with
estimates of changes in surplus. Adding these cost estimates together would result in double-counting.
82
For example, for products where consumer demand is relatively inelastic, producers may have greater ability to pass compliance costs on to consumers in
the form of higher prices.
83
Examples of computable general equilibrium models used by Federal agencies to estimate impacts to the U.S. economy include the Global Trade Analysis
Project (GTAP) model, the USAGE model, the Intertemporal General Equilibrium Model, and EMPAX-CGE.

35
Chapter 5
Account for Timing
The costs, benefits, and other impacts of regulations often accrue over several years, requiring that analysts take
into account how affected individuals value impacts that occur in different time periods. In addition, RIAs
generally involve applying data that reflects past rather than current price levels. Thus analysts must both inflate
prices from prior years to the same dollar year, and discount future impacts back to the base year in which the
regulation is first implemented.84 Carrying out these steps involves distinguishing between inflation and real
changes in value, and understanding how to appropriately account for time preferences. Below, we first discuss
the underlying concepts and basic approach, then describe how to adjust for inflation, calculate discounted
present values, and determine impacts on an annualized basis.

5.1 BASIC CONCEPTS AND APPROACH

Consider the four streams of payments in Table 5.1. If they represent the net benefits of different policy options
in each year, how might we choose among them? Options A, B, and C sum to the same total over the 10 year
period, but the net benefits vary across years. Under Option A, costs substantially exceed benefits for two years
after which benefits exceed costs; under Option B, costs also initially exceed benefits but by lesser amounts and
for a longer period; under Option C, benefits exceed costs by the same amount in all years. Option D sums to a
smaller total with net benefits that decline over time. Comparing such streams of payments, regardless of
whether they represent costs, benefits, or net benefits, requires (1) understanding whether they include the
effects of inflation and (2) addressing time preferences through discounting. In combination, considering these
issues allows us to determine which option is preferable.
TA B L E 5 . 1 . C O M PA R I N G U N D I S C O U N T E D A N N U A L N E T B E N E F I T S

YEAR 0 1 2 3 4 5 6 7 8 9 TOTAL

Option A ($2,000) ($1,000) $200 $300 $400 $500 $600 $600 $700 $700 $1,000
Option B ($600) ($500) ($400) ($300) $100 $200 $300 $600 $700 $900 $1,000
Option C $100 $100 $100 $100 $100 $100 $100 $100 $100 $100 $1,000
Option D $200 $200 $200 $50 $50 $50 $50 $50 $50 $0 $900

The first question is whether values are measured in real or nominal dollars. Observed prices are measured in
nominal (current-year) dollars. Because these prices may be affected by economy-wide inflation, values in
different years are not necessarily comparable. If there is inflation, the quantity of goods one can buy for $1.00
decreases over time. Real (constant or inflation-adjusted) dollars net-out the effect of inflation so that dollars
have equal purchasing power over time and are comparable across different periods.
To avoid misleading comparisons, regulatory analyses should always be conducted in constant (real, inflation-
adjusted) dollars.85 This approach has the advantage of allowing analysts to avoid the difficult task of attempting
to project future inflation rates. As discussed in more detail in section 5.2, all values should be first converted to
the same year dollars, then the analysis should be conducted in real dollars from that point forward. For
example, a regulatory analysis prepared in 2015, which projects benefits and costs over the next 10 or 20 years,

84
As discussed in more detail later, the dollar year is likely to differ from the base year used for discounting.
85
For the remainder of this discussion, we assume that the values in Table 5.1 are undiscounted and expressed in real terms.

36
may be conducted in constant 2014 dollars. Generally, a dollar year should be selected that is reasonably close
to the current year.86
The second question is how to weight real benefits and costs that accrue in different time periods. There are
two interrelated reasons why values are not likely to be weighted equally over time. One is individual time
preferences; people generally prefer to receive benefits as soon as possible and to defer costs. The other is
opportunity costs; resources received today can be invested to yield a positive return while resources expended
are no longer available for investment.87
Analysts account for the effects of timing by discounting future impacts to the base year of the analysis. This
base year commonly reflects the first year in which the regulation is implemented, and is likely to differ from the
dollar year selected for the analysis. For example, an analysis
conducted in 2015 may express all values in 2014 dollars. HOW SHOULD ANALYSTS ACCOUNT FOR
However, if the rule will not be implemented until 2017, the THE EFFECTS OF TIMING?
analysts may use 2017 as the base year for discounting. The dollar
year and the base year must be clearly identified throughout the Analysts should first select a common dollar
analysis. year, and inflate all unit values to that year.
They should then calculate the benefits,
Although there are conceptual differences between a discount costs, and net benefits expected to accrue in
rate and an interest rate, they are closely related. An interest rate each future year of the analysis, report the
is the market rate at which money can be borrowed or loaned, undiscounted stream of benefits and costs,
and results from the interaction between market participants’ and report their present value applying
willingness to save and demand for borrowing. The discount rate discount rates of 3 and 7 percent. Analysts
reflects preferences for receiving benefits or bearing costs at should also report annualized values
different dates, which may be influenced by the opportunity costs (calculated using each discount rate). The
imposed by regulations or by similar actions that divert resources base year used when calculating present
from other investments or consumption. In practice, discount values should be the year in which the
rates are often based on market interest rates. regulation is initially implemented, and may
As in the case of prices, care must be taken to distinguish differ from the dollar year.
between nominal and real discount rates. Because regulatory
analyses are conducted in real dollars, a real discount rate must
be applied. Below, we first discuss how to adjust for inflation. Section 5.3 then discusses discounting in more
detail, and Section 5.4 describes how to convert discounted amounts to annualized dollars.

5.2 ADJUSTING FOR INFLATION

In regulatory analysis, analysts often work with data from many different time periods; an analysis conducted
this year is likely to rely on unit cost and benefit data collected in several previous years. Adjusting for inflation
involves using an index to convert all dollar values to the same year dollars. Indices commonly used to reflect
economy-wide trends are the Consumer Price Index (CPI), and the gross domestic product (GDP) implicit price
deflator. Because the CPI is more easily accessible, it is more frequently applied.88 In general, the two
approaches yield similar estimates of the inflation rate.

86
A year prior to the current year is generally used as the dollar year because the rate of inflation for the current year is not yet known.
87
The two reasons are related. When real interest rates are positive, individuals can purchase more goods and services if they postpone those purchases by
saving more or borrowing less. To maximize utility (well-being), they should allocate their spending over time such that their preference for incremental
current over future spending equals the interest rate. Real interest rates are typically positive because individuals require compensation for deferring
consumption.
88
As discussed in Chapter 8, the GDP deflator must be used in preparing the accounting statement required under OMB Circular A-4. It is also used to
determine the threshold for conducting analyses under the Unfunded Mandates Reform Act as discussed in Chapter 7 (see, for example, HHS 2015). However,
within the analysis itself, the CPI (or more specialized indices) may be used instead of the GDP deflator to adjust benefits and costs to the same year dollars.

37
Other, more specialized indices are also available that reflect price trends in particular market segments (such as
producer prices or medical services) or in particular geographic areas. In this section, we focus on the CPI and
GDP deflator, since these are the indices most commonly used in regulatory analysis. However, in some cases
analysts may instead apply more specialized indices. The inflation index or indices used, and the rationale for
applying them, must be clearly documented in the RIA.
The CPI is developed by the Bureau of Labor Statistics within the U.S. Department of Labor. It measures the
average change over time in the prices paid by urban consumers for a market basket of goods and services.
(According to the Bureau of Labor Statistics, these urban consumers represent about 87 percent of the U.S.
population.89) The CPI is based on detailed information on actual expenditures by a statistically-representative
sample of individuals and families, including all consumption goods and services.
The CPI website includes an inflation calculator ([Link] that can be
easily used to convert values to the same year dollars, based on purchases of all goods and services nationally. If
an analyst prefers to directly apply the index values from the CPI tables (for example, including these values in a
spreadsheet used to calculate benefits and costs), the index values must be converted to reflect the rate of
change, expressed as a proportion or percentage. For example, if the analyst wishes to inflate a value from
dollar year “a,” in which the index was 120, to a dollar year “b,” in which the index was 140 (a 20 point
difference), the increase would be (20/120) * 100 = 17 percent.90 Thus a unit cost of $100 in year “a” dollars
would become $117 if expressed as year “b” dollars.
More generally, to inflate a value from year “a” to year “b,” the percentage change is calculated as:
Percent inflation (CPI) =

((CPIyear b – CPIyear a) / CPIyear a) * 100

The GDP deflator is instead based on the value of all goods and services produced within the U.S. economy; it
also can be calculated for subsectors of the economy. It is developed by the Bureau of Economic Analysis in the
U.S. Department of Commerce, and includes personal consumption, domestic investment, net exports, and
government consumption and investment.91 It is not derived from a market basket of goods; rather it changes
depending on investment and consumption patterns.
The GDP deflator is provided in Table 1.1.9 of the National Income and Product Accounts, which can be accessed
through the Bureau of Economic Analysis website.92 Again, as in the case of the CPI, the index values need to be
converted to a proportional or percentage change to be applied in the analysis. This conversion follows the same
formula as provided above for the CPI.

5.3 DETERMINING PRESENT VALUES

Once all unit benefits and costs are converted to the same dollar year (i.e., to constant dollars) and the year in
which they occur is identified, the next step is to calculate their discounted present value. This value indicates
how much dollars paid or received at a later time are worth in the base year (i.e., the year in which the
regulation is first implemented), given time preferences and opportunity costs as discussed earlier.93 For

89
This and other basic information on the CPI is available at [Link]
90
The change can also be expressed as a multiplier, applying the formula CPIyear b / CPIyear a (140/120 = 1.17 percent in the example). In spreadsheet analysis,
converting the proportion into a percentage is not necessary; the analyst may simply enter the proportion and multiply the year “a” value by the result.
91
A glossary of related terms is available at: [Link] more information on the underlying concepts and methodology is
available at: [Link]
92
To access this table: (1) click on the “Interactive Data” tab at the top of [Link] (2) select “GDP & Personal Income” under “National Data;”
(3) click on “Begin Using the Data;” (4) under “National Income and Product Account Tables,” click on Section 1, and select T able 1.1.9, “Implicit Price
Deflators;” (5) click on the “Options” icon to choose the desired time period and to indicate annual as the frequency, then select “Update” to regenerate the
table.
93
In some cases, regulated entities may begin to respond to the regulation before it becomes effective, and related costs, benefits, and net benefits will need
to be carried forward to the base year rather than discounted. In this case, their value will increase rather than decrease between the time when they are
incurred and the base year.

38
regulatory analysis, the OMB guidance in Circular A-4 (2003) requires agencies to report the results of their
analyses applying discount rates of three and seven percent per year.94 The use of two rates reflects uncertainty
about whether regulation is likely to displace investment or consumption.95 In a simple theoretical model,
investment- and consumption-based discount rates would be equal, but in reality distortions such as taxes lead
to differences.
The seven percent rate is intended to reflect the opportunity costs associated with displacing private
investment, and was based on the estimated average before-tax rate of return to private capital in the U.S.
economy at the time when the OMB guidance was developed. The three percent rate is intended to reflect the
opportunity costs associated with displacing consumption (often referred to as the marginal “social rate of time
preference”), and was based on the before-tax rate of return on long-term government debt to approximate the
interest paid on savings. This approach assumes that the savings rate represents the average by which
consumers discount future consumption. Both are real rates, consistent with the use of real dollars when
estimating benefits and costs.
The formulae for calculating present values are provided in Figure 5.1.
F I G U R E 5 . 1 . C A L C U L AT I N G P R E S E N T VA L U E S

If:
• PV = present value as of the base year
• FVt = future value in the year (t) when the benefit or cost accrues
• NPV = net present value of benefits and costs combined across all time periods
• r = the discount rate
• t = the number of years in the future (measured from the base year) when the cost or benefit accrues
• n = the number of years included in the analysis

Then the discount factor for costs or benefits that accrue at the end of year t is:
t
1/(1+r)

The present value of a future cost or benefit that accrues in year t is:
t
PV = FVt (1/(1+r) )

The net present value for a stream of future benefits and costs is:
2 3 n
NPV = Vt=0 + (FVt=1 /(1+r) ) + (FVt=2 /(1+r) ) + (FVt=3 /(1+r) )...(FVt=n /(1+r) )

Most spreadsheet programs automate these calculations, as do many calculators. In Excel, the function is NPV(r,
[range or list of cells with flows ordered from “now” to the last period]).96 Financial calculators typically have an
NPV function into which you can enter a stream of costs, benefits, or net benefits as well as a discount (interest)
rate. While in Excel r should be entered as a decimal (e.g., 0.03 if the discount rate is three percent), in many
calculators r instead must be entered as a percentage (e.g., 3). The Excel function also has some optional
arguments, such as whether the payments occur at the start or end of each period. While the end of the period

94
While OMB allows agencies to apply other rates if justified, in practice agencies usually apply only the three and seven percent rates for intra-generational
impacts. Discounting inter-generational impacts (for policies such as those addressing climate change or radioactive waste storage) raises several difficult
issues related to forecasting future preferences and opportunity costs as well as inter-generational equity. HHS analysts rarely need to address these concerns
because most HHS analyses cover shorter time periods; i.e., 10 to 20 years as noted earlier in this this guidance. OMB Circular A-4 (2003) provides more
discussion of these issues.
95
On occasion, it may be informative to estimate the internal rate of return, which is the discount rate at which benefits equal costs (i.e., the net present
value is zero). Calculating the internal rate of return is generally not useful for selecting among regulatory alternatives, however. A policy may have more
than one internal rate of return if net benefits change from positive to negative (or vice-versa) more than once over the time period addressed. In addition,
as is the case for both benefit-cost and cost-effectiveness ratios, the internal rate of return is not sensitive to scale. It does not indicate the amount by which
benefits exceed costs, and hence does not provide information on which policy maximizes net benefits when policies differ in size.
96
Excel also provides a present value function [PV(r, n, payment per period)] that is useful when the values are the same in each period.

39
is the Excel default, analysts often instead assume that payments occur at the beginning of each period, which
means that the impacts in the base year (year “0” in the examples) are not discounted. In this case, a value of
“1” must be entered into the Excel formula under “type,” to change the default from the end to the beginning of
each period.97
As discussed in Chapter 2, benefits and costs generally should be assessed over a 10-to-20 year period,
consistent with the OMB guidance, unless the policy terminates sooner. Analysts should select a period that is
adequate to encompass the time needed for the regulation to become fully effective, without requiring
extrapolation so far into the future that predicting impacts become highly speculative given changes in the
population, economy, technology, and other factors. Impacts further in the future often add relatively little to
the present value of benefits and costs, given the effects of discounting, and are unlikely to alter the policy
implications of the analysis. However, longer time periods may be considered if clearly justified.
We can use the streams of undiscounted net benefits in Table 5.1 to provide an example of this process. First,
consistent with the OMB guidance in Circular A-4, analysts should present the stream of undiscounted costs,
benefits, and net benefits (as illustrated for net benefits in Table 5.1), to aid decision-makers in understanding
the timing. Presenting these graphically is often useful, as illustrated in Figure 5.2 for Option A from Table 5.1.

F I G U R E 5 . 2 . E X A M P L E P R E S E N TAT I O N O F U N D I S C O U N T E D N E T B E N E F I T S

Second, OMB requires that the results be presented using different discount rates, as illustrated in Table 5.2.
This table presents the same four streams of net benefits as Table 5.1, undiscounted as well as discounted
applying rates of the three and seven percent. Calculating present values makes it clear that the preferred
option depends on the discount rate. Without discounting, Options A, B, and C all appear preferable to Option
D. Discounted at a three percent rate, Option C is the best option. If the discount rate is seven percent, then
Option D becomes best. At a seven percent rate, the net benefits of Option A also become negative. Thus at this
rate, Option A would not be preferred to the “no action” baseline even if it were the only option being
considered, since its costs exceed its benefits.

97
While typically impacts incurred in the base year are not discounted, assuming payments occur at the beginning of each period, for some regulations analysts
may find that it is more appropriate to assume end-of-period payments. In that case, base year impacts should be discounted and the Excel default
assumption is appropriate. In the RIA, analysts should report the timing assumption used and the same assumption should be applied throughout the analysis.

40
 TA B L E 5 . 2 . C O M PA R I N G D I S C O U N T E D N E T B E N E F I T S

YEAR 0 1 2 3 4 5 6 7 8 9 NPV

Undiscounted
Option A ($2,000) ($1,000) $200 $300 $400 $500 $600 $600 $700 $700 $1,000
Option B ($600) ($500) ($400) ($300) $100 $200 $300 $600 $700 $900 $1,000
Option C $100 $100 $100 $100 $100 $100 $100 $100 $100 $100 $1,000
Option D $200 $200 $200 $50 $50 $50 $50 $50 $50 $0 $900
Discounted to Year “0” at 3 Percent
Option A ($2,000) ($971) $189 $275 $355 $431 $502 $488 $553 $536 $358
Option B ($600) ($485) ($377) ($275) $89 $173 $251 $488 $553 $690 $506
Option C $100 $97 $94 $92 $89 $86 $84 $81 $79 $77 $879
Option D $200 $194 $189 $46 $44 $43 $42 $41 $39 $0 $838
Discounted to Year “0” at 7 Percent
Option A ($2,000) ($935) $175 $245 $305 $356 $400 $374 $407 $381 ($292)
Option B ($600) ($467) ($349) ($245) $76 $143 $200 $374 $407 $490 $28
Option C $100 $93 $87 $82 $76 $71 $67 $62 $58 $54 $752
Option D $200 $187 $175 $41 $38 $36 $33 $31 $29 $0 $770

As demonstrated by Table 5.2, the choice of a discount rate can have a significant effect on the estimated net
benefits. Whether the discount rate will affect the conclusions of the analysis will depend on the pattern of
benefits and costs over time for each alternative considered. The option that provides the largest net benefits
will depend on the magnitude of the impacts and their timing, as well as on the discount rate. Generally, the
decision rule is that if only one policy is considered, then the policy should be implemented if the present value
of net benefits is greater than zero. For regulatory analyses, which should consider multiple options (as
discussed in Chapter 2 of this guidance), the option that is preferable in terms of economic efficiency will be the
option with the largest net benefits, as long as the net present value is greater than zero.

