NCEE Working Paper
The Importance of Source-Side
Effects for the Incidence of Single
Sector Technology Mandates and
Vintage Differentiated Regulation
Alex Marten
Working Paper 19-03
March, 2019
U.S. Environmental Protection Agency	fgf
National Center for Environmental Economics	fw
https://www.epa.gov/environmental-economics	e n v' r o n m e n taI\c°n o m i cs

-------
The Importance of Source-Side Effects for the Incidence of Single Sector
Technology Mandates and Vintage Differentiated Regulation
Alex Marten1,2
Abstract
Understanding the distribution of regulatory costs is key to evaluating whether a rulemaking exacerbates
or ameliorates preexisting economic disparities and is of stated interest to many stakeholders and policy
makers. Previous studies on the incidence of command-and-control environmental regulations have
predominantly focused on the distribution of costs through final goods prices (the use side). However, the
impact of regulations on household income (the source side) can be of first-order importance in
determining the overall incidence. Using a detailed computable general equilibrium model of the U.S.
economy we study the incidence of single-sector technology mandates across a broad set of industrial
sectors. We find the use-side incidence is notably regressive but the source-side effects are progressive
on average and tend to dominate the overall incidence of costs. This occurs as a significant share of
regulatory costs is passed on through lower returns to capital and natural resources, which predominantly
affects upper-income households, while indexed transfer payments partially shields the purchasing power
of low-income households from increases in output prices. However, when the regulated sector
predominantly produces final goods with inelastic demand and low trade exposure (e.g., utility services)
we find that the use-side incidence can dominate leading to regressive distribution of regulatory costs.
Finally, we find that the common practice of vintage differentiation, whereby only new sources of
pollution are covered, can cause a significantly more regressive distribution of costs, all else equal.
Keywords: environmental regulation, incidence, general equilibrium
JEL Classification: Q52, C68
1	National Center for Environmental Economics, U.S. Environmental Protection Agency, 1200 Pennsylvania Avenue,
Washington, DC 20460, USA
2	The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of
the U.S. Environmental Protection Agency.
1

-------
1 Introduction
The costs and benefits of environmental regulation are not expected to be felt uniformly across the
population. Understanding the distribution of these impacts is important for determining whether a
rulemaking exacerbates or ameliorates preexisting disparities. In recent years, there has been a push to
provide decision makers and the public with better information as to the distribution of benefits from
reducing pollution exposure across income groups.3 However, assessments of how the costs are
distributed across income groups remain absent from federal analyses of environmental regulations in
the United States (Robinson et al., 2016).4
The economic incidence of regulation has two primary components: the "source side" associated with
changes in the payments to primary factors of production (capital, labor, land, natural resources) and thus
the sources of income; and the "use side" associated with increases in final good prices which affect the
use of income. In a partial equilibrium setting and under perfect competition, the share of costs borne by
consumers versus factors of production will depend upon the relative elasticities of supply and demand.
If demand is more inelastic than supply, then consumers will absorb a greater share of the costs than the
factors of production. Under the assumption that supply is perfectly elastic in the long-run, all regulatory
costs will be passed on to consumers in the form of higher prices. Based on this assumption many early
studies on the distribution of regulatory costs focused exclusively on the use side (e.g., Gianessi et al.,
1979; Robison, 1985; and Casler and Rafiqui, 1993). When focusing on the use side, environmental policies
that affect goods which are a larger share of the budget for low income households than for high income
households, such as energy, appear regressive (e.g., Burtraw et al., 2010; Hasset et al., 2009).
However, the source side can have an important role in determining both the short-run and long-run
incidence of regulatory costs (Fullerton and Heutel, 2007, 2010). Limited mobility of workers or capital
across space and sectors will cause those factors to bear some of the regulatory costs (Fullerton and
Muehlegger, 2019). Similarly, the presence of fixed factors, such as land and natural resources, will lead
to non-homothetic production where the fixed factors bear some of the regulatory costs (Bento and
Jacobsen, 2007). Imperfect substitution between imports and exports can also cause a share of the
3	For example, Executive Order 12898 on Federal Action to Address Environmental Justice in Minority Populations
and Low-Income Populations requires federal agencies to identify "disproportionately high and adverse human
health or environmental effects of their actions on minority and low-income populations."
4	Some regulatory analyses, such as the 2010 Portland cement National Emissions Standards for Hazardous Air
Pollutants (NESHAP) consider how costs are split between consumer and producer surplus, but they do not consider
how that affects the overall distribution across households. https://www.regulations.gov/document?D=EPA-HQ-
OAR-2002-0051-2042
2