5.4 ANNUALIZING IMPACTS

For regulatory analyses, OMB Circular A-4 (2003) also requires that analysts present benefits, costs, and net
benefits on an annualized basis to facilitate comparisons across analyses that cover different time periods. The
annualized value of a stream of benefits, costs, or net benefits is the constant annual amount that, if maintained
for the same number of years as the initial stream, has the same present value. In other words, annualization
spreads the costs, benefits, or net benefits equally over the time period assessed, taking the discount rate into
account. The concept is similar to amortization of a loan, in which the principal and interest are paid through a
series of constant payments.
The formula for annualization is provided in Figure 5.3; the expression in brackets transforms a value into an
annuity of n years at a discount rate r. Note that applying this formula requires first estimating the present
value, following the formulae in Figure 5.1 as discussed in the preceding section.

41
 F I G U R E 5 . 3 . C A L C U L AT I N G A N N U A L I Z E D VA L U E S

If:
• PV = net present value of costs, benefits, or net benefits
• r = the discount rate
• n = the number of years included in the analysis
• AV = annualized value
The annualized value is:
n n
AV = PV * [(r * ( 1 + r) ) / (( 1 + r) - 1)]

Once a present value is calculated, it can be easily converted to an annualized value using spreadsheet
software or a financial calculator. In Excel, the function is PMT (r, nper [number of periods], and PV). Because
the PMT function is designed to calculate loan payments, it will provide a value with the opposite sign of the
present value; simply reversing the sign will provide the correct amount for the purpose of regulatory
analysis. OMB’s 2011 Regulatory Impact Analysis: Frequently Asked Questions provides more detailed, step-by-
step guidance on these calculations.
The annualized value is an alternative method for expressing the net benefits; the ranking of policies by
annualized value will be the same as the ranking by present value net benefits when estimated over the same
time period. To illustrate, in Table 5.3 we provide the results for the same streams of net benefits as assessed in
Table 5.2. The conclusions are the same: Option C has the largest annualized value under a three percent rate;
while Option D has the largest annualized value under a seven percent rate. If, however, these options were
implemented over different time periods, the results could vary.
TA B L E 5 . 3 . C O M PA R I N G A N N U A L I Z E D N E T B E N E F I T S

OPTION ANNUALIZED
Undiscounted
Option A $100
Option B $100
Option C $100
Option D $90
Discounted at 3 Percent
Option A $41
Option B $58
Option C $100
Option D $95
Discounted at 7 Percent
Option A ($39)
Option B $4
Option C $100
Option D $102

Because annualization provides a different perspective than the estimate of net present values, both annualized
and present values should be reported in the RIA along with information on the time period over which these
measures are calculated. The annualized value measures the average flow over the years included; the net
present value measures the total. Annualized estimates are also needed to complete the accounting statement
that must be submitted to OMB along with the RIA, as discussed in more detail in Chapter 8.

42
Chapter 6
Address Uncertainty and Nonquantifiable
Effects
Any analysis involves uncertainties, including difficulties related to quantifying some potentially important
effects. The challenge for the analyst is to determine how to best assess or quantify these uncertainties to
support decision-making. The goal is to ensure that decision-makers and other stakeholders understand the
extent to which key uncertainties – in the data, models, and assumptions – affect the main analytic conclusions.
For example, if the agency’s best estimates suggest that benefits exceed costs for a particular regulatory option,
how likely is it that this conclusion would be reversed given uncertainty about the magnitudes of the quantified
effects and the potential impact of nonquantified effects? Might these uncertainties affect the relative rankings
of the policy options? Answering these questions requires quantifying impacts to the greatest extent possible,
and identifying key uncertainties and exploring them in both quantitative and qualitative terms. Over time,
analysts should work to reduce these uncertainties and minimize the types of effects that cannot be quantified,
by anticipating future analytic needs and investing in research that will be useful across a variety of regulatory
analyses.
This chapter discusses strategies for characterizing the uncertainty in quantified effects as well as the potential
impacts of nonquantified effects. It focuses on the benefit-cost analysis, as discussed in the prior chapters, but
the approaches it describes are applicable to the supplemental analyses discussed later in this guidance as well.
As with other analytic components, the uncertainty analysis is often iterative; the initial analysis may lead to
decisions to conduct more research or to change the assumptions used, and perhaps to explore other policy
options.
Although the assessment of uncertainty (including nonquantified effects) may be described along with the
analytic methods when documenting the RIA (see Chapter 8), it is often helpful to summarize key uncertainties
in a separate section. For example, the chapter describing the benefits analysis could first describe the analytic
approach, then present the results, and conclude by discussing uncertainty and its implications. The executive
summary, and the chapter that compares costs to benefits, could consolidate the most important findings from
the individual chapters and describe how the uncertainties affect the overall conclusions.

6.1 CHARACTERIZING UNCERTAINTY IN QUANTIFIED EFFECTS

The data and models used to estimate costs, benefits, and other impacts inevitably involve limitations. These
may relate to the quality of the methods used to collect the data, the extent to which the data address the same
population, industries, or geographic area as the regulatory impacts, and the degree to which conditions may
change between when the data were collected and when the regulation is implemented. In addition, the models
used in the analysis, which may range from simple formulae to complex computer simulations, involve making
assumptions about the relationships between various factors. All analyses require predicting how those affected
will respond to the regulation, which adds to the uncertainty. The challenge for the analyst is to clearly describe
(in qualitative and quantitative terms) the uncertainties related to the data, models, and assumptions in a way
that aids decision-makers in understanding the confidence they should have in the results and the likely
direction and magnitude of any bias.

6.1.1 BASIC CO NCEPTS

Conceptually, one should distinguish uncertainty and variability. Variability refers to heterogeneity; for example,
differences in the ages of those affected by a regulation. While variability can be described by statistical
measures such as the standard deviation, it may be difficult to characterize precisely given that data may be

43
available for only a small (and perhaps non-representative) sample of those affected or for a limited geographic
area or time period. The usual measure of uncertainty about a parameter when estimated from a sample of the
population ("sampling variability") will be larger when there is more variability in the population (if there were
no variability, even a small sample would yield an exact estimate of the parameter).98
In contrast, uncertainty describes lack of knowledge. For example, data on the relationship between exposure to
a pathogen and the risk of mortality may be available for only a particular age group, and the agency may be
uncertain whether individuals of different ages would respond similarly to the exposure. Variability is a
characteristic of the real world that cannot be reduced by research (although research can lead to a better
understanding of variability). In contrast, uncertainty concerns lack of knowledge and can be reduced by
research.
Regulatory analysts often lack the time and resources needed to engage in substantial new primary research,
and must determine how to best target their efforts. Such targeting requires using screening analysis (see
Chapter 2) to identify areas where more work will have the most important implications for decision-making.
Analysts must then determine how to best combine the available data with reasonable models and assumptions
to characterize regulatory impacts. The limitations and uncertainties in these data, models, and assumptions
must be clearly disclosed in the RIA.
The requirements in OMB Circular A-4 (2003) encompass both variability and lack of knowledge when discussing
treatment of uncertainty. OMB urges analysts to fully disclose any uncertainties inherent in the analysis and to
evaluate and justify their analytical choices. OMB cautions that, at times, uncertainties may be significant
enough to warrant delaying a decision until more information can be collected and assessed. This is especially
true in situations where uncertainties have a significant effect on which regulatory decision appears to be best.
When considering whether to recommend a delay, analysts must take into account both costs (e.g., of further
data gathering efforts) and benefits (e.g., of the knowledge likely to be obtained from the new data). Delay may
also have consequences for social welfare (for instance if it allows dangerous practices to continue), which must
also be considered along with the impacts of any interim protective measures. If the timing of the regulation is
determined by statute or court order, delay may not be possible.

6.1.2 GENERAL APPROA CH

There are many options for addressing uncertainty in quantified effects. In Circular A-4, OMB outlines three
approaches with increasing levels of complexity: qualitative discussion, numerical sensitivity analysis, and
probabilistic analysis. These three methods are summarized in Table 6.1 and described in more detail below.
Additional information on these approaches is provided in Morgan and Henrion (1990), Boardman et al. (2011),
and other texts.

98
Statistical or sampling variability is the variability in a statistical estimate that results when the estimate is calculated from a sample, not the full
population. For example, the average height in the sample may not equal the average height in the population because a disproportionate number of tall
people were sampled by chance.

44
 TA B L E 6 . 1 . A P P R O A C H E S F O R A D D R E S S I N G U N C E RTA I N T Y I N Q U A N T I F I E D E F F E C T S

APPROACH APPLICABILITY CONDUCT

 For all analyses.

 May suffice if: Disclose key assumptions
– the rule involves annual economic effects less than $1 billion; and uncertainties and
Qualitative
include information on the
Discussion – the analyst is able to demonstrate that the results are robust
implications for decision-
to uncertainties; and, making.
– the consequences of the rule are modest.
 For rules involving annual economic effects less than $1 billion,
where: Vary one or many
Numerical parameters to calculate
– the qualitative discussion raises questions about the
Sensitivity Analysis distinct sets of results for
robustness of the results; or, comparison.
– the consequences of the rule are large.
 For rules involving annual economic effects of $1 billion or more Develop distributions for the
(required). uncertain parameters and
Probabilistic
conduct Monte Carlo
Analysis  For rules with smaller impacts where numerical sensitivity
analysis to determine the
analysis raises questions about the robustness of the results. distribution of the results.

Qualitative discussion of uncertainties: Qualitative discussion is the least rigorous approach, but is of significant
importance. It should always be included in the RIA. This approach involves disclosing key assumptions and
uncertainties and including information on the implications. To the greatest extent possible, the qualitative
discussion should include both the likely direction of the potential bias (i.e., whether the assumption may lead
to an under- or over-estimate of the impacts) and the likely magnitude of the effect (e.g., whether it is major or
minor). Such information will help decision-makers and others better understand the implications of the
analysis.
Numerical sensitivity analysis: Numerical sensitivity analysis allows the analyst to explore the effects of varying
the values of key parameters and is often useful to determine whether uncertainty about particular components
or assumptions may substantially affect the analytic result, as well as when data limitations or constrained
resources prevent full probabilistic analysis. Sensitivity analysis can be conducted by: (1) by changing one
variable or assumption at a time and calculating a new set of estimates (sometimes referred to as “partial
sensitivity analysis”); or (2) by varying several variables simultaneously to learn more about the robustness of
the results to widespread changes.
When conducting partial sensitivity analysis, it is generally infeasible to test all assumptions. Attention should be
devoted to analyzing those assumptions or variables that are most important (in that they may have the
greatest effect on the result) or are most uncertain. The analyst should vary key parameters one at a time using
plausible alternative values while holding all other parameters constant. Partial sensitivity analysis can be
conducted as a breakeven, or threshold, analysis; for example, where the analyst seeks to find the value of one
key parameter at which quantified benefits equal costs (i.e., net benefits equal zero), as discussed later in this
chapter.
Varying a combination of parameters simultaneously may obscure the effect that a single variable or assumption
has on the estimates, but can be particularly useful when a group of parameters are closely related (e.g.,
changing demographics and participation in the labor market) or when conducting a bounding analysis. In a
bounding analysis, the most- or least-favorable assumptions are selected to calculate best- or worst-case results.
These two sets of results represent high-end and low-end estimates that bound the primary results of the
analysis. However, care should be taken in conducting and interpreting this type of analysis, because it is

45
extremely unlikely that all of the parameters will simultaneously be at their highest or lowest values. Thus the
outcome of an analysis that uses lower (or upper) bound estimates for all parameters is very improbable.
If the sign of the net benefits or the relative ranking of the regulatory alternatives does not change in response
to sensitivity tests, analysts and decision-makers can conclude that the results are relatively robust and have
greater confidence in them. Otherwise, the analyst should (1) further investigate whether it is likely that the
alternative assumptions are more appropriate than the assumptions used in the original analysis; and
(2) conduct more rigorous probabilistic analysis if possible.
Probabilistic analysis: Probabilistic analysis is generally most informative because it quantifies the likelihood
that different results will occur. However, in some cases such analysis may not be warranted or feasible given
data limitations and constrained time and resources. OMB Circular A-4 indicates that probabilistic analysis “is
appropriate for complex rules where there are large, multiple uncertainties whose analysis raises technical
challenges, or where the effects cascade; it is required for rules that exceed the $1 billion annual threshold”
(OMB 2003, p. 41).
Probabilistic analysis often involves the use of simulation models to quantify the probability distributions of the
effects. It provides decision-makers with information about the variance, or spread, of the statistical distribution
of the impacts. This information may be particularly useful when the expected value of the net benefits is close
to zero or similar across multiple policy alternatives. In such cases, decision-makers may feel more confident
about the results if they have a smaller variance, because the realized results are more likely to be near the
expected value.
To conduct a formal probabilistic analysis, analysts must determine the joint distribution of the uncertain
parameters; i.e., the distribution of each parameter together with any dependencies among them. For some
parameters, such as the average body mass index (BMI) of the population when BMI has been measured for a
large representative sample, the distribution can be well estimated from the sample distribution. In other cases,
the probability distribution may be estimated from other data (e.g., by regression analysis), or it may be
necessary to assume a distribution (e.g., uniform or triangular between upper and lower bounds) and to test
whether the results are very sensitive to the assumed distribution.
Even when data are limited, distributions can be developed through formal, structured expert elicitation. Such
elicitation is designed to avoid well-known heuristics and biases that can lead to poor judgment, and may be
worthwhile if (1) assumptions about the distribution are likely to significantly affect the analytic results;
(2) additional primary data collection is not feasible or cost-effective; and (3) sufficient time and resources are
available.
Conducting structured expert elicitation requires substantial effort. Researchers first develop a well-defined
question to be addressed, as well as an extensive elicitation protocol designed to ensure that the experts each
interpret the questions similarly and explain the bases for their responses. Experts are identified through a
formal process intended to provide a range of perspectives. The elicitation often includes supplying the experts
with background materials and holding a pre-elicitation workshop to share and critique information. The
elicitation is then conducted with each expert individually, frequently through a lengthy interview following a
pre-determined protocol. More information on this process can be found in the expert elicitation literature (e.g.,
Morgan and Henrion 1990, Cooke 1991, and O’Hagan et al. 2006).
Once the joint distribution of the key parameters is estimated, Monte Carlo simulation techniques are applied to
derive a probability distribution of the outcome measure, which may be total costs, total benefits, net benefits,
or another impact of concern. Monte Carlo analysis involves taking a random draw from the joint distribution of
the uncertain parameters (or from the distributions for each parameter if they are independent) to produce a
value for each parameter; these values are then used to calculate the outcome measure. This process is

46
repeated many times to produce a distribution of the outcome measure, the average of which provides an
estimate of its expected value.
An advantage of Monte Carlo analysis is that it provides information on the full distribution of effects, from
which one can determine how likely it is that the effect exceeds any particular threshold (e.g., zero). A limitation
is that the results can be sensitive to the probability distributions that are used for the input parameters, and
these are often not known with much accuracy.
In sum, HHS analysts should quantify the impacts of the regulatory alternatives to the greatest extent practical.99
The analysis should be accompanied by clear discussion of the evidence of causality as well as the quality of the
studies and the statistical rigor of the methods used. However, even if the available data are of low quality or
inconsistent, the impact should be quantified and accompanied by an appropriate assessment of uncertainty
that clearly communicates the limitations of the analysis.100 When time and resource constraints restrict the
extent to which less significant impacts can be quantified, the evidence used to support the analytic decision
should be reported. Potentially significant effects should be left unquantified only when there is no feasible
approach for quantifying them.
Regardless of which approach is used to assess uncertainty, analysts should take care to avoid the appearance of
false precision. Calculations should be performed without any intermediate rounding, but the results should
generally be rounded for presentation in the RIA. While a variety of conventions are used in different disciplines
to determine the number of significant figures to present, generally the results should be rounded to reflect the
number of significant digits in the input data. For example, total costs should not be reported to the penny if the
unit costs used as an input are reported in tens or hundreds of dollars.

6.2 CHARACTERIZING NONQUANTIFIED EFFECTS

Another challenge is addressing outcomes that cannot be quantified but may have important implications for
decision-making. For example, available data may suggest that a regulated hazard affects the risk of both
mortality and morbidity, but may not be adequate to estimate the change in some types of morbidity risks
associated with each regulatory option. Without quantification, it is difficult to appropriately balance the risk
reductions associated with each option against its costs, or to determine the relative importance of these
different types of benefits.101
Quantification with appropriate treatment of uncertainty is desired (as discussed above) because it provides a
clearer indication of the likely direction and magnitude of the impacts. If quantification is not possible, analysts
must determine how to best provide related information. Ignoring potentially important nonquantified effects
may lead to poor decisions, but there is also a danger of overemphasizing them. In the absence of information,
decision-makers and others may weight nonquantified effects in a manner consistent with their own
(unarticulated and perhaps unconscious) beliefs, without sufficiently probing the rationale or the weighting.
Clear presentation of the available evidence is needed to counterbalance this tendency.102
Thus analysts should first quantify regulatory impacts to the greatest degree possible, using tools such as
sensitivity and probabilistic analysis to evaluate the effects of uncertainty as discussed previously. They then
should determine how to best describe those effects that remain unquantifiable, to provide insights into their
significance in comparison to each other and to the quantified impacts, as discussed below.

99
OMB Circular A-4 states: “[t]o the extent feasible, you should quantify all potential incremental benefits and costs” (OMB 2003, p. 45).
100
Determining how to best apply the available research requires careful review of the evidence and substantial professional judgment. A number of
approaches, such as criteria-driven systematic review, meta-analysis, and structured expert elicitation, can be used to develop estimates in cases where the
research varies in quality and provides inconsistent results. The benefit transfer framework, discussed in Chapter 3, also can be applied to other types of
quantities to develop estimates from data on somewhat dissimilar effects.
101
We use the term “quantification” to refer to the consequences of the regulation (generally measured in physical units, such as cases averted), and
monetization to refer to the dollar value of those consequences.
102
OMB Circular A-4 (2003) indicates that nonquantified effects should be included in the summary table discussed in Chapter 8.

47
6.2.1 BASIC CO NCEPTS
Analysts may be unable to estimate some potentially important regulatory impacts due to gaps in the available
data, the nature of the impacts themselves, or the need to focus on assessing more significant effects due to
time and resource constraints. For example, analysts may not have the data needed to estimate the effect of the
regulation on disease incidence, even though the available research suggests that the disease is associated with
the regulated hazard. In the case of costs, analysts may have evidence that the regulation will lead to significant
innovation, but may not be able to predict or describe the likely innovations adequately to estimate the impacts
in monetary terms.
Another example is information provision. Some regulations increase the type or quality of information available
and its dissemination, but research may be lacking on how recipients are likely to respond. Thus while an
intermediate measure may be available, such as the number of patients who receive information on potentially
beneficial lifestyle changes, it may not be possible to translate this measure into a quantity that can be
monetized to estimate benefits. The latter requires an estimate of the change in behavior that results and of
how the behavioral change affects individual welfare; e.g., of the degree to which the risk of illness or death is
reduced. While these types of deficiencies ideally would be remedied through additional primary research, such
research may require more time and resources than immediately available. HHS agencies should, however, try
to anticipate future analytic needs and invest in research that will be useful across several regulatory analyses.
In other cases, the lack of quantification may result because the effects are less tangible and more subject to
normative judgment. They may involve important human values, such as dignity, equity, and privacy. While it
may be difficult to quantify the change in these values attributable to a particular regulation, it may be possible
to count the number of people affected or report other intermediate measures.
Any intermediate measures, such as these counts, should be presented in the analysis as indicators of potential
costs or benefits. For example, if analysts have information on the number of organizations subject to a
regulatory provision, but lack the information needed to estimate related costs, or they have information on the
number of individuals affected, but lack the data needed to estimate a particular benefit, these counts should be
reported. Such intermediate measures should also be reported when the resulting benefits and costs are fully
quantified, to promote better understanding of the analytic results.

6.2.2 GENERAL APPROA CH

Options for incorporating nonquantified effects into the
regulatory analysis depend on the available data and include both HOW SHOULD NONQUANTIFIABLE EFFECTS
quantitative and qualitative approaches. Approaches that involve BE ADDRESSED?
some calculation (and may be particularly useful when comparing If it is not possible to quantify an impact,
benefits and costs) include breakeven, cost-effectiveness, and analysts should consider using breakeven,
bounding analysis, but care must be taken to avoid cost-effectiveness, or bounding analysis, as
misinterpretation of the results. More qualitative approaches well as tables and text, to illustrate the
include the use of tables and graphics as well as text discussions. potential implications.
Breakeven analysis: Breakeven analysis, sometimes referred to as
threshold analysis, asks the question “how large would the
nonquantified effect(s) have to be, to bridge the gap between quantified benefits and costs?” Figure 6.1
provides an example of this concept. Part (a) shows the case where only some of the benefits can be quantified;
part (b) illustrates the case where only some of the costs can be quantified.

48
 F I G U R E 6 . 1 . B R E A K E V E N A N A LY S I S

Generally, breakeven analysis can only be conducted for a single quantity. Thus breakeven analysis is useful
when analysts are particularly uncertain about one key parameter. For example, analysts may have information
on the value of the effect (e.g., the VSL in the case of mortality risk reductions) but not the physical effects (e.g.,
the number of statistical cases averted). In this case, the breakeven analysis would be used to estimate the
number of averted cases needed for benefits to exceed costs, given the VSL. Similarly, for costs, it may be
possible to estimate the number of firms affected by a particular provision, but not the cost per firm. Breakeven
analysis can be used to provide insight into how large the cost per firm would need to be for the costs to exceed
the benefits of that provision. It can also be used to identify the breakeven probability of occurrence that would
equalize costs and benefits.
Once the analysis is conducted, decision-makers and stakeholders can inspect the results to judge whether it is
likely that the nonquantified effects are large enough to fill the gap. Breakeven analysis is most useful when
some information is available on the potential magnitude of the impact, to provide a basis for judging whether
the nonquantified effects can plausibly exceed the breakeven amount. It also may be informative when data are
available but not public. For example, confidential information on the likelihood and consequences of terrorist
attacks may be available to decision-makers but not to regulatory analysts or the general public. This
information could provide context for decision-makers’ review of the breakeven results for a regulation that
addresses homeland security.
Cost-effectiveness analysis: Cost-effectiveness analysis is another approach that can provide insights when an
impact can be quantified but cannot be assigned a monetary value (see Institute of Medicine 2006, Drummond
et al. 2015). Under this approach, a monetary estimate of the costs (net of any monetized benefits) is divided by
an effects measure to determine the cost per unit of effect. The effect could be the number of deaths averted,
QALYs gained (see Chapter 3 and Appendix C), individuals treated, or another measure. Care must be taken,
however, in interpreting the results. Cost-effectiveness ratios do not indicate whether an intervention is worth
undertaking (i.e., whether the value of the benefits exceeds the costs), nor which option is likely to yield the
largest net benefits.
Bounding or “what-if” analysis: Bounding analysis considers the extent to which benefits are likely to exceed
costs based on lower- or upper-bound estimates of the magnitude of the nonquantified effects. For example, if
the available data are sufficient to estimate that the mortality risk reductions associated with the regulation are
unlikely to be greater than 1,000 statistical cases or fewer than 10 statistical cases, then the results could be
presented using both estimates. “What if” analysis is similar, and involves investigating the impact of various
hypothetical, but plausible, scenarios on the results. For example, the analyst could compare benefits and costs
49
for mortality risk reductions ranging from 10 to 1,000 statistical cases, if he or she believes that outcomes within
this range are possible, and report the extent to which benefits exceed costs under each scenario.
The dividing line between these approaches and standard sensitivity analysis (discussed above) is somewhat
vague. In concept, bounding or “what-if” analysis in this case would involve very wide ranges based on relatively
little data or supporting evidence, and would be presented separately from the primary estimates of benefits
and costs due to the high degree of speculation involved.
Tables and graphics: Tables and graphics are often useful for highlighting nonquantified effects, to ensure that
they are not overlooked by decision-makers and others. One option is to simply list the effects in a table;
however, the list is likely to be more useful if the effects can be categorized in a way that indicates the
implications for decision-making. This categorization could include whether the effects are likely to be large or
small, and to lead to over- or underestimates. Separate categories or exhibits could be used to report the
strength of the evidence that links the effect to the regulation, the likelihood of its occurrence (e.g., high or low),
or the extent to which it is reversible, as well as other attributes that will be salient for decision-making.
Table 6.2 below provides an example that uses symbols to highlight the potential magnitude of the impacts.103
Alternatively or in addition, analysts could insert text into the table to provide more information than can be
conveyed by a symbol. Such tables can also be used to separately indicate the effects on benefits and costs,
rather than solely focusing on net benefits as in the example.
TA B L E 6 . 2 . E X A M P L E O F S U M M A RY O F N O N Q U A N T I F I E D E F F E C T S

EFFECT OF NONQUANTIFIED IMPACTS ON NET BENEFITS POTENTIAL MAGNITUDE*

Analysis may overstate net benefits

 impact “a”
 impact “b”
 etc.
Analysis may understate net benefits
 impact “c”
 impact “d”
 etc.
Analysis may under- or overstate net benefits
 impact “e”
 impact “f”
 etc.
*Dashed vertical line indicates quantified net benefits.

Text discussion: All of the approaches described above must be accompanied by text that clearly defines the
nonquantified effects, explores the causal evidence that links them to the regulatory action, summarizes
available information on their direction and magnitude, and discusses the conduct and interpretation of related
analysis, including both the results and related uncertainties.

103
Microsoft Excel and similar programs allow the user to represent quantities graphically; for example, to automatically size an arrow that represents the
quantity “10” so that it is twice the size of an arrow that represents the quantity “5.” While such features may be useful when analysts have some
information on relative magnitudes, care should be taken to not mislead readers about the extent to which the size of the symbols represents evidence on
the expected size of the effect.

50
In sum, the treatment of nonquantified impacts should be tailored to the characteristics of the effect (such as
whether it involves intangibles or normative values), the extent to which relevant data are available, and the
importance of the effect for decision-making. These impacts should be clearly defined and distinguished from
the quantified impacts, to avoid the potential for double-counting.
At minimum, analysts should list significant nonquantified effects in a table and discuss them qualitatively. To
the extent possible, the effects should be categorized or ranked in terms of their importance and implications
for choosing among the regulatory alternatives (including the option of no action). Where some data exist, but
are not sufficient to reasonably quantify the effect, analysts should consider whether breakeven, cost-
effectiveness, or bounding analysis will provide useful insights. Intermediate measures, such as the number of
individuals affected, should be reported where available. Where impacts can be monetized but not quantified,
the monetary value per unit of impact (e.g., the value per averted statistical case in the case of health impacts)
should be reported.

51
Chapter 7
Conduct Distributional and Other
Supplementary Analyses
The previous chapters focus largely on the benefit-cost analysis that is the core of the RIA. However, agencies
must also comply with a number of other analytic requirements. These include considering the distribution of
benefits and costs across demographic or other population subgroups as well as complying with several other
executive orders and statutes. In addition, for those regulations with impacts outside of the U.S., analysis of
international impacts is required. These analyses should be reported in clearly labeled, separate sections of the
regulatory analysis (see Chapter 8), which discuss the available evidence and related uncertainties as well as the
implications for decision-making.

7.1 ASSESS DISTRIBUTION ACROSS DEMOGRAPHIC GROUPS

In addition to estimating the national net benefits of the policy options, HHS and other regulatory agencies are
required to separately address how the benefits and costs of their economically significant regulations are
distributed. The benefit-cost analysis discussed previously focuses on the net impact of the regulation on social
welfare, while the distributional analysis focuses on the incidence of the benefits and costs.
In this section, we discuss the distribution of impacts across individuals with differing demographic or other
characteristics. Such analysis is encouraged under Executive Orders 12866 and 13563 (Clinton 1993, Obama
2011), as well as by OMB Circular A-4 (2003), and includes analyses required by Executive Order 13045,
“Protection of Children from Environmental Health Risks and Safety Risks” (Clinton 1997), and Executive Order
12898, “Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations”
(Clinton 1994), where applicable.104
This analysis is intended to provide descriptive information for consideration by decision-makers and
stakeholders; it should not assign values to reflect distributional preferences or make normative judgments
related to the fairness or equity of the impacts. In many cases this analysis will be primarily qualitative or rely
largely on simple screening; in those cases where distributional concerns are more significant, it will be more
extensive and detailed.

7.1.1 BASIC CO NCEPTS

The goal of distributional analysis is to provide information on how benefits and costs affect different groups, so
as to make trade-offs between economic efficiency and distributional concerns more explicit. Decision-makers
may choose the economically-efficient regulatory option that maximizes net benefits, or may choose a less
efficient option to ameliorate distributional impacts or achieve other policy goals.
Generally, the distribution of both benefits and costs should be considered, so that decision-makers and others
can consider the extent to which the impacts are counterbalancing for each group as well as the overall
distribution of net benefits across groups. In addition to understanding the incremental effects of the regulation,
analysts may wish to provide information on the distribution under the “without new regulation” baseline as
well as on the distribution that results under each policy alternative.
The starting point for distributional analysis is the national assessment of social benefits and costs, discussed in
Chapters 3 and 4. However, as noted in Chapter 4, transfer payments are generally not included in the benefit-
cost analysis, but must be considered in the distributional analysis.

104
See the National Archives website for a complete set of executive orders ([Link]

52
A key step in the analysis involves identifying which population groups should be considered.105 In some cases,
groups of concern may be defined by statute. In addition, Executive Order 12898, “Federal Actions to Address
Environmental Justice in Minority Populations and Low-Income Populations” (Clinton 1994), requires agencies to
identify and address “disproportionately high and adverse human health or environmental effects” on these
groups. Executive Order 13045, “Protection of Children from Environmental Health Risks and Safety Risks”
(Clinton 1997), requires agencies to identify and address risks that may disproportionately affect children. Other
groups of concern may emerge in the course of the analysis. For example, analysts may find that the effects of
the regulations are likely to be concentrated in certain geographic areas or among groups with particular
characteristics, such as the homeless, the HIV-infected, or those with specific dietary habits.
It is often tempting to focus solely on adverse effects on disadvantaged groups. However, such focus is
problematic because it leads analysts to ignore potential beneficial effects that may be of equal or greater
importance. Any distributional effect involves both “from” and “to” sides of the equation; who gains may be as
important as who loses. The benefits and costs of the regulation may be counterbalancing, or may differentially
affect the advantaged and the disadvantaged.
When describing these effects, one option is to provide a table or graph that reports the percentage and value
of the costs, benefits, and net benefits that accrue to individuals or households at different points in the
distribution; e.g., to income quintiles. Other measures for describing inequality are available; their advantages
and disadvantages are discussed in detail in several sources.106

7.1.2 GENERAL APPROA CH

Assessing the distribution of regulatory benefits and costs, as well as net benefits, can be challenging. As noted
earlier, the conduct of such analysis is likely to vary significantly depending on the nature of the regulation, the
characteristics of its benefits and costs, the population groups of
interest, and the data and other analytic resources available. WHAT ARE THE REQUIREMENTS FOR
Screening analysis (see Chapter 2) can be useful in determining DISTRIBUTIONAL ANALYSIS?
how to best focus this effort. Below, we discuss some of the
At minimum, analysts should include a short
challenges related to assessing the distribution of regulatory costs
description of the likely distribution of
and health-related benefits, which affect analysts’ ability to
benefits and costs across individuals or
address each independently as well as their net effect.
households in different population groups,
Distribution of regulatory costs: In the case of regulatory costs including low income and minority groups
(and off-setting savings), we are typically interested in the and children as discussed in Executive Orders
monetary expenditures needed to comply with the regulatory 12898 and 13045. Requirements for other
requirements (including transfers), measured in dollar terms, and types of distributional analysis are discussed
the ultimate effect on the disposable income of the groups of in the next section.
concern. Where regulatory costs are borne directly by individuals
and households, the main challenge is determining how the costs
are distributed across those who belong to different groups, which may be identified, for example, by income
quintile, minority status, or degree of health impairment.107 Where the costs are borne initially by firms,
assessing the effects on individuals and households in different groups requires additional steps.108 We first
need to know how regulatory costs imposed on these entities translate into changes in unit prices paid by

105
OMB Circular A-4 (2003) defines distributional effects broadly as including, for example, how regulatory impacts are divided across “income groups, race,
sex, industrial sector, geography” as well as over time.
106
For a general overview of options for addressing distributional concerns in policy analysis, see Weimer and Vining (2011), Chapter 7. For further discussion,
see Boardman et al. (2011).
107
Consumer behavior will also affect the distribution of these costs. For example, if the price of a food is increased, some may substitute an alternative food.
This substitution may affect both the costs and the benefits incurred, and such behavioral responses may vary across population groups.
108
If the organizations are not-for-profit, similar principles apply although the nature of the impacts may differ. If the costs are initially incurred by a
government unit, then the analysis would address how that unit is funded; i.e., the distribution of taxes, users fees, or other revenue sources.

53
consumers (including both income and substitution effects), in wages paid to employees, and in returns to
capital that accrue to owners, as illustrated in Figure 7.1.

F I G U R E 7 . 1 . D I S T R I B U T I O N O F I N D U S T RY C O S T S A C R O S S I N D I V I D U A L S A N D H O U S E H O L D S

As in the benefit-cost analysis, the distributional analysis must clearly differentiate the impacts of the new policy
from the impacts of other factors that should be reflected in the “no new regulation” baseline projections. At
times, retrospective analysis may be available that addresses similar regulations and uses statistical tools to
distinguish the effects of regulatory costs.109 Interviews with members of the affected industry may also be
useful. Otherwise, the extent to which each of the pathways in Figure 7.1 can be assessed will depend largely on
the data available from the benefit-cost analysis. If only direct compliance costs are estimated, then it may be
difficult to estimate how the costs are allocated across consumers and producers. If partial equilibrium modeling
is included (which estimates changes in consumer and producer surplus), more sophisticated distributional
analysis is possible. In a few cases, where regulations are expected to have significant impacts throughout the
economy, results for model households from general equilibrium modeling may also be available. In all cases,
the analysis of total social costs will exclude transfers, which will need to be estimated to assess the distribution
of the impacts.
The allocation of costs across producers and consumers will also depend on the timeframe considered. Some
costs that are fixed in the short run will be variable in the long run. For example, in the near term firms may not
be able to make major changes in their physical plant (and some may close due to the costs of complying with
the regulation), but such changes become more possible in the future, affecting how the costs are distributed.
Distribution of health benefits: In the case of benefits, some regulations may primarily provide savings in
monetary costs, in which case the distributional analysis would proceed along the same lines as described above
although the effects are likely to be in the opposite direction – savings potentially decrease prices, increase
wages, and increase returns to capital. When the benefits involve reduced mortality and morbidity risks, there
are several options for measuring the effects on each group. We can count the number of statistical cases
averted (by multiplying the expected individual risk reduction by the number of people affected); we can use
integrated measures (such as quality-adjusted life years, QALYs) to estimate the net effect on health-related

109
For employment impacts, see Morgenstern (2013) for a comprehensive review.

54
quality of life and longevity; and we can use monetary measures that indicate the amount those affected would
be willing to pay for the risk reductions (see Chapter 3).
In general, the distribution of health effect incidence is easier to calculate than the distribution of costs. The
benefit analysis is likely to provide estimates of the number of people affected; the challenge is then to identify
how the effects are allocated across the groups of concern.110 In some cases, the characteristics of the regulation
may aid in estimating this distribution. For example, if a food safety regulation affects the risks associated with
drinking juice, and the distribution of juice drinking across groups (categorized by income, age, or other
demographic attributes) is known, the analysis may be relatively straightforward. The risk assessment that
supports the regulation will often provide related information. It typically summarizes or references available
data on populations that may be particularly sensitive or vulnerable to the effects of the regulated hazard,
including those who may be disproportionately affected due to health conditions, age, or socioeconomic status.
In addition, HHS maintains several population databases that provide information on the characteristics of those
who experience various types of health effects. Examples include the National Health Interview Survey and the
Medical Expenditure Panel Survey.
In sum, the discussion above suggests that distributional analysis may be quite complex, and requires thinking
carefully about what types of information will be most useful to decision-makers given the characteristics of the
regulation and of those it is likely to affect. In some cases, the analysis may be primarily qualitative; in others
more detailed quantitative assessment will be warranted. Analysts should follow a phased approach to ensure
that the assessment is well-focused and useful for decision-making, using screening analysis as discussed in
Chapter 2. Both gains and losses among advantaged and disadvantaged groups should be considered, to ensure
that any counterbalancing or exacerbating impacts are taken into account.