-------
regulatory burden to fall on owners of capital (Gravelle and Smetters, 2006). Finally, behavioral response
affecting investment can impact the burden borne by owners of capital (Gravelle and Smetters, 2006).
Accounting for the source-side effects typically makes the distribution of regulatory costs appear less
regressive. First, returns to capital make up a greater proportion of overall income for high income
households and therefore, when capital bears some of the regulatory costs the distribution tends to be
more progressive (Rausch et al., 2010). Second, transfer payments fund a significant share of low income
households' consumption leaving them less exposed to costs distributed through the source side. The
common practice of, implicitly or explicitly, indexing transfer payments to inflation can add to the
progressivity of the source-side distribution (Fullerton et al., 2011, Cronin et al., 2017). Together, these
effects can cause costs passed through the source side to be progressively distributed, whereby costs as
a share of income are larger for higher income households (Fullerton et al., 2011; Blonz et al., 2012).
Several studies have used a general equilibrium (GE) framework to capture both the use and source-side
effects for market-based policies targeting carbon pollution from fossil-fuel combustion. When the source
side is considered in conjunction with the use side, the distribution of costs for first-best environmental
policies may be modestly progressive, even prior to redistributing potential revenue from emissions taxes
or auctioned allowances (Rausch et al., 2011). In other words, even for energy goods, where the use side
is notably regressive, the progressive distribution of costs on the source side may be strong enough to
dominate the overall incidence. However, for such market-based policies the complete design, including
how revenue is recycled (e.g., Caron et al., 2018; Rausch et al., 2010; Rausch et al., 2011), or if it is even
raised in the first place (e.g., Dinan and Rogers, 2002; Rose and Oladosu, 2002; and Parry, 2004), is critical
for determining the overall incidence. Second best policy designs, such as renewable energy standards or
cap-and-trade with incomplete coverage, may have cost distributions that are dominated by the use-side
effects and therefore, remain regressive even when accounting for the source side (Rausch and Mowers,
2014).
Research on the overall cost incidence of environmental policies (i.e., use and source-side effects) has
predominately focused on the distributional impacts of market-based policies, often with economy-wide
coverage. However, in practice market-based policies are rare and command-and-control policies, usually
affecting only a single sector, are the most common. The scant attention given to the incidence of
command-and-control regulations has been primarily focused only on the use side.5 Fullerton and Heutel
5 Fullerton (2008,2011), Bento (2013), and Robinson et al. (2016) provide excellent reviews of the previous literature
on the incidence of environmental regulations.
3

-------
(2010) use a stylized two-sector static GE model to demonstrate that source-side effects are also
important for determining the distribution of costs for command-and-control environmental policies,
such as performance standards and technology mandates. However, the strength of the source-side effect
for sector-specific technology mandates remains an open question, including its sensitivity to the affected
sector and the input composition of the abatement technology. Such information is critical for interpreting
the results of prior use-side studies of command-and-control regulations and for determining the
appropriate scope of future distributional analyses.
We use a detailed computable general equilibrium (CGE) model of the U.S. economy to study the
incidence of regulatory costs from single sector technology mandates and evaluate the strength of the
source-side effects in defining the distribution of costs. The application of a CGE framework allows us to
capture both the use and source-side effects of environmental regulations in a consistent manner. The
use of a CGE model also allows us to capture many of the conditions previously found to be important for
estimating source-side effects, including limited sectoral and spatial mobility of capital, fixed factors in
production, imperfect substitution between imports and exports, and indexed transfer payments. Similar
to previous studies, we find that the use-side incidence of technology mandates is notably regressive.
However, we find that the source-side effects are progressive on average and, in the scenarios we study,
dominate the overall incidence of costs from environmental regulations. We find that this result is robust
across the sector subjected to the regulation and the composition of inputs required by the abatement
process.
In addition, we examine the distributional implications of vintage differentiation, which is a common
feature of environmental regulations that can have important source-side effects. The practice of
exempting or setting less stringent standards for existing sources, relative to new facilities, has the effect
of generating scarcity rents that may be captured by the un- or less-regulated firms (Fullerton and Metcalf,
2001). The distribution of scarcity rents to owners of existing capital can result in a more regressive source-
side distribution of costs.6 Parry (2004) shows that even when the regulated goods represent a small share
of low income households' total consumption, policy designs, such as grandfathered permits, that transfer
scarcity rents to high income households can be highly regressive. For the case of technology mandates,
6 For example, in 1973 the U.S. Supreme Court decision in favor of the prevention of significant deterioration under
the National Ambient Air Quality Standards imposed potential entry costs on new firms even in attainment areas.
Despite the implied increase in regulatory stringency the ruling resulted in an increase in asset prices for regulated
firms with existing facilities signaling an increase in the value of existing capital due to the increased market entry
costs (Maloney and McCormick, 1982).
4