7.2 CONDUCT SUPPLEMENTARY ANALYSES

Several other types of analysis are required by various statutes and executive orders. In general, all of these
requirements should be addressed; however, the extent to which detailed analysis is required will depend on
the characteristics of the specific rule. Table 7.1 summarizes these requirements and directs the analyst to
additional guidance documents. The basic requirements are discussed in more detail below.

110
As noted earlier, to the extent that people may alter their behavior in response to the regulation (e.g., taking less precaution in handling food when
packaging is improved), any difference in this response can affect the distribution of benefits.

55
 TA B L E 7 . 1 . R E Q U I R E M E N T S F O R S U P P L E M E N TA RY A N A LY S E S

REQUIREMENT APPLICABILITY GUIDANCE DOCUMENTS

Regulatory Flexibility Act: All regulations subject to notice and comment  A Guide for Government Agencies: How to
Requires agencies to consider under section 553(b) of the Administrative Comply with the Regulatory Flexibility Act
the impact of regulatory Procedures Act. (SBA 2012)
actions on small entities, Note: a full regulatory flexibility analysis is not  Guidance on Proper Consideration of Small
analyze effective alternatives required if the agency can certify that the Entities in Rulemakings of the U.S. Department
that minimize small entity proposed rule will not “have a significant of Health and Human Services (HHS 2003)
impacts, and make their economic impact on a substantial number of
analyses available for public small entities” (5 U.S.C. §605(b)). HHS provides
comment. guidance defining a “substantial number” and
“significant effect” (see HHS 2003).
Unfunded Mandates Reform All “significant” rulemakings – defined as  “Guidance for Implementing Title II of S.1”
Act: Requires agencies to those likely to result in the expenditure by (OMB 1995)
assess the effects of State, local, or tribal governments, in the  Annual memorandum from HHS updating
regulatory actions on State, aggregate, or by the private sector, of $100 “significant rulemaking” threshold value (e.g.,
local, and tribal governments, million or more in any one year in 1995 HHS 2014)
and the private sector. dollars, adjusted for inflation.
Executive Order 13132 All policies that have “substantial direct None.
(“Federalism”): Requires effects on the States, on the relationship
agencies to develop a process between the national government and the
to ensure meaningful and States, or on the distribution of power and
timely input by State and local responsibilities among the various levels of
officials. government.”
Paperwork Reduction Act: All policies that require generation,  Paperwork Reduction Act Primer
Requires agencies to estimate maintenance, or provision of information to (Sunstein 2010b)
the information collection or for a Federal agency. Agencies must obtain  OMB’s website Federal Collection of Information
(reporting, recordkeeping, approval from OMB prior to requesting the
and third-party disclosure) same information from 10 or more  HHS’s website Frequently Asked Questions
burden associated with their individuals. about PRA/Information Collection
actions.  Agency’s designated PRA team

7.2.1 REGULATORY FLEXIBILI TY ACT

The Regulatory Flexibility Act of 1980 (RFA), as amended by the Small Business Regulatory Enforcement Fairness
Act (SBREFA) (5 U.S.C. 601, et seq.), “requires agencies to consider the impact of their regulatory proposals on
small entities, analyze effective alternatives that minimize small entity impacts, and make their analyses
available for public comment” (SBA 2012). Small entities include small businesses, not-for-profit organizations,
and governmental jurisdictions, definitions of which can be found within Section 601 of the RFA. In addition, the
U.S. Small Business Administration (SBA) has developed size standards to define small businesses, which can be
found in 13 CFR 121.201. The RFA requirements have been extended to small rural hospitals through Section
1102(b) of the Social Security Act (42 U.S.C. §1302). Definitions of “small,” “rural,” and “hospital” are provided in
the Medicare regulations at 42 CFR 412.
If a proposed rule is not expected to have a significant impact on a substantial number of small entities, the
agency may certify that this is the case, and must provide a statement providing the factual basis for this
determination. If the agency cannot provide this certification, or is uncertain about the rule’s impact, it should
prepare an Initial Regulatory Flexibility Analysis (IRFA) for publication with the proposed rule. Section 603 of the
RFA lists the information that must be included in the IRFA.

56
For the final rule, if the agency cannot provide this certification or remains uncertain after reviewing public
comment on the proposed rule, a Final Regulatory Flexibility Analysis (FRFA) should be prepared and published.
The requirements for the FRFA are similar to those for the IRFA and are outlined in Section 604 of the RFA.
When it prepares a FRFA, the agency must also publish one or more small entity compliance guides to inform
small entities of their obligations and responsibilities under the rule.
Detailed guidance on compliance with the RFA and preparation of the regulatory flexibility analysis can be found
in the SBA’s A Guide for Government Agencies: How to Comply with the Regulatory Flexibility Act (SBA 2012).111
This document walks agencies through the process of preparing screening analyses and initial and final
regulatory flexibility analyses. In addition, HHS’s Guidance on Proper Consideration of Small Entities in
Rulemakings of the U.S. Department of Health and Human Services (HHS 2003) supplements the SBA guidance,
providing examples of issues that commonly arise in applying the RFA and SBREFA to HHS rulemakings.

7.2.2 UNFUNDED MANDATES REFORM ACT

The Unfunded Mandates Reform Act (UMRA) (2 U.S.C. §1501 et seq.) seeks to curb the practice of imposing
unfunded Federal mandates on State and local governments. UMRA Section 1531 requires Federal agencies to
assess the effects of their regulatory actions on State, local, and tribal governments, and the private sector.
Section 1532 requires them to prepare a written statement that assesses the costs, benefits, and other effects of
proposed or final rules for significant regulatory actions (2 U.S.C. §1532(a)). UMRA defines significant regulatory
actions as those that include a Federal mandate likely to result in the expenditure by State, local, or tribal
governments, in the aggregate, or by the private sector, of $100 million or more in any one year (in 1995 dollars)
(2 U.S.C. §1532(a)). This threshold is adjusted each year for inflation.
Most of UMRA’s requirements are fulfilled by the RIA that is prepared to comply with Executive Orders 12866
and 13563 (Clinton 1993, Obama 2011) and OMB Circular A-4 (2003), as discussed in the earlier chapters of this
guidance. Additional guidance on the preparation of written statements under UMRA can be found in OMB’s
1995 “Guidance for Implementing Title II of S.1.” In addition, HHS releases an annual memorandum updating the
threshold value (adjusted for inflation) for a significant regulatory action (see, for example, HHS 2015).112

7.2.3 FEDERALISM
Executive Order 13132, “Federalism” (Clinton 1999), emphasizes consultations with State governments and
enhanced sensitivity to their concerns in cases where regulatory or other policy actions impinge on their
constitutionally established role as sovereign entities. It requires Federal agencies to develop an accountable
process to ensure “meaningful and timely input by state and local officials in the development of regulatory
policies that have federalism implications.” Section 1(a) defines policies that have federalism implications to
include regulations that have “substantial direct effects on the States, on the relationship between the national
government and the States, or on the distribution of power and responsibilities among the various levels of
government.”
Under Executive Order 13132, Federal agencies may not issue a regulation with Federalism implications that
imposes substantial direct compliance costs and that is not required by statute unless the Federal government
provides the funds necessary to pay the direct compliance costs incurred by State and local governments or the
agency consults with State and local governments in the process of developing the proposed regulation. The
agency also may not issue a regulation with Federalism implications that preempts a State law without
consulting with State and local officials.

111
This and other material related to implementation of the RFA is available on the Regulatory Flexibility Act ([Link]
navigation-structure/regulatory-flexibility-act) page of the SBA website.
112
The method and sources used to update this threshold value are described in HHS (2015).

57
7.2.4 PAPERWORK REDU CTION ACT
The Paperwork Reduction Act (PRA) (44 U.S.C. §3501 et seq.) requires Federal agencies to estimate the
information collection burden associated with all of their actions. The term “burden” means the time, effort, or
financial resources expended by persons to generate, maintain, or provide information to or for a Federal
agency. Agencies must obtain approval from OMB prior to requesting the same information from ten or more
individuals. Thus, if a proposed regulation will impose such a burden (e.g., a regulation may require regular
reporting of compliance data to HHS), the agency must prepare an information collection request (ICR) for
review and approval by OMB.
Paperwork burdens or costs are a subset of the total costs of a regulation and should be included in those costs
(see Chapter 4).113 The paperwork burden of a regulation includes the incremental cost of required record
keeping, reporting, and public disclosure. It includes only the incremental data collected as a result of the
regulation; data collections required by the rule that are already undertaken for other purposes are considered
part of the baseline and are not part of the collection burden under the PRA. For example, a rule requiring
facilities to maintain records on health and safety-related maintenance practices may not result in an
incremental collection burden if these records are already collected by the facility for other purposes, such as
payroll. As with estimates of other compliance costs (see Chapter 4), it is important to isolate the incremental
burden of the regulation when preparing the ICR.

7.3 ADDRESS INTERNATIONAL EFFECTS

The regulatory analysis should generally focus on benefits and costs that accrue to U.S. citizens and residents.
However, regulations that address trade barriers and other market failures may have an effect on both the
United States and its trading partners. In cases where regulations have impacts outside of the United States,
they should be addressed in a supplementary analysis. Following the guidance in OMB Circular A-4 (2003), these
international effects should be reported separately from those occurring within the U.S.114
International effects may include direct economic impacts (e.g., related to increases or decreases in
international trade) as well as any other potentially significant effects. For example, increasing safety
requirements for U.S.-based food manufacturing may provide health benefits to countries that import this food;
decreasing the transmission of disease in the U.S. is likely to decrease the risk of transmission to residents of
other countries.
In general, analysis of international effects should include impacts on imports and exports. Partial equilibrium
analysis using publicly available information on import supply and demand elasticities can be used to model how
a regulation might change the flow of imports and exports. More complicated general equilibrium analysis may
be required if an entire sector of the economy is affected. For additional information on these types of
modeling, see Chapter 4.
The analysis of international effects may also include impacts on foreign entities whose U.S. operations are
affected. It is often difficult to identify U.S. subsidiaries of foreign entities and report impacts to their operations
separately from those to U.S.-based businesses. Therefore, impacts on U.S. subsidiaries are often included in the
main analysis. If this is the case, and the analyst thinks that impacts on U.S. subsidiaries of foreign entities may
be substantial, the analysis should include a qualitative discussion of the effect.

113
There are important differences in the requirements of the PRA and the best practices for preparing RIAs as discussed in the prior chapters. For a detailed
discussion of the PRA requirements, see the sources referenced in Table 7.1.
114
Executive Order 13069 (Obama 2012) includes requirements for identifying regulations that may have significant international impacts.

58
For more information on how to address international effects, see OMB’s 2008 Report to Congress on the
Benefits and Costs of Federal Regulations and Unfunded Mandates on State, Local, and Tribal Entities (OMB
2008) and the Review of the Application of EU and US Regulatory Impact Assessment Guidelines on the Analysis
of Impacts on International Trade and Investment (OMB and the Secretariat General of the European
Commission 2008).115

115
OMB’s reports to Congress are available on the OIRA Reports to Congress ([Link] page of its
website.

59
Chapter 8
Communicate the Approach and Results
Regulatory analyses must be clearly and comprehensively documented in an RIA, which may be published in full
in the preamble to the Federal Register notice for the proposed or final rule, or as a separate report, in which
case it must be summarized in the preamble. The RIA must describe the rationale for the regulation, the options
considered, the analytic approach, and the results, as well as the implications of uncertainties. For regulations
with particularly large or complex impacts, it may be necessary to provide additional information in technical
reports that supplement the main analysis.
Without clear communication, the RIA will not meet its intended goal of informing related decisions. This
communication should address two audiences. First, it should be written so that members of the lay public can
understand the analysis and conclusions. Second, it should provide enough detail so that competent analysts
could ideally reconstruct the analysis, or at minimum explore the implications of changing key assumptions. This
chapter briefly describes related practices.

8.1 DESCRIBE THE ANALYSIS AND RESULTS

The audience for the RIA is diverse and includes many who lack the technical expertise and knowledge of those
who conducted the analysis. Given that the purpose of the analysis is to inform decision-makers and other
stakeholders, it is critical that it be described in terms that can be easily understood by a lay audience. At the
same time, the documentation must be sufficient to support future work, including replication, testing the
effects of alternative assumptions, applying the same or similar approaches in a future analysis, or
reconstructing the analysis as part of a retrospective assessment.
The main text should provide a succinct and clear summary of the analysis. Technical details should be provided
in appendices or supporting documents. The main text may, for example, include the following major sections,
reflecting the requirements in OMB Circular A-4 (2003) as well as the requirements provided in this guidance
document. Those sections that provide analytic results should also include a subsection that discusses the
implications of uncertainties, as described in Chapter 6.
1) Executive Summary (see additional discussion below)
2) Statement of the need for the regulation
3) Characterization of the without-regulation baseline
4) Description of the regulatory alternatives (including the preferred alternative)116
5) Benefits of the regulatory alternatives
6) Costs of the regulatory alternatives
7) Comparison of benefits and costs
8) Supplementary analyses
a) Distribution of benefits and costs
b) Regulatory Flexibility Act analysis
c) Unfunded Mandates Reform Act analysis
d) Other analyses
e) International effects

116
These alternatives may include both regulatory and non-regulatory approaches, as described in Chapter 2.

60
In particular, the executive summary must use plain English and be designed to promote public understanding.
OMB (2012) suggests that executive summaries include a statement of need for the regulation; a summary of
the major provisions of the regulatory action; and, for economically significant rulemakings, a table summarizing
the benefits and costs. For additional guidance on the format for Executive Summaries see “Clarifying Regulatory
Requirements: Executive Summaries” (OMB 2012).

8.2 PROVIDE SUMMARY TABLES AND FIGURES

The RIA should include tables and figures that clearly convey the results of the analysis.
Key information to be summarized includes:
 Annual benefits and costs (undiscounted);
 Annualized and present value costs;
 Annualized and present value benefits;
 Net benefits (i.e., benefits minus costs) presented on an annualized basis and, as appropriate, in present
value terms.
These quantified results should be accompanied by information on important nonquantified impacts.
In addition to “central” or “best” estimates, information on uncertainty must also be presented. When reporting
annualized or present value impacts, analysts must indicate the time period over which impacts are
estimated.117 Results should be presented for discount rates of both three and seven percent.
Depending on the complexity of the analysis and the number of cost and benefit categories, the results may be
summarized in a single or multiple tables or figures. Each should reference the information sources and note key
assumptions. While such exhibits are essential to focus attention on key findings, analysts should keep in mind
that some readers will skip over the more detailed technical information in the text. Thus clear labeling is
needed to ensure that the contents of the tables and figures are not misinterpreted. Additionally, the associated
text should interpret each table or figure for the reader. It may improve communication to supplement the
results tables with charts and graphs that summarize and highlight key steps in the analysis as well as the major
conclusions and their implications.
For economically significant rules, agencies are also required to provide OMB with an accounting statement that
includes a standard table reporting benefit and cost estimates. Figure 8.1 provides a suggested format for this
accounting statement, adapted from OMB Circular A-4. The accounting statement summarizes the information
presented in the RIA and should include:
 Annualized incremental benefit and cost estimates, using real discount rates of three and seven percent,
within the following three categories: monetized; quantified, but not monetized; and qualitative, but not
quantified or monetized. The primary benefit and cost estimates should reflect the expected values. The
minimum and maximum estimates should, if possible, reflect the 5th and 95th percent confidence bounds.
 Annualized incremental transfer estimates, which occur when wealth or income is redistributed without any
direct change in aggregate social welfare.
 Information on the effects on State, local, and tribal governments, small businesses, wages, and economic
growth.

117
As discussed in Chapter 2, for meaningful comparison, benefits and costs should be measured over the same time period. When some impacts are assessed
over longer periods than others to provide important information for decision-making, the results for the additional period should be reported separately to
avoid misleading comparisons.

61
 F I G U R E 8 . 1 . T E M P L AT E F O R O M B A C C O U N T I N G S TAT E M E N T

OMB #: Agency/Program Office:

Rule Title:
RIN#: Date:
Economic Data: Costs and Benefits Statement
. . . . Units .
Primary Low High Year Discount Period
Category Notes
Estimate Estimate Estimate Dollars Rate Covered
Benefits
Annualized . . . . 7% . .
Monetized $
millions/year . . . . 3% . .

Annualized . . . . 7% . .
Quantified . . . . 3% . .
Qualitative . . . . . . .
Costs
Annualized . . . . 7% . .
Monetized $
millions/year . . . . 3% . .

Annualized . . . . 7% . .
Quantified . . . . 3% . .
Qualitative . . . . . . .
Transfers
Federal . . . . 7% . .
Annualized
Monetized $ . . . . 3% . .
millions/year
From/To From: . . To: . . .
Other Annualized . . . . 7% . .
Monetized $
millions/year . . . . 3% . .

From/To From: . . To: . . .

.
Effects
State, Local or Tribal Government:
.
Small Business:
.
Wages:
.
Growth:

62
In addition to the present value and annualized results, OMB Circular A-4 suggests that the analyst include
separate schedules of undiscounted monetized benefits and costs showing the type and timing of these effects,
as discussed in Chapter 5. These undiscounted results should be presented in constant dollars for each year of
the analytic time horizon. Again, this schedule could be presented in a table or as a bar chart or other graphic.
In sum, presenting the analysis so that it can be easily understood by decision-makers and stakeholders may
require significant effort to clearly and concisely describe the options assessed, the analytic approach, and the
results. Without such effort, the analysis may not play its intended role in the decision-making process, and may
be misconstrued in ways that lead to significant and unnecessary controversy. Avoiding technical jargon, and
using tables and graphics to illustrate key points, will aid in ensuring that the analysis is useful for decision-
making.

63
Chapter 9
Conduct Retrospective Analysis
Executive Order 13563 directs each Federal agency to establish a plan for ongoing retrospective review of
existing significant regulations to identify those that can be eliminated as obsolete, unnecessary, burdensome,
or counterproductive, or that can be modified to be more effective, efficient, flexible, and streamlined (Obama
2011, HHS 2011).118 The initial HHS plan was finalized in August 2011 and has been subsequently updated.119
The plan describes HHS’s approach for identifying regulations for review as part of an ongoing process and lists
factors HHS routinely considers in this review (HHS 2011). The factors include many that can be evaluated
qualitatively; for example, identifying redundant or obsolete regulations or requirements. While important for a
broader program of retrospective review, such qualitative analysis is not the focus of this guidance. Rather, this
chapter describes HHS’s approach for quantitative retrospective analysis of the benefits and costs of selected
economically significant regulations.
Quantitative retrospective benefit-cost analysis may serve several purposes, ranging from assessing the
effectiveness of a single regulation to evaluating the overall use of benefit-cost analysis in the regulatory
development process. The next section discusses the conceptual framework in greater detail. We follow with an
overview of the approach to retrospective benefit-cost analysis, including a generalized discussion of analytic
steps.

9.1 BASIC CONCEPTS

The general purpose of the prospective, or ex ante, analysis discussed in the previous chapters of this guidance
is to determine whether the benefits of the regulation are likely to exceed costs (i.e., whether benefits minus
costs, or net benefits, are positive) and to identify the regulatory alternative likely to generate the largest net
benefits. Figure 9.1 identifies several ways in which subsequent retrospective, or ex post, analysis of benefits
and costs may be useful.120
F I G U R E 9 . 1 . U S E S O F R E T R O S P E C T I V E B E N E F I T C O S T A N A LY S I S

1. Evaluate whether existing regulations continue to be justified in economic terms (i.e., produce positive net
benefits).
2. Support identification of changes to existing regulations that will decrease their costs or increase their
benefits.
3. Provide insight into the accuracy of ex ante estimates of regulatory benefits and costs, particularly whether
they tend to be over- or underestimated.
4. Identify ways to improve the accuracy of future cost-benefit analyses.

A primary goal is to assess whether the regulation has achieved the desired outcome. For example, if its purpose
was to reduce new cases of heart disease, analysts would seek empirical evidence of this impact. While
potentially difficult to obtain, this information is a necessary to determine whether net benefits are positive.
Additionally, retrospective benefit-cost analysis “can help identify specific regulations that are ripe for
regulatory reform, since their benefit-cost balance may be more or less favorable than originally expected”
(OMB 2005). Importantly, OMB notes that “a validation study designed to determine the accuracy of ex ante

118
Aldy (2014) discusses the historical development of the retrospective review process within the Federal government and potential improvements.
119
See the HHS website for the 2011 plan ([Link] as well as updates and an opportunity for public input regarding
which regulations to review.
120
This discussion is based largely on OMB (2005); more information is provided in subsequent reports such as OMB (2011c).

64
estimates does not by itself provide full guidance on the desirability of reforming the existing regulation” (OMB
2005, p. 41). For example, regulated entities may have incurred costs that will not be recovered if the regulation
is retracted.121
Retrospective analysis may also inform the modification of an existing regulation with the goal of increasing its
net benefits, regardless of whether net benefits are positive or negative as currently implemented. New
information about key assumptions or inputs may suggest opportunities for optimizing the regulation.
After an agency has completed retrospective review of multiple regulations, it can identify whether it has a
tendency to systematically over- or underestimate costs or benefits, and the extent to which over- or
underestimation is attributable to various factors.122 This information might highlight the need for additional
uncertainty analysis, as well as ways in which future analyses can be improved.123 It might also provide insight
into how much weight should be granted to the cost-benefit analysis in the decision-making process as agencies
promulgate new regulations (OMB 2005).
Finally, such information may identify ways to improve the accuracy of future ex ante analyses. For example, it
may demonstrate that agencies routinely underestimate the ability of regulated entities to reduce costs as they
gain experience with a particular regulation.124 In certain cases, a regulation may motivate affected entities to go
beyond the required compliance standards, resulting in additional health or other improvements not included in
ex ante benefits estimates.125 A better understanding of how affected entities respond to regulation will help
improve the accuracy of future ex ante analysis.126

9.2 GENERAL APPROACH

In general, analysts should pursue retrospective benefit-cost analysis for those economically significant
regulations identified in the HHS Plan for Retrospective Review where the need for regulatory reform is not
obvious for other reasons (such as where the regulation requires obsolete technology) and where available data
allow for meaningful assessment of impacts. Below, we first discuss challenges to estimating the effect of the
regulation and addressing the time frame over which the impacts occurred, then describe the overall framework
for the analysis.

9.2.1 ESTIMATING THE IMPACT OF THE REGULATION

The public or decision-makers may presume that retrospective analysis will be more accurate than prospective
analysis because analysts can simply “tally” benefits and costs that have actually occurred. In other words,
retrospective analysis may be perceived as a simple accounting exercise. However, correctly measuring
incremental effects on a retrospective basis presents similar challenges to estimating impacts prospectively and
is also subject to substantial uncertainty. The key challenge to ex post analysis is isolating the incremental effects

121
Such incurred, or “sunk” costs have zero opportunity costs because these resources have already been used and cannot be used again. When the analytic
goal is to determine whether to revise or vacate an existing regulation, and significant costs have been incurred, a prospective analysis of retracting the
existing regulation may be more appropriate than a retrospective evaluation.
122
An agency might also use retrospective review to better understand the cumulative effect of multiple regulations aimed at reducing the same risk.
123
For review of the results of retrospective benefit-cost analyses of Federal regulations see, for example, Harrington et al. (2000), Harrington (2006), and
Morgenstern (2015).
124
A substantial body of literature on “learning by doing” examines declines in the per-unit cost of producing or using a new technology as experience with the
technology increases over time, as discussed in Chapter 4 and EPA (2010).
125
For example, in 2003, FDA promulgated a final regulation requiring that trans fatty acids be declared in the nutrition label of conventional foods and dietary
supplements on a separate line immediately under the line for the declaration of saturated fatty acids. Subsequent review of industry compliance with the
regulation revealed that the informational nature of the regulation and the desire to maintain market share for certain food products created incentives for
the industry to find ways to reduce trans fatty acids in foods to a degree that exceeded FDA’s expectations. As a result, ex ante estimates of costs and health
benefits may have been understated.
126
In addition to improving the accuracy of future ex ante analysis, a better understanding of how affected entities respond to regulations may identify more
efficient methods of achieving similar policy objectives. For example, if analysts learn that assuming complete compliance with future regulations overstates
actual compliance rates, they may determine that increasing enforcement resources for existing regulations will achieve better health outcomes for less cost
than introducing additional regulations.

65
of the regulation. As with ex ante analysis, identifying incremental effects requires comparing two scenarios: the
world with the regulation (the “incremental scenario”) and the world without the regulation (the “baseline
scenario” in ex ante analysis, as discussed in Chapter 2, or “counterfactual scenario” in ex post analysis).127
In ex ante analysis both scenarios occur in the future; neither is observed. Ideally, analysts do not assume that
current conditions will persist in the future; the baseline is the evolution of the existing, observed world. Both
the baseline and incremental scenarios are subject to significant uncertainty associated with assumptions about
likely future health and economic conditions without the regulation, compliance with the regulation, and
behavioral responses that may affect implementation (e.g., innovation by the regulated community).
In ex post analysis, uncertainty may be reduced because the world with the regulation (the incremental
scenario) can be observed. What were included as probabilities or expected values in the ex ante analysis can be
replaced with actual outcomes, to the extent that it is possible to separate the effects of the regulation from
other factors. The agency may have data on compliance rates, or it may be able to obtain more accurate
information on key assumptions, such as the number of units of a drug sold. In other cases, it may be difficult to
separate the effects of the regulation from other factors. For example, the incidence of the health conditions
addressed by the regulation may be rising or falling due to medical innovations, changing demographics, or
other causes. The extent to which the regulation has accelerated a decrease in incidence, or offset what would
have otherwise been an even larger increase, may be difficult to isolate, even using sophisticated statistical
tools. Furthermore, analysts must still model the counterfactual scenario, which cannot be observed.
Assumptions about what the world would have been like without the regulation introduce uncertainty to
estimates of the incremental impacts.128,129
The major components of a retrospective benefit-cost analysis are the same as the RIA components illustrated in
Figure 1.1. Figure 9.2 illustrates the process used to construct new models and highlights differences in the data
and information potentially available for retrospective benefit-cost analysis. The process begins with the
evaluation of existing information and the collection of new data. Relevant information may be obtained from a
variety of sources, including the ex ante analysis previously developed in support of the regulation, newly
available public information, surveys, or other sources. Retrospective analysis, like prospective analysis, is
subject to the requirements of the PRA, which may limit an agency’s ability to conveniently collect new data.

127
The relevant comparison is the world with and without the regulation, not the world before and after the regulation is implemented. For example, a
regulated entity’s operating costs after a regulation takes effect may be influenced by market conditions or other factors unrelated to the regulation. Simply
comparing costs before and after the regulation takes effect, without accounting for these other changes, could be misleading.
128
Ex ante analysis estimates net benefits conditional on one or more sets of assumptions about the future, and sometimes uncertainty with regard to
estimated net benefits may be aggregated over sets of future uncertain factors. In contrast, ex post analysis estimates net benefits conditional on specific
realization of at least some of the ex ante uncertain factors. When using ex post analysis to judge the accuracy of ex ante estimates, this difference affects
the interpretation and must be recognized.
129
Many ex ante actions are undertaken to protect against uncertain adverse events. If the event does not occur, that does not necessarily mean that the
regulation was unwarranted. For example, a vaccination policy does not necessarily have negative net benefits if the disease does not materialize; the
insurance against possible disease provides its own benefit. Uncertainty about the likelihood of occurrence should be considered in the retrospective analysis
as well as the prospective analysis.

66
F I G U R E 9 . 2 . R E T R O S P E C T I V E A N A LY S I S P R O C E S S

67
As indicated by the figure, data from two types of experiments might be available for these analyses: controlled
or quasi-experiments. In the best case, the agency would design the regulation to allow for a controlled
experiment, enabling analysts to empirically estimate the impact of the regulation with a high degree of
confidence by comparing otherwise-identical treatment (i.e., subject to the regulation) and control (i.e., not
subject to the regulation) groups.130 This information on actual effects can replace assumptions about likely
effects in the cost and benefit models. However, implementation of a controlled experiment is often at odds
with regulatory design, which targets the populations in need of intervention or, for fairness, applies equally to
everyone. Alternatively, in certain circumstances opportunities for natural or quasi-experimental designs, where
natural randomization is exploited, may exist. For example, analysts may be able to identify unregulated
comparison groups if (1) the regulation is phased in through time (new products are subject to the regulation
while similar, older products are exempt); or (2) the regulation is not implemented uniformly across all
geographic areas (e.g., implementation may differ across states).131
Such controlled or quasi-experiments may provide the best assessment of the actual effects of existing
regulations because they are based on observed outcomes and data.132 However, in practice, they may be too
small in scale to be extrapolated to a national level, or the conditions necessary for successful experiments may
be unavailable. For example, in many cases Federal regulations apply broadly to the general population. Thus,
comparable control groups do not exist. Comparing populations through time may be more feasible; however,
changes in underlying economic or health conditions may complicate such comparisons. Some of these
challenges may be overcome using simple regression analysis or more sophisticated econometric modeling
techniques.133 In addition, regulations should be designed to ensure that monitoring or other data are available
for use in future retrospective assessments.
The input data and additional analytic results are then used to update existing ex ante models or create new ex
post models, as discussed in greater detail later in this chapter. At the conclusion of the process, decision-
makers would use the results of the ex post modeling effort to evaluate the regulations. The process may be
iterative as new data or insights are identified.

9.2.2 ADDRESSING THE TIMING OF THE I MPACT S

Below, we address two additional technical issues. They include defining the period of analysis and calculating
present value and annualized impacts. While these issues are also relevant to prospective analysis, they may be
addressed differently in retrospective analysis, depending on the period of interest.
Determining the time horizon: In retrospective analysis, defining the relevant time horizon is fairly simple. As
with prospective analysis, the retrospective analysis should start in the year the impacts were first incurred,
even if that period predates the effective date of the final regulation. For example, many regulated entities may
incur costs in anticipation of upcoming regulations as they prepare to meet the regulation’s effective date.
These costs should be included in the analysis.
The end date is determined by the date when the retrospective analysis is undertaken or the most recent date
for which retrospective data are available. To the extent that agencies wish to project impacts into the future
based on new information collected in the retrospective analysis, additional prospective results should be clearly

130
For a discussion of regulatory design intended to foster such experiments, see Greenstone (2009).
131
For example, the National Traffic Highway Safety Administration (NHTSA) often issues standards applying only to new vehicles. Thus, it can estimate the
efficacy of new safety equipment by comparing contemporaneous accident reports for new vehicles to similar records for older vehicles manufactured prior
to the effective date of the final regulation (Lutter 2013).
132
For two examples of these types of experiments conducted in the context of public health policy, see Newhouse and the Insurance Experiment Group (1996)
and Baicker et al. (2013).
133
For an informative discussion of the use of controlled and quasi-experiments in policy evaluation and the statistical analysis of such empirical data, see HM
Treasury (2011), Chapter 9. For additional guidance on the design and conduct of such experiments, see Box et al. (2005).

68
separated and reported, as prospective analysis requires a different set of assumptions to address the future
baseline and incremental scenarios.
Where the benefits and costs of a regulation are expected to occur unevenly through time, the analysts should
consider the full time period over which the regulation was implemented. Longer timeframes may be
particularly important when positive health impacts are not expected to be measurable until many years after
the regulation goes into effect. In such cases, a longer timeframe ensures that all significant one-time benefits
and costs are captured in the analysis. However, if benefits and costs are likely to remain constant through the
period of the analysis, it may be sufficient to model impacts for a single year.134
Finally, if the agency wishes to compare the results of ex ante and ex post analyses, it must model the same time
periods. However, this may not always be possible, particularly if the agency reviews the regulation within the
first few years of implementation. In such cases, analysts should adjust ex ante estimates to exclude years not
analyzed in the ex post analysis. Agencies should also ensure that the identical time periods are covered when
comparing ex ante and ex post estimates of annualized impacts.
Calculating present value and annualized impacts: Regardless of whether impacts occur in the future or the
past, time preferences matter. Resources allocated to compliance in prior years could have been used for other
purposes. Benefits accrued earlier are generally more valuable than those accrued later. If analysts are
interested in comparing the results of the retrospective analysis to the prospective analysis, they should report
benefits and costs in present value terms using the same base year (see Chapter 5). Generally, the starting point
(base year) is the year the regulation went into effect or the first year costs or benefits were incurred.
Alternatively, impacts may be reported on an annualized basis. In either case, the stream of benefits and costs
should also be reported by year and in constant, undiscounted dollars for those years.

9.2.3 FRAMING THE EX POST MODELING EFFORT

Earlier in this chapter, Figure 9.2 describes the general components of retrospective analysis, including inputs,
analysis, and outputs. This section provides additional discussion of the choices analysts face during the analysis,
particularly the ex post modeling effort. Generally, analysts should follow a phased approach to ensure that their
work is carefully focused and useful for decision-making, following the steps listed in Figure 9.3 as discussed
below.
Prior to initiating any retrospective modeling effort, analysts should consider the purpose of the effort, as the
goal may affect the content of the analysis (see Figure 9.1). Based on that purpose, they should develop
reasonable stopping rules to define the scope of the analysis. These rules are designed to focus analysts on
answering the pertinent question related to a particular regulation, while avoiding unnecessary and expensive
data collection and analysis.
Analysts should follow a stepwise progression: (1) simple screening analysis; (2) revisions to existing models
developed for the ex ante regulatory analysis; and (3) entirely new modeling efforts, as indicated in greater
detail in Figure 9.3. For example, if the purpose of the effort is to determine whether the benefits of a regulation
exceed costs, and a simple screening analysis can answer this question, additional modeling efforts may not be
necessary.

134
Such a situation seems unlikely given continued changes in the size of the U.S. economy. But per capita effects might be roughly constant and reporting
them would be perhaps as useful as reporting totals.

69
 FIGURE 9.3. SUGGESTED STEPS FOR EX POST MODELING

LEVEL OF EFFORT STEPS

Screening Analysis
Lowest
 Conduct case studies of incurred costs or benefits.
 Conduct a simple bounding analysis with assumptions based
on observed data.
Adjust Existing Ex Ante Model Assumptions and Data

In addition to the above screening tools,

 Identify the key assumptions or data sources influencing the
impact estimates in the ex ante model.
 Focus retrospective research efforts on refining these
assumptions and data, such as through natural or controlled
experiments or other data collection efforts.
 Update counterfactual and incremental scenarios using the
existing model and this new information.
 Evaluate the validity of existing models and whether they will
achieve the goals of the retrospective analysis (e.g., whether
they accurately depict the response of the regulated
community).
Construct a New Model

In addition to, or in place of, adjustments to the ex ante model,

 Use existing and new information to construct a new model of
Highest impacts.
 Ensure the new model captures missing categories of benefits
and costs or unanticipated responses by the affected
community.

If analysts are interested not just in whether the regulation was effective, but also in the accuracy of the ex ante
cost and benefit estimates, additional modeling may be required. They should first review the key assumptions
or data sources driving the results of the ex ante analysis, particularly if the appropriateness of these is uncertain
and they substantially affect the results. Analysts should focus their research on refining or updating these key
factors.135 Variations of the original ex ante models could be used to estimate the incremental change compared
with the counterfactual scenario. However, such an approach assumes that the original models accurately
characterize the implementation of the regulation and linkages to resulting benefits and costs.136
In some cases, through interviews with affected entities, additional data collection, or the results of controlled
or quasi-experiments, analysts may determine that the ex ante models did not accurately characterize the
impacts of the regulation. For example, compliance costs may be lower than anticipated if affected entities
develop innovative methods of compliance, or improvements in overall productivity reduce all costs, including
compliance costs. Or underlying market conditions may fundamentally change, making substitute sources of

135
It is particularly helpful if the original ex ante analysis clearly identifies key assumptions and sources of uncertainty. Sensitivity analysis can be used to
demonstrate the importance of each uncertain variable.
136
A particularly well-known example of a regulation where the ex ante models did not accurately predict the behavioral response of the regulated community
is the case of EPA’s regulation of sulfur dioxide (SO2) emissions. As described in Harrington et al. (2000), emissions reductions exceeded expectations for
several reasons, including greater than expected efficacy of pollution control equipment, innovation by the regulated community, and changes in market
conditions. The SO2 regulation illustrates circumstances that would necessitate new cost and benefit modeling to accurately estimate the net benefits of the
regulation.

70
goods or materials available to offset costs or benefits. In other cases, the agency may learn that key categories
of benefits and costs were omitted from the original analysis. Based on this new information, analysts may
decide to develop new models of benefits and costs.
In sum, conducting retrospective analysis requires thinking carefully about its goals. In some cases, revisiting the
prospective analysis from an ex post perspective will provide important insights into the benefits and costs of
the regulation. In other cases, prospective analysis of the benefits and costs of eliminating or modifying the
regulation may be useful – instead of, or in addition to, the ex post analysis. In either case, the level of effort
should be tailored to the purpose of the review.

71
Appendix A
Agency Checklist: Regulatory Impact Analysis
(OMB 2010)
This appendix replicates OMB’s 2010 Checklist, which is also available at:
[Link]

With this document, the Office of Information and Regulatory Affairs is providing a checklist to assist agencies in
producing RIAs, as required for economically significant rules by Executive Order 12866 and OMB Circular A-4.
Nothing herein alters, adds to, or reformulates existing requirements in any way. Moreover, this checklist is
limited to the requirements of Executive Order 12866 and Circular A-4; it does not address requirements
imposed by other authorities, such as the National Environmental Policy Act, the Regulatory Flexibility Act, the
Unfunded Mandates Reform Act, the Paperwork Reduction Act, and various Executive Orders that require
analysis. Executive Order 12866 and Circular A-4, as well as those other authorities, should be consulted for
further information.
Checklist for Regulatory Impact Analysis:
 Does the RIA include a reasonably detailed description of the need for the regulatory action?1,2
 Does the RIA include an explanation of how the regulatory action will meet that need?3
 Does the RIA use an appropriate baseline (i.e., best assessment of how the world would look in the absence
of the proposed action)?4
 Is the information in the RIA based on the best reasonably obtainable scientific, technical, and economic
information and is it presented in an accurate, clear, complete, and unbiased manner?5
 Are the data, sources, and methods used in the RIA provided to the public on the Internet so that a qualified
person can reproduce the analysis?6
 To the extent feasible, does the RIA quantify and monetize the anticipated benefits from the regulatory
action?7,8
 To the extent feasible, does the RIA quantify and monetize the anticipated costs?9
 Does the RIA explain and support a reasoned determination that the benefits of the intended regulation
justify its costs (recognizing that some benefits and costs are difficult to quantify)?10
 Does the RIA assess the potentially effective and reasonably feasible alternatives?11
 Does the RIA assess the benefits and costs of different regulatory provisions separately if the rule
includes a number of distinct provisions? 12
 Does the RIA assess at least one alternative that is less stringent and at least one alternative that is
more stringent? 13
 Does the RIA consider setting different requirements for large and small firms? 14
 Does the preferred option have the highest net benefits (including potential economic, environmental,
public health and safety, and other advantages; distributive impacts; and equity), unless a statute requires a
different approach? 15
 Does the RIA include an explanation of why the planned regulatory action is preferable to the identified
potential alternatives?16
 Does the RIA use appropriate discount rates for benefits and costs that are expected to occur in the
future?17
 Does the RIA include, if and where relevant, an appropriate uncertainty analysis?18

A-1
  Does the RIA include, if and where relevant, a separate description of distributive impacts and equity?19
 Does the RIA provide a description/accounting of transfer payments? 20
 Does the RIA analyze relevant effects on disadvantaged or vulnerable populations (e.g., disabled or
poor)? 21
 Does the analysis include a clear, plain-language executive summary, including an accounting statement
that summarizes the benefit and cost estimates for the regulatory action under consideration, including the
qualitative and non-monetized benefits and costs?22
 Does the analysis include a clear and transparent table presenting (to the extent feasible) anticipated
benefits and costs (quantitative and qualitative)?23

NOTES
1. Required under Executive Order 12866, Section 6(a)(3)(B)(i): “The text of the draft regulatory action,
together with a reasonably detailed description of the need for the regulatory action and an explanation
of how the regulatory action will meet that need.”
2. Circular A-4 states: “If the regulation is designed to correct a significant market failure, you should
describe the failure both qualitatively and (where feasible) quantitatively.” (P. 4)
3. See note 1 above.
4. Circular A-4 states: “You need to measure the benefits and costs of a rule against a baseline. This baseline
should be the best assessment of the way the world would look absent the proposed action… In some
cases, substantial portions of a rule may simply restate statutory requirements that would be self-
implementing, even in the absence of the regulatory action. In these cases, you should use a pre-statute
baseline.” (P. 15-16)
5. Circular A-4 states: “Because of its influential nature and its special role in the rulemaking process, it is
appropriate to set minimum quality standards for regulatory analysis. You should provide documentation
that the analysis is based on the best reasonably obtainable scientific, technical, and economic
information available… you should assure compliance with the Information Quality Guidelines for your
agency and OMB’s Guidelines for Ensuring and Maximizing the Quality, Objectivity, Utility, and Integrity
of Information Disseminated by Federal Agencies...” (P. 17). The IQ Guidelines (paragraph V.3.a) define
objectivity to include “whether disseminated information is being presented in an accurate, clear,
complete, and unbiased manner.”
[Link]
6. Circular A-4 states: “A good analysis should be transparent and your results must be reproducible. You
should clearly set out the basic assumptions, methods, and data underlying the analysis and discuss the
uncertainties associated with the estimates. A qualified third party reading the analysis should be able to
understand the basic elements of your analysis and the way in which you developed your estimates. To
provide greater access to your analysis, you should generally post it, with all the supporting documents,
on the internet so the public can review the findings.” (P. 17). OMB IQ Guidelines (paragraph [Link])
further states: “If an agency is responsible for disseminating influential scientific, financial, or statistical
information, agency guidelines shall include a high degree of transparency about data and methods to
facilitate the reproducibility of such information by qualified third parties.”
7. Required under Executive Order 12866, Section 6(a)(3)(C)(i): “An assessment, including the underlying
analysis, of benefits anticipated from the regulatory action (such as, but not limited to, the promotion of
the efficient functioning of the economy and private markets, the enhancement of health and safety, the
protection of the natural environment, and the elimination or reduction of discrimination or bias)
together with, to the extent feasible, a quantification of those benefits.”

A-2
8. Circular A-4 states: “You should monetize quantitative estimates whenever possible. Use sound and
defensible values or procedures to monetize benefits and costs, and ensure that key analytical
assumptions are defensible. If monetization is impossible, explain why and present all available
quantitative information.” (P. 19). Circular A-4 also offers a discussion of appropriate methods for
monetizing benefits that might not easily be turned into monetary equivalents.
9. Required under Executive Order 12866, Section 6(a)(3)(C)(ii): “An assessment, including the underlying
analysis, of costs anticipated from the regulatory action (such as, but not limited to, the direct cost both
to the government in administering the regulation and to businesses and others in complying with the
regulation, and any adverse effects on the efficient functioning of the economy, private markets
(including productivity, employment, and competitiveness), health, safety, and the natural environment),
together with, to the extent feasible, a quantification of those costs;” See also note 6 above.
10. Executive Order 12866, Section 1(b)(6) states that to the extent permitted by law, “[e]ach agency shall
assess both the costs and the benefits of the intended regulation and, recognizing that some costs and
benefits are difficult to quantify, propose or adopt a regulation only upon a reasoned determination that
the benefits of the intended regulation justify its costs.” As Executive Order 12866 recognizes, a statute
may require an agency to proceed with a regulation even if the benefits do not justify the costs; in such a
case, the agency’s analysis may not show any such justification.
11. Required under Executive Order 12866, Section 6(a)(3)(C)(iii): “An assessment, including the underlying
analysis, of costs and benefits of potentially effective and reasonably feasible alternatives to the planned
regulation, identified by the agencies or the public (including improving the current regulation and
reasonably viable nonregulatory actions)...”
12. Circular A-4 states: “You should analyze the benefits and costs of different regulatory provisions
separately when a rule includes a number of distinct provisions.” (P. 17)
13. Circular A-4 states: “you generally should analyze at least three options: the preferred option; a more
stringent option that achieves additional benefits (and presumably costs more) beyond those realized by
the preferred option; and a less stringent option that costs less (and presumably generates fewer
benefits) than the preferred option.” (P. 16)
14. Circular A-4 states: “You should consider setting different requirements for large and small firms, basing
the requirements on estimated differences in the expected costs of compliance or in the expected
benefits. The balance of benefits and costs can shift depending on the size of the firms being regulated.
Small firms may find it more costly to comply with regulation, especially if there are large fixed costs
required for regulatory compliance. On the other hand, it is not efficient to place a heavier burden on
one segment of a regulated industry solely because it can better afford the higher cost. This has the
potential to load costs on the most productive firms, costs that are disproportionate to the damages
they create. You should also remember that a rule with a significant impact on a substantial number
of small entities will trigger the requirements set forth in the Regulatory Flexibility Act. (5 U.S.C. 603(c),
604).” (P. 8)
15. Executive Order 12866, Section 1(a) states: “agencies should select those approaches that maximize net
benefits (including potential economic, environmental, public health and safety, and other advantages;
distributive impacts; and equity) unless a statute requires another regulatory approach.”
16. Required under Executive Order 12866, Section 6(a)(3)(C)(iii): “An assessment, including the underlying
analysis, of costs and benefits of potentially effective and reasonably feasible alternatives to the planned
regulation, identified by the agencies or the public (including improving the current regulation and
reasonably viable nonregulatory actions), and an explanation why the planned regulatory action is
preferable to the identified potential alternatives.”

A-3
17. Circular A-4 contains a detailed discussion, generally calling for discount rates of 7 percent and 3 percent
for both benefits and costs. It states: “Benefits and costs do not always take place in the same time
period. When they do not, it is incorrect simply to add all of the expected net benefits or costs without
taking account of when they actually occur. If benefits or costs are delayed or otherwise separated in
time from each other, the difference in timing should be reflected in your analysis.... For regulatory
analysis, you should provide estimates of net benefits using both 3 percent and 7 percent.... If your rule
will have important intergenerational benefits or costs you might consider a further sensitivity analysis
using a lower but positive discount rate in addition to calculating net benefits using discount rates of 3
and 7 percent.” (PP. 31, 34, 36)
18. Circular A-4 provides a detailed discussion. Among other things, it states: “Examples of quantitative
analysis, broadly defined, would include formal estimates of the probabilities of environmental damage
to soil or water, the possible loss of habitat, or risks to endangered species as well as probabilities of
harm to human health and safety. There are also uncertainties associated with estimates of economic
benefits and costs, such as the cost savings associated with increased energy efficiency. Thus, your
analysis should include two fundamental components: a quantitative analysis characterizing the
probabilities of the relevant outcomes and an assignment of economic value to the projected outcomes.”
(P. 40). Circular A-4 also states: “You should clearly set out the basic assumptions, methods, and data
underlying the analysis and discuss the uncertainties associated with the estimates.” (P. 17)
19. Executive Order 12866, Section 1(b)(5) states; “When an agency determines that a regulation is the best
available method of achieving the regulatory objective, it shall design its regulations in the most cost-
effective manner to achieve the regulatory objective. In doing so, each agency shall consider incentives
for innovation, consistency, predictability, the costs of enforcement and compliance (to the government,
regulated entities, and the public), flexibility, distributive impacts, and equity” (emphasis added). Circular
A-4 states: “The term ‘distributional effect’ refers to the impact of a regulatory action across the
population and economy, divided up in various ways (e.g., income groups, race, sex, industrial sector,
geography)… Your regulatory analysis should provide a separate description of distributional effects (i.e.,
how both benefits and costs are distributed among sub-populations of particular concern) so that
decision makers can properly consider them along with the effects on economic efficiency… Where
distributive effects are thought to be important, the effects of various regulatory alternatives should be
described quantitatively to the extent possible, including the magnitude, likelihood, and severity of
impacts on particular groups.” (P. 14)
20. Circular A-4 states: “Distinguishing between real costs and transfer payments is an important, but
sometimes difficult, problem in cost estimation. . . . Transfer payments are monetary payments from one
group to another that do not affect total resources available to society. . . . You should not include
transfers in the estimates of the benefits and costs of a regulation. Instead, address them in a separate
discussion of the regulation's distributional effects.” (P. 14)
21. Circular A-4 states: “Your regulatory analysis should provide a separate description of distributional
effects (i.e., how both benefits and costs are distributed among sub-populations of particular concern) so
that decision makers can properly consider them along with the effects on economic efficiency. Executive
Order 12866 authorizes this approach. Where distributive effects are thought to be important, the
effects of various regulatory alternatives should be described quantitatively to the extent possible,
including the magnitude, likelihood, and severity of impacts on particular groups.” (P. 14)
22. Circular A-4 states: “Your analysis should also have an executive summary, including a standardized
accounting statement.” (P. 3). OMB recommends that: “Regulatory analysis should be made as
transparent as possible by a prominent and accessible executive summary—written in a “plain language”
manner designed to be understandable to the public—that outlines the central judgments that support

A-4
 regulations, including the key findings of the analysis (such as central assumptions and uncertainties)…If
an agency has analyzed the costs and benefits of regulatory alternatives to the planned action (as is
required for economically significant regulatory actions), the summary should include such information.”
See 2010 Report to Congress on the Benefits and Costs of Federal Regulations and Unfunded Mandates on
State, Local, and Tribal Entities, page 51. Available at:
[Link]
23. Circular A-4 states: “You need to provide an accounting statement with tables reporting benefit and cost
estimates for each major final rule for your agency.” (P. 44). Circular A-4 includes an example of a format
for agency consideration. OMB recommends “that agencies should clearly and prominently present, in
the preamble and in the executive summary of the regulatory impact analysis, one or more tables
summarizing the assessment of costs and benefits required under Executive Order 12866 Section
6(a)(3)(C)(i)-(iii). The tables should provide a transparent statement of both quantitative and qualitative
benefits and costs of the proposed or planned action as well as of reasonable alternatives. The tables
should include all relevant information that can be quantified and monetized, along with relevant
information that can be described only in qualitative terms. It will often be useful to accompany a
simple, clear table of aggregated costs and benefits with a separate table offering disaggregated figures,
showing the components of the aggregate figures. To the extent feasible in light of the nature of the issue
and the relevant data, all benefits and costs should be quantified and monetized. To communicate any
uncertainties, we recommend that the table should offer a range of values, in addition to best estimates,
and it should clearly indicate impacts that cannot be quantified or monetized. If nonquantifiable variables
are involved, they should be clearly identified. Agencies should attempt, to the extent feasible, not
merely to identify such variables but also to signify their importance.” See 2010 Report to Congress on
the Benefits and Costs of Federal Regulations and Unfunded Mandates on State, Local, and Tribal Entities,
page 51. Available at:
[Link]

A-5
Appendix B
Consumer and Producer Surplus
As discussed in Chapter 3, a key assumption that underlies benefit-cost analysis is that benefit values are
determined by the change in the amount by which aggregate WTP exceeds the market price, or “consumer
surplus.” When WTP exceeds price, the individual benefits from the fact that he or she can acquire the good or
service for less than his or her willingness to pay. If price exceeds WTP, the individual would not purchase the
good or service, choosing to use the money for other things. The difference between WTP and price can be
aggregated across individuals to determine the consumer surplus associated with different price levels.
Consumers generally benefit from price decreases, because WTP then exceeds price by a larger amount, and
vice-versa.
This relationship is illustrated by Figure B.1. The horizontal axis represents the quantity of the good (q), the
vertical axis represents its price (p). The market demand curve (D) indicates both consumers’ WTP at each
quantity and the quantity that would be purchased at each price.137 Similarly, the supply curve (S) indicates both
the marginal cost of supply at each quantity and the quantity that would be supplied at each price. The
equilibrium market price is determined by where the two curves intersect. At this point, only consumers whose
WTP exceeds the price purchase the good, and only producers whose cost of supply is less than the price
produce it. For example, at price p1, consumers would purchase quantity q1. The shaded area above the price
line and below the demand curve indicates the amount by which WTP exceeds price; i.e., consumer surplus at
price p1.
F I G U R E B . 1 . C H A N G E I N C O N S U M E R S U R P L U S D U E TO A P R I C E D E C R E A S E

137
Depending on the good or service, the prices represented in this schedule may reflect time costs or other factors that influence demand, in addition to the
“sticker price” viewed by the consumer. Demand curves can also be developed for nonmarket goods, using the techniques described in Chapter 3 to estimate
WTP.

B-1
If the price decreases, the quantity demanded rises as some consumers choose to purchase the good at the
lower price rather than buying other goods or services. If changes in supply lead price to drop from p1 to p2,
consumers would increase their purchases to quantity q2.
When the price falls from p1 to p2, consumers benefit in two ways. First, they pay less for the q1 units they
continue to buy. Second, they buy q2 – q1 additional units for which WTP exceeds p2 but does not exceed p1.
(The size of the increase in q is often summarized by the “demand elasticity,” defined as the proportional change
in q divided by the proportional change in p.) The area marked with diagonal lines indicates the gain in
consumer surplus that results from the price decrease from p1 to p2.138
Similar concepts apply to producers. Regulatory compliance costs may affect the price and quantity of goods
exchanged in the market, leading to changes in producer surplus. These relationships are illustrated by Figure
B.2 for a competitive market.139 In this case, we illustrate a cost increase that results from compliance with a
new regulation. As in the earlier figure, the horizontal axis represents the quantity of the good (q) and the
vertical axis represents its price (p); the market demand curve (D) indicates both consumers’ WTP at each
quantity and the quantity that would be purchased at each price; the supply curve (S) indicates both the
marginal cost of supply at each quantity and the quantity that would be supplied at each price; and the
equilibrium market price is determined by where the supply and demand curves intersect.
If the cost of supplying the good increases as a result of the regulation, the supply curve shifts upwards, from s1
to s2, reducing consumer surplus (the area between the demand curve and the price line). Producer surplus,
which reflects the difference between the market price and supply costs (the area above the supply curve and
below the price line), also decreases. For example, at price p1 producers will supply quantity q1. When supply
costs increase, producers will provide a smaller quantity for each price and demand a higher price for each
quantity. Thus the market price will increase to p2 and the quantity sold will decrease to q2.
The area bounded by the two supply curves and the new quantity line represents the increased cost of
producing the quantity that is demanded at the new price.140 In addition, the reduction in output results in a
deadweight loss represented by the solid triangle, indicating forgone net benefits. This deadweight loss is part of
the costs of the regulation.141 Thus the net reduction in the total surplus (consumer plus producer) is a real cost
to society. The question for analysts is whether these costs are greater or less than aggregate WTP for the
regulation’s benefits.

138
When the price falls, some consumers who purchase the good at p 1 might purchase more units and some who do not purchase it at p1 may purchase at the
lower price p2. In this case, the graph displays aggregate demand by all consumers; it does not indicate what quantity each consumer purchases. A similar
graph could be drawn for an individual consumer.
139
For a more detailed discussion of these concepts, see Boardman et al. (2011).
140
As noted elsewhere, the real resource cost of producing a good may differ from the supply cost when the resource costs are not equal to the private costs,
due to externalities, taxes, subsidies, or monopoly producers.
141
Note that the deadweight loss results from changes in both producer and consumer surplus.

B-2
F I G U R E B . 2 . C H A N G E I N P R O D U C E R S U R P L U S D U E TO A C O S T I N C R E A S E

B-3
Appendix C
Methods for Estimating QALYs
As discussed in Chapter 3, estimating QALYs involves first determining the effect of a health state on HRQL, then
multiplying HRQL by the duration of the health state. While the HRQL associated with a health state is likely to
vary among individuals, in practice a common value is typically used for each state, representing a population
average. This appendix introduces methods for estimating HRQL; more information on the implementation of
these methods and their advantages and limitations is provided in Institute of Medicine (2006).
HRQL can be estimated directly or indirectly. Commonly used direct methods include the standard gamble, time
tradeoff, and visual analog scale, administered in interviews or a survey. The standard gamble approach asks
respondents to compare living the rest of their life (T years) in the health state of interest with a gamble
between living the rest of their life in full health (with probability p) and immediate death (with probability 1 –
p). The probability p* at which the individual is indifferent is his or her HRQL for that health state. This follows
because living the rest of his or her life in the specified health state yields p* T QALYs (i.e., T years weighted by
an HRQL of p*) and the gamble provides an expected value of p* T QALYs (i.e., a p* chance of T QALYs (T years
weighted by an HRQL of 1) plus a complementary chance of zero QALYs (immediate death)).
The time tradeoff approach asks respondents to compare living the rest of their life (T years) in the health state
of interest with living a shorter period (qT years) in full health, followed by death. The value q* at which the
individual is indifferent is his or her HRQL for the health state. This follows because living T years with HRQL q*
provides q* T QALYs, and living q* T years in full health also provides q* T QALYs.
The visual analog scale does not require a comparison of different future lives. It simply asks the individual to
rate the health state of interest on a visual scale where one end is described as being as bad as dead and labeled
0, and the other is described as full health and labeled 100. (Alternatively, the individual may be asked to report
a number between 0 and 100 rather than marking it on the scale.) HRQL is then defined as the response divided
by 100.
An indirect method to estimate HRQL is to apply one of several generic HRQL indices, examples of which include
the EurQol- (EQ)-5D, the Health Utilities Index (HUI), and the Quality of Well-Being (QWB) scale. Each describes
health status by employing a classification system with several dimensions. In the case of the EQ-5D, these
include mobility, self-care, usual activities, pain, and anxiety and depression. A particular health state is rated
within each dimension; for example, as causing no, some, or extreme mobility problems. The HRQL associated
with each health state is then calculated by applying a scoring function, developed by eliciting HRQL for some of
the health states through a population survey using one of the direct methods described earlier. These indices
have the advantage of standardizing the approach for describing each health state and providing a convenient
method to calculate HRQL. The results will vary, however, depending on which index is applied, given
differences in the attributes they include and in the scoring functions.
Once HRQL is determined for a particular health state, it is multiplied by the duration of that state to estimate
the associated QALYs. The QALYs can then be summed across health states (e.g., acute and chronic phases)
associated with a particular illness, and across the illnesses associated with a particular hazard. For regulatory
analysis, health status with the regulation must be compared to health status in the absence of the regulation,
which is likely to be less than full health. In particular, health status generally deteriorates with age, so that
average HRQL for older individuals is generally less than 1.0 (see, for example, Hamner et al. 2006). Expected
QALYs are calculated by weighting the HRQL experienced in each future year of life by the probability of living
that year (i.e., by the survival curve). In addition, future QALYs are usually discounted using the same discount
rates as for monetary values.

C-1
References
Akobundu, E., J. Ju, L. Blatt, and C.D. Mullins. 2006. “Cost-of-Illness Studies: A Review of Current Methods.”
Pharmacoeconomics. 24(9): 869-890.
Aldy, J.E. 2014. Learning from Experience: An Assessment of the Retrospective Reviews of Agency Rules and the
Evidence for Improving the Design and Implementation of Regulatory Policy. Prepared for the Administrative
Conference of the United States. [Link]
Aldy, J.E. and W.K. Viscusi. 2007. “Age Differences in the Value of Statistical Life: Revealed Preference Evidence.”
Review of Environmental Economics and Policy. 1(2): 241-260.
Alolayan, M.A., J.S. Evans, and J.K. Hammitt. 2015. “Valuing Mortality Risk in Kuwait: Stated-Preference with a
New Consistency Test.” Environmental and Resource Economics. Early view.
Arias, E. 2014. “United States Life Tables, 2009.” National Vital Statistics Reports. 62(7).
[Link]
Auerswald, P., S. Kauffman, J. Lobo, and K. Shell. 2000. “The Production Recipes Approach to Modeling
Technological Innovation: An Application to Learning by Doing.” Journal of Economic Dynamics & Control. 24:
389-450.
Baicker, K., S.L. Taubman, H.L. Allen, M. Bernstein, J.H. Gruber, J.P. Newhouse, E.C. Schneider, B.J. Wright, A.M.
Zaslasky, and A.N. Finkelstein. 2013. “The Oregon Experiment – Effects of Medicaid on Clinical Outcomes.” New
England Journal of Medicine. 368: 1713-1722.
Baloff, N. 1971. “Extension of the Learning Curve – Some Empirical Results.” Operational Research Quarterly
(1970-1977). 22(4): 329-340.
Berck, P. and S. Hoffman. 2002. “Assessing the Employment Impacts of Environmental and Natural Resource
Policy.” Environmental and Resource Economics. 22: 133-156.
Bloom, B.S., D.J. Bruno, D.Y. Maman, and R. Jayadevappa. 2001. “Usefulness of U.S. Cost of Illness Studies in
Healthcare Decision Making.” Pharmacoeconomics. 19(2): 207-213.
Boardman, A.E., D.H. Greenberg, A.R. Vining, and D.L. Weimer. 2011. Cost-Benefit Analysis: Concepts and
Practice (Fourth Edition). Upper Saddle River, N.J.: Pearson.
Box, G.E., W.G. Hunter, J.S. Hunter. 2005. Statistics for Experimenters: Design, Innovation, and Discovery.
Hoboken, N.J.: John Wiley & Sons.
Cameron, T.A. and J.R. DeShazo. 2013. “Demand for Health Risk Reductions.” Journal of Environmental
Economics and Management. 65: 87-109.
Centers for Disease Control and Prevention, Division for Heart Disease and Stroke Prevention, State Heart
Disease and Stroke Prevention Program. 2007. Evaluation Guide: Developing and Using a Logic Model.
[Link]
Clinton, W.J. 1993. “Executive Order 12866: Regulatory Planning and Review.” Federal Register. 58(190): 51735-
51744. [Link]
Clinton, W.J. 1994. “Executive Order 12898: Federal Actions to Address Environmental Justice in Minority
Populations and Low-Income Populations.” Federal Register. 59(32): 7629-7633.
[Link]

References-1
Clinton, W.J. 1997. “Executive Order 13045: Protection of Children from Environmental Health Risks and Safety
Risks.” Federal Register. 62(78): 19885-19888. [Link]
orders/[Link]
Clinton, W.J. 1999. “Executive Order 13132: Federalism.” Federal Register. 64(153):43255-43259.
[Link]
Congressional Budget Office. 2015. The 2015 Long-Term Budget Outlook.
[Link]
Cooke, R.M. 1991. Experts in Uncertainty: Opinion and Subjective Probability in Science. New York, N.Y.: Oxford
University Press.
Corso, P.S., J.K. Hammitt, and J.D. Graham. 2001. “Valuing Mortality-Risk Reduction: Using Visual Aids to
Improve the Validity of Contingent Valuation.” Journal of Risk and Uncertainty. 23(2): 165-184.
Cropper, M., J.K. Hammitt, and L.A. Robinson. 2011. “Valuing Mortality Risk Reductions: Progress and
Challenges.” Annual Review of Resource Economics. 3: 313-336.
Cutler, D.M., A. Jessup, D. Kenkel, and M.A. Starr. 2015. “Valuing Regulations Affecting Addictive or Habitual
Goods.” Journal of Benefit-Cost Analysis. 6:247-280.
[Link]
S2194588815000445
Drummond, M.F., M.J. Sculpher, K. Claxton, G.L. Stoddart, and G.W. Torrance. 2015, Methods for the Economic
Evaluation of Health Care Programmes, 4th Edition. Oxford, U.K.: Oxford University Press.
Dutton, J.M. and A. Thomas. 1984. “Treating Progress Functions as a Managerial Opportunity.” Academy of
Management Review. 9(2): 235-247.
Epple, D., L. Argote, and R. Davadas. 1991. “Organizational Learning Curves: A Method for Investigating Intra-
plant Transfer of Knowledge Acquired Through Learning by Doing.” Organizational Science. 2(1): 58-70.
Freeman, A.M. III, J.A. Herriges, and C.L. Kling. 2014. The Measurement of Environmental and Resource Values:
Theory and Methods (Third Edition). New York, N.Y.: RFF Press.
Greenstone, M. 2009. “Toward a Culture of Persistent Regulatory Experimentation and Evaluation.” Published in
New Perspectives on Regulation. D. Moss and J. Cisterino (eds.). Cambridge, M.A.: The Tobin Project, Inc.
Hammitt, J.K. 2000. “Valuing Mortality Risk: Theory and Practice.” Environmental Science and Technology. 34:
1396-1400.
Hammitt, J.K. 2002. “QALYs versus WTP.” Risk Analysis. 22(5): 985-1001.
Hammitt, J.K. 2007. “Valuing Changes in Mortality Risk: Lives Saved versus Life Years Saved.” Review of
Environmental Economics and Policy. 1(2): 228-240.
Hammitt, J.K. 2013. “Admissible Utility Functions for Health, Longevity, and Wealth: Integrating Monetary and
Life-Year Measures.” Journal of Risk and Uncertainty. 47: 311-325
Hammitt, J.K. and K. Haninger. 2010. “Valuing Fatal Risks to Children and Adults: Effects of Disease, Latency, and
Risk Aversion.” Journal of Risk and Uncertainty. 40: 57-83.
Hammitt, J.K. and L.A. Robinson. 2011. “The Income Elasticity of the Value per Statistical Life: Transferring
Estimates Between High and Low Income Populations.” Journal of Benefit-Cost Analysis. 2: Art. 1.
[Link]
S2152281200000024

References-2
Hanmer, J. W.F. Lawrence, J.P. Anderson, R.M. Kaplan, D.G. Fryback. 2006. “Report of Nationally Representative
Values for the Noninstitutionalized US Adult Population for 7 Health-Related Quality-of-Life Scores.” Medical
Decision Making. 26(4): 391-400.
Haninger, K. and J.K. Hammitt. 2011. “Diminishing Willingness to Pay per Quality-Adjusted Life Year: Valuing
Acute Foodborne Illness.” Risk Analysis. 31(9): 1363-1380.
Harrington, W., R. Morgenstern, and P. Nelson. 2000. “On Accuracy of Regulatory Cost Estimates.” Journal of
Policy Analysis and Management. 19(2): 297-322.
Harrington, W. 2006. Grading Estimates of the Benefits and Costs of Federal Regulation: A Review of Reviews.
Resources for the Future Discussion Paper 06-39.
[Link]
Hersch, J. and W.K. Viscusi. 2010. “Immigrant Status and the Value of Statistical Life.” Journal of Human
Resources. 45: 749-771.
Hirth, R.A., M.E. Chernew, E. Miller, A.M. Fendrick, and W.G. Weissert. 2000. “Willingness to Pay for a Quality-
adjusted Life Year: In Search of a Standard.” Medical Decision Making. 20: 332-342.
HM Treasury. 2011. The Magenta Book: Guidance for Evaluation.
[Link]
Institute of Medicine. 2006. Valuing Health for Regulatory Cost-Effectiveness Analysis. (W. Miller, L.A. Robinson,
and R.S. Lawrence, eds.) Washington, D.C.: The National Academies Press.
[Link]
Kling, C.L. et al. 2011. “Review of ‘Valuing Mortality Risk Reductions for Environmental Policy: A White Paper’
(December 10, 2010).” Memorandum to Lisa P. Jackson, EPA Administrator, from the EPA Science Advisory
Board and Environmental Economics Advisory Committee. EPA-SAB-11-011.
[Link]
0400712aeb!OpenDocument&TableRow=2.3#2
Kniesner, T.J. and W.K. Viscusi. 2005. “Value of a Statistical Life: Relative Position vs. Relative Age.” American
Economic Review. 95(2): 142-146.
Kniesner, T. J., W.K. Viscusi, and J.P. Ziliak. 2010. “Policy Relevant Heterogeneity in the Value of Statistical Life:
New Evidence from Panel Data Quantile Regressions.” Journal of Risk and Uncertainty. 40: 15-31.
Krupnick, A. 2007. “Mortality-risk Valuation and Age: Stated Preference Evidence.” Review of Environmental
Economics and Policy. 1(2): 261-282.
Larg, A. and John R. Moss. 2011. “Cost-of-Illness Studies: A Guide to Critical Evaluation.” Pharmacoeconomics.
29(8): 653-671.
Lee, J.M. and L.O. Taylor. 2013. “Randomized Safety Inspections and Risk Exposure on the Job: Quasi-
Experimental Estimates of the Value of a Statistical Life.” Center for Economic Studies, U.S. Census Bureau.
Discussion Paper CES 14-05.
Lipscomb, J., M. Drummond, D. Fryback, M. Gold, and D. Revicki. 2009. “Retaining, and Enhancing, the QALY.”
Value in Health. 12(Supp. 1): S18-S26.
Lofgren, H., R.L. Harris, and S. Robinson. 2002. “A Standard Computable General Equilibrium (CGE) Model in
GAMS.” International Food Policy Research Institute.
Lund, J.L., K.R. Yabroff, Y. Ibuka, L.B. Russell, P.G. Barnett, J. Lipscomb, W.F. Lawrence, and M.L. Brown. 2009.
“Inventory of Data Sources for Estimating Health Care Costs in the United States.” Medical Care. 47(7, Supp. 1):
S127-S142.
References-3
Lutter, R. 2013. “Regulatory Policy: What Role for Retrospective Analysis and Review?” Journal of Benefit-Cost
Analysis. 4(1): 17-38.
Morgan, M.G. and M. Henrion. 1990. Uncertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and
Policy Analysis. Cambridge, U.K.: Cambridge University Press
Morgenstern, R. 2013. “Analyzing the Employment Impacts of Regulation.” Published in Does Regulation Kill
Jobs? C. Coglianese, A. Finkel, and C. Carrigan (eds.). Philadelphia, P.A.: University of Pennsylvania Press.
Morgenstern, R.D. 2015. The RFF Regulatory Performance Initiative: What Have We Learned? Resources for the
Future Discussion Paper 15-47. [Link]
Newhouse, J.P. and the Insurance Experiment Group. 1996. Free for All? Lessons from the RAND Health
Insurance Experiment. Cambridge, M.A.: Harvard University Press.
O’Hagan, A., C.E. Buck, A. Daneshkah, J. Eiser, P. Garthwaite, D. Jenkinson, J. Oakely, T. Rakow. 2006. Uncertain
Judgements: Eliciting Experts’ Probabilities. Chichester, U.K.: John Wiley & Sons Ltd.
Obama, B. 2011. “Executive Order 13563: Improving Regulation and Regulatory Review.” Federal Register.
76(14): 3821-3823. [Link]
Obama, B. 2012. “Executive Order 13609: Promoting International Regulatory Cooperation.” Federal Register.
77(87): 26413-26415. [Link]
international-regulatory-cooperation
Raiffa, H. 1968. Decision Analysis. Reading, M.A.: Addison-Wesley.
Robinson, L.A. and J.K. Hammitt. 2011. “Behavioral Economics and Regulatory Analysis.” Risk Analysis. 31(9):
1408-1422.
Robinson, L.A. and J.K. Hammitt. 2013. “Skills of the Trade: Valuing Health Risk Reductions in Benefit-Cost
Analysis.” Journal of Benefit-Cost Analysis. 4(1): 107-130.
[Link]
S2194588800000518
Robinson, L.A. and J.K. Hammitt. 2016. “Valuing Reductions in Fatal Illness Risks: Implications of Recent
Research.” Health Economics. 25(8):1039-1052.
Scotton, C.R. 2013. “New Risk Rates, Inter-Industry Differentials and the Magnitude of VSL Estimates.” Journal of
Benefit-Cost Analysis. 4(1): 39-80.
Shogren, J.F. 2005. “Experimental Methods and Valuation.” Published in Handbook of Environmental Economics.
K-G Maler and J.R. Vincent (eds.). 2: 969-1027.
Small Business Administration. 2012. A Guide for Government Agencies: How to Comply with the Regulatory
Flexibility Act. [Link]
Smith, V.K., S.K. Pattanayak, and G.L. van Houtven. 2006. “Structural Benefits Transfer: An Example Using VSL
Estimates.” Ecological Economics. 60: 361-371.
Sullivan, P.W. and V. Ghushchyan. 2006. “Preference-Based EQ-5D Index Scores for Chronic Conditions in the
United States.” Medical Decision Making. 26(4): 410-420. [Link]
Sundra, D.L., J. Scherer, and L. Anderson. 2003. A guide on logic model development for CDC’s Prevention
Research Centers. Atlanta, GA: Centers for Disease Control and Prevention, Prevention Research Centers
Program Office. [Link]

References-4
Sunstein, C. 2010a. “Disclosure and Simplification as Regulatory Tools.” Memorandum for the Heads of
Executive Departments and Agencies from the Administrator, Office of Information and Regulatory Affairs, U.S.
Office of Management and Budget. [Link]
Sunstein, C. 2010b. “Information Collection under the Paperwork Reduction Act.” Memorandum for the Heads
of Executive Departments and Agencies, and Independent Regulatory Agencies from the Administrator, Office of
Information and Regulatory Affairs, U.S. Office of Management and Budget.
[Link]
Thorat, T., M. Cangelosi, and P.J. Neumann. 2012. “Skills of the Trade: The Tufts Cost-Effectiveness Analysis
Registry.” Journal of Benefit-Cost Analysis. 3(1): Art. 1.
[Link]
S2194588800000336
U.S. Department of Health and Human Services. 2003. Guidance on Proper Consideration of Small Entities in
Rulemakings of the U.S. Department of Health and Human Services.
U.S. Department of Health and Human Services. 2011. Plan for Retrospective Review of Existing Rules.
[Link]
U.S. Department of Health and Human Services. 2015. Annual Update to the Unfunded Mandates Reform Act
Threshold for 2014 is $141 million.
U.S. Department of Transportation. 2015a. “Guidance on Treatment of the Economic Value of a Statistical Life
(VSL) in Departmental Analyses – 2015 Adjustment.” Memorandum to Secretarial Officers and Modal
Administrators from K. Thomson, General Counsel, and C. Monje, Assistant Secretary for Policy.
[Link]
U.S. Department of Transportation. 2015b. “Revised Departmental Guidance on Valuation of Travel Time in
Economic Analysis.” Memorandum to Secretarial Officers and Modal Administrators from C. Monje, Assistant
Secretary for Transportation Policy; prepared by R. Endorf, Economist.
[Link]
U.S. Environmental Protection Agency. 2007. Benefits and Costs of the Clean Air Act – Direct Costs and
Uncertainty Analysis. Report to the Honorable Stephen L. Johnson, Administrator, U.S. Environmental Protection
Agency from James K. Hammitt, Chair, Advisory Council on Clean Air Compliance Analysis. EPA-Council-07-002.
U.S. Environmental Protection Agency. 2010. Valuing Mortality Risk Reductions for Environmental Policy: A
White Paper (Review Draft). Prepared by the National Center for Environmental Economics for consultation with
the Science Advisory Board – Environmental Economics Advisory Committee.
[Link]
U.S. Environmental Protection Agency. 2014. Guidelines for Preparing Economic Analysis. EPA 240-R-10-001.
[Link]
U.S. Food and Drug Administration. 2003. “Food Labeling: Trans Fatty Acids in Nutrition Labeling, Nutrient
Content Claims, and Health Claims: Final Regulation.” Federal Register. 68: 41434-41506.
U.S. Office of Management and Budget. 1995. “Guidance for Implementing Title II of S.1.” Memorandum to the
Heads of Executive Departments and Agencies from Alice M. Rivlin, Director.
[Link]
U.S. Office of Management and Budget. 2003. Circular A-4: Regulatory Analysis.
[Link]

References-5
U.S. Office of Management and Budget. 2005. Validating Regulatory Analysis: 2005 Report to Congress on the
Costs and Benefits of Federal Regulations and Unfunded Mandates on State, Local, and Tribal Entities.
[Link]
U.S. Office of Management and Budget. 2008. 2008 Report to Congress on the Benefits and Costs of Federal
Regulations and Unfunded Mandates on State, Local, and Tribal Entities.
[Link]
U.S. Office of Management and Budget. 2010. Agency Checklist: Regulatory Impact Analysis.
[Link]
U.S. Office of Management and Budget. 2011a. Regulatory Impact Analysis: Frequently Asked Questions (FAQs).
[Link]
U.S. Office of Management and Budget. 2011b. Regulatory Impact Analysis: A Primer.
[Link]
U.S. Office of Management and Budget. 2012. Clarifying Regulatory Requirements: Executive Summaries.
[Link]
U.S. Office of Management and Budget and the Secretariat General of the European Commission. 2008. Review
of the Application of EU and US Regulatory Impact Assessment Guidelines on the Analysis of Impacts on
International Trade and Investment. [Link]
Viscusi, W.K. 2004. “The Value of Life: Estimates with Risks by Occupation and Industry.” Economic Inquiry.
42(1): 29-48.
Viscusi, W.K. 2013. “Using Data from the Census of Fatal Occupational Injuries (CFOI) to Estimate the ‘Value of a
Statistical Life’.” Monthly Labor Review. Bureau of Labor Statistics
Viscusi, W.K. 2015. “The Role of Publication Selection Bias in Estimates of the Value of a Statistical Life.”
American Journal of Health Economics. 1(1): 27-52.
Viscusi W.K. and J.E. Aldy. 2003. The Value of a Statistical Life: A Critical Review of Market Estimates throughout
the World. Journal of Risk and Uncertainty. 27(1): 5–76.
Weimer, D.L. and A.R. Vining. 2011. Policy Analysis: Concepts and Practices (Fifth Edition). Upper Saddle River,
N.J.: Pearson.
Wholey, J.S., H.P. Hatry, and K.E. Newcomer (eds.). 2010. Handbook of Practical Program Evaluation. San
Francisco, CA: Jossey-Bass.
Yabroff, K.R. et al. (eds.). 2009. “Health Care Costing: Data, Methods, Future Directions.” Medical Care. 47(7):
Supplement 1.

References-6
Index
Accounting Costs, 23 Net Present Value, 38-42
Accounting Statement, U.S. Office of Management Nominal Value, 36-37
and Budget (OMB) , 42, 61-63, A-2-5 Nonquantified Effects, 24, 43, 47-51, 61
Administrative Costs, 25, 31 OMB Circular A-4, 2-3, 5-8, 22, 37, 39-41, 44, 46-47,
Annualization, 41-42 52-53, 57-58, 60-61, A-1-5.
Baseline,Without Regulation, Counterfactual, 1, 5- Operations and Maintenance (O&M) Costs, 25, 32-
7, 10, 20, 40, 52, 54, 58, 60, 66, 69, A-1, A-2 33
Benefit Transfer, 12-13, 18-19, 47 Opportunity Cost, 6, 23-30, 32-34, 37, 38-39, 65-66
Bounding Analysis, 45, 48-51, 70 Paperwork Reduction Act, 2, 58, A-1
Breakeven Analysis, 18, 46, 8-49 Partial Equilibrium Models, 35, 54, 58
Capital Costs, , 32-34 Present Value, 17, 36-42, 63, 68-69
Compliance Costs, 24-26, 34-35, 54, 57-58, 70, B-2 Probabilistic Analysis, 44-47
Consumer Price Index, 16, 37-38 Producer Surplus, 23-24, 35, 54, B-1-3
Consumer Surplus, 11, 24, B-1-2 Quality Adjusted Life Year (QALY), 16-22, 33, 49, 54,
Cost-Effectiveness Analysis, 19, 49 C-1
Deadweight Loss, 23-24, B-2 Real Value, 36-37, 39
Discount, Discount Rate, Discounting, 8, 14, 17-21, Regulatory Alternatives, 6-8, 39, 41, 46-47, 51, 57,
36-42, 61-63, 69, A-1, A-4, C-1 60, 64, A-1, A-3-5
Economically Significant Regulation, 2-3, 52, 64-65, Regulatory Flexibility Act (RFA), 56-57
A-5 Revealed Preference Method, 12, 15
Elasticity, 16, B-2 Screening Analysis, 2, 5-9, 24-25, 44, 52-53, 55, 57,
EQ-5D, 19-20, C-1 69-70
Executive Order 12866, 1-3, 5-6, 52, 57, A-1-5 Sensitivity Analysis, 7, 15-17, 30-31, 44-47, 50, 70,
Executive Order 12898, 52-53 A-4
Executive Order 13045, 52-53 Significant Regulation, 2-3, 52, 56-57, 64-65, A1, A5
Executive Order 13132, 2, 56-57 Small Business Regulatory Enforcement Fairness Act
Executive Order 13563, 1-2, 52, 57, 64 (SBREFA), 2, 56-57,
Experiments, Controlled or Quasi, 68-70, Social Cost, 24-25, 35, 52, 54
Federalism, 2, 56-57 Standing, 7
Final Regulatory Flexibility Analysis (FRFA), 58 Stated Preference Method, 12, 15
General Equilibrium Models, 24, 35, 54, 58 Statistical Case, 10, 14, 17, 49-51, 54
Gross Domestic Product (GDP) Implicit Price Timeframe, Analytic, 7, 54, 69
Deflator,35, 37-38 Transfer Payment, 2-3, 23-25, 33-34, 53-54, 61-62,
Health-Related Quality of Life, Health-Related A-2, A-4
Quality of Life (HRQL), 18-21, 54-55, C-1 Uncertainty, 2-4, 8, 43-45, 47, 61, 65-66, 70, A-1
Health Utilities Index, C-1 Unfunded Mandates Reform Act (UMRA), 2, 56-57,
Income Elasticity, 16 60, A-1, A-5
Inflation, 15-16, 20, 36-38, 57-58 Value of Time, 25-32
Initial Regulatory Flexibility Analysis (IRFA), 56 Value per Statisical Life (VSL), 13-18, 20-21, 33, 49
International Effects, 35, 58, 59-60 Value per Statistical Life Year (VSLY), 13-14, 17, 20-
Market Failure, 5, 58, A-2 21
Median vs. Mean, Use of, 16, 29-32 Variability, 43-44
Medical Costs, 18, 21, 25, 33-34 Willingness to Pay (WTP), 11-14, 16, 18, 20-23, 23,
Monte Carlo Analysis, 45-47 33, B-1-2

Index-1
Glossary
Accounting Costs: Actual expenses plus depreciation of capital equipment (Chapter 4).
Annualized Value: The constant annual amount, which, if paid each year over a defined time period, has the
same present value as a specified series of unequal payments over the same period (Chapter 5).
Baseline: Expected future conditions in the absence of a new regulation or other policy change (Chapter 2).
Benefits: For the purpose of HHS regulatory analysis, the value of the intended outcomes of a regulation or
other policy, such as reductions in mortality or morbidity risks, as well as any countervailing effects on these
outcomes, such as health risk increases. Note that analyses not subject to this guidance may use differing
definitions when categorizing outcomes as benefits or costs (Chapter 2).
Benefit Transfer: The application of values from the available research to a policy context that differs in some
respects from the context studied. Involves evaluating the quality of the research and its applicability to the
policy context (Chapter 3).
Bounding Analysis: The application of reasonable high and low parameter values to determine the extent to
which the analytic results might change given the likely variation in the values (Chapter 6).
Breakeven Analysis: The value of an unknown or uncertain parameter at which benefits and costs would be
equal, indicating how large the value would need to be to bridge the gap between the quantified benefits and
costs. Also referred to as “threshold” analysis (Chapter 6).
Capital Cost: The value of resources, including equipment, buildings, and land, that are not immediately
consumed in the production process (Chapter 4).
Compliance Cost: The value of resources, including labor, capital, and materials, used to implement a regulation
or other policy. Includes only those resources expended by the entities and individuals directly responsible for
implementation; excludes impacts on prices or other market conditions (Chapter 4).
Consumer Price Index (CPI): An index maintained by the U.S. Bureau of Labor Statistics that indicates changes in
the prices paid by consumers for a market basket of goods and services over time. May be used to adjust values
measured in current dollars to a common dollar year so that analyses can be conducted in real dollars, avoiding
the need to adjust for expected inflation (Chapter 5).
Consumer Surplus: The difference between the maximum an individual would be willing to pay for a good or
service and the market price (Chapter 3, Appendix B).
Costs: For the purpose of HHS regulatory analysis, the value of the inputs required to implement a regulation or
other policy, including labor, capital, and materials, as well as any offsetting savings. Note that analyses not
subject to this guidance may use differing definitions when categorizing outcomes as benefits or costs (Chapter
2).
Deadweight Loss: The net loss in consumer and producer surplus that accrues when government intervention or
other factors prevent the market from reaching a competitive equilibrium (Appendix B).
Discounting: The process for converting values that accrue in different years to their present value, to reflect
individual time preferences and the value of investments forgone (Chapter 5).
Distribution: The allocation of benefits, costs, or net benefits across different population groups, defined, for
example, by income level (Chapter 7).
Experiments: Comparison of outcomes across groups who are similar or identical except for their exposure to a
regulation or other policy (Chapter 9).
Glossary-1
Gross Domestic Product (GDP) Implicit Price Deflator: A measure reported by the U.S. Bureau of Economic
Analysis that indicates the ratio of the market value of goods and services in current dollars to its the value in
chained (constant) dollars. May be used to adjust values measured in current dollars to a common dollar year so
that analyses can be conducted in real dollars, avoiding the need to adjust for expected inflation (Chapter 5).
General Equilibrium Models: Models that can be used to estimate the economy-wide impact of a regulation or
other policy with large impacts (Chapter 4).
Health-Related Quality of Life (HRQL): A numerical indicator of health status estimated using a scale anchored
at zero and one, where one corresponds to full health and zero corresponds to a state that is as bad as dead
(Chapter 3, Appendix C).
Income Elasticity: The proportional change in price or quantity associated with a change in real income. When
used in estimating the VSL, it indicates the proportional change in value (i.e., unit price) associated with an
income change (Chapter 3).
Inflation: Economy-wide increases in prices (Chapter 5).
Net Benefits: The difference, benefits minus costs (Chapter 2).
Nominal Value: Values expressed in current-year dollars, reflecting the effects of both inflation and real changes
in value over time (Chapter 5).
Opportunity Cost: The benefits of the best alternative use of specified resources, which is forgone when
resources are used for one purpose and hence cannot be used for other purposes (Chapter 4).
Partial Equilibrium Models: Models that describe the effects of a regulation or other policy in one market,
which can be used to estimate the impact on an industry or group of industries (Chapter 4).
Present Value: The value of a stream of benefits, costs, or net benefits discounted to reflect their value in a
common year (Chapter 5).
Probabilistic Analysis: The use of distributions of parameter values to explore the effects of uncertainty on an
analytic result. Often employs Monte Carlo simulation techniques, which involve taking multiple random draws
from the distribution for each critical parameter, calculating the model output for each draw, and using the
results to represent the distribution of the outcome measure (Chapter 6).
Producer Surplus: The difference between the revenue producers receive and their cost of production (Chapter
4, Appendix B).
Quality-Adjusted Life Year (QALY): A nonmonetary measure that integrates the duration and severity of illness.
Calculated by multiplying the amount of time an individual spends in a health state by the HRQL associated with
that state, and summing over health states (Chapter 3, Appendix C).
Real Value: Values adjusted to a common dollar year (constant dollars), removing the effects of inflation
(Chapter 5).
Retrospective Analysis (ex post): Assessment of the impacts of a regulation or a policy after it has been
implemented, looking back to compare its impacts to what might have otherwise occurred, in contrast to
prospective (ex ante) analysis which involves predicting future impacts (Chapter 9).
Revealed Preference Methods: Estimation of values based on observed market prices or behaviors (Chapter 3).
Screening Analysis: Use of readily available information and simple assumptions to provide preliminary
information on potential impacts; may aid in targeting future work (Chapter 2).
Sensitivity Analysis: Varying one or more key parameter values to explore the effects of uncertainty on the
analytic results (Chapter 6).

Glossary-2
Social cost: The sum of the opportunity costs associated with the implementation of a regulation or other policy
(Chapter 4).
Standing: The definition of whose benefits and costs are to be counted in an analysis. For HHS regulatory
analysis, generally includes all U.S. residents (Chapter 2).
Stated Preference Methods: Estimation of values based on surveys or other self-reported data (Chapter 3).
Statistical Cases: Risk changes summed over the affected population; for example, if 10,000 people each
experience a risk reduction of 1 in 10,000, then one statistical case has been averted (Chapter 3).
Transfer Payment: Monetary payments between individuals or groups that do not affect the total resources
available to society (Chapter 4).
Uncertainty: Lack of knowledge about a parameter value that could be addressed by more research (Chapter 6).
Value per Quality-Adjusted Life Year (QALY): The marginal rate of substitution between money in a defined
period and health-adjusted life years remaining; often approximated by dividing a value per statistical life (VSL)
estimate by expected remaining QALYs (Chapter 3).
Value per Statistical Life (VSL): The marginal rate of substitution between money in a defined time period and
mortality risk; often approximated by dividing individual willingness to pay for a small risk change by the risk
change (Chapter 3).
Value per Statistical Life Year (VSLY): The marginal rate of substitution between money in a defined period and
life years remaining; often approximated by dividing a VSL estimate by remaining life expectancy (Chapter 3).
Variability: “Real world” heterogeneity of a parameter value (Chapter 6).
Willingness to Pay (WTP): The maximum amount of money an individual would exchange to obtain an
improvement, given his or her budget constraints, such that his or her wellbeing is as good with the
improvement and having made the payment as without (Chapter 3).

Glossary-3