-------
we compare the case where all sources of production are affected to the case where only new sources
are regulated. We find that increased returns to owners of existing capital under vintage differentiated
regulations, can lead the incidence of costs to be significantly more regressive on average compared to
the case where all production sources face similar abatement costs.
The remainder of the paper is organized as follows. Section 2 presents details of the CGE models used,
along with the stylized regulations studied and the approach to quantitatively measuring the average
progressivity of regulatory costs. Section 3 presents the results of our regulatory simulations and Section
4 provides discussion.
2 Methods
The most common approach to estimating the social cost of a regulation in a general equilibrium setting
is a computable general equilibrium (CGE) model. CGE models assume that during a discrete period of
time an economy can be characterized by a set of conditions in which supply equals demand in all markets.
When a government policy, such as a tax or a regulation, alters conditions in one market, a general
equilibrium model determines a new set of relative prices for all markets that return the economy to
equilibrium. These relative prices determine changes in sector outputs, demand for factors of production,
intra-national and international trade, investment, and household consumption of goods, services, and
leisure (U.S. EPA, 2010). As such, a CGE model is able to capture the distribution of regulatory costs on
both the use and source side in an integrated framework. Section 2.1 describes the CGE model we use to
examine the cost incidence of regulation. Section 2.2 describes how we model command-and-control
environmental regulations and Section 2.3 introduces the approach we use to quantitatively measure the
average progressivity of the cost distribution.
2.1 Model
SAGE is an inter-temporal CGE model of the U.S. economy covering the period 2016 through 2061 and is
resolved at a subnational level.7,8 The model is similar to the class of calibrated CGE models regularly used
to analyze environmental and energy policies (e.g., Caron and Rausch, 2013; Chateau et al., 2014; Ross,
7	We use a recursive naming convention, where SAGE stands for SAGE is an Applied General Equilibrium (SAGE)
model.
8	In practice, the capital stock is not fixed and the behavioral response of savers to the regulation can affect the share
of the burden borne by capital owners making an intertemporal model important for assessing the source side
incidence (Gravelle and Smetters, 2006).
5

-------
2014). In this section, we provide a general description of the version of the SAGE model used in this
paper. Marten and Garbaccio (2018) provide detailed technical documentation of the model,

Pacific
Mountain
West South Central
East South Central
South Atlantic
West North Central
East North Central
Middle Atlantic
New England
Figure 1: SAGE Regions
The model represents the nine Census regions of the United States (Figure 1). Labor is assumed to be
immobile across regions, as is capital once it is installed; however, savings is mobile across regions. Trade
in goods follows an Armington specification, where goods are differentiated by their origin (Armington,
1969). For a given region, the model assumes differentiation between local goods, intra-national imports,
and international imports. Substitution possibilities across these sources are defined by a nested constant
elasticity of substitution (CES) function (Figure 2).
Armington
Composite
International
Imports
Domestic
Goods
se dn
Intra-national
Imports
Local
Goods
Intra-national International
Exports Exports
te dx
Regional
Output
Figure 2: Armington Trade Specification
6

-------
On the demand side, the first decision in defining the Armington composite is between consuming locally
produced goods and those imported from other regions within the United States (center-left of Figure 2).
Intra-national imports are assumed to be homogeneous with a single national market-clearing price. Next,
the local and national bundle is combined with international imports to form an aggregate Armington
composite good (top-left of Figure 2). Similarly, on the supply side regional output can be consumed
locally, exported intra-nationally, or exported internationally (bottom-right of Figure 2). The ability to
move regional output between markets is controlled by a constant elasticity of transformation (CET)
function (Figure 2). While the price of foreign exchange is endogenously determined, international
demand and supply are assumed to be perfectly elastic following the small open economy assumption.
Within each region, production is disaggregated into 23 sectors, with a focus on manufacturing and energy
as these sectors are the typical purview of environmental regulation at the federal level (Table 1). In most
sectors, production is assumed to be constant returns to scale where the production function is defined
by a nested CES function (Figure 3). Firms make decisions about the relative use of primary factors (i.e.,
capital and labor) and energy, and then the relative use of other intermediate material inputs compared
to the energy and value-added composite. The energy good is a composite of primary energy sources (i.e.,
coal, natural gas, and refined petroleum products) and electricity. It is assumed that firms determine the
relative use of primary energy sources followed by the relative use of primary fuels compared to
electricity. The sub-nest combining non-energy intermediate inputs is assumed to be Leontief.
Table 1: SAGE Sectors
Manufacturing
Energy
bom
Balance of manufacturing
col
Coal mining
cem
Cement, concrete, & lime manufacturing
cru
Crude oil extraction
chm
Chemical manufacturing
ele
Electric power
con
Construction
gas
Natural gas extraction & distribution
cpu
Electronics and technology
ref
Petroleum refineries
fbm
Food & beverage manufacturing


fmm
Fabricated metal product manufacturing
Other

pmm
Primary metal manufacturing
agf
Agriculture, forestry, fishing & hunting
prm
Plastics & rubber products
hit
Healthcare services
tem
Transportation equipment
min
Metal ore & nonmetallic mineral mining
wpm
Wood & paper product manufacturing
srv
Services


trn
Transportation


ttn
Truck transportation


wsu
Water, sewage, & other utilities
7

-------
Sectors associated with fixed factor inputs, such as land or natural resources, have a production structure
that deviates from the one presented in Figure 3. The presence of a fixed factor suggests that the
production function in those sectors should exhibit decreasing returns to scale to more accurately
represent the responsiveness of production to changes in relative prices. Therefore, in the resource
extraction sectors (col, gas, cru, and min) and the agriculture and forestry sector (agf) we include an
additional top-level nest which combines the fixed factor with the capital-labor-energy-materials (KLEM)
composite. The substitution elasticity between the fixed factor and KLEM composite is calibrated, so that
the price elasticity of supply in these sectors matches empirical estimates.
International
Intra-national
Local
Regional
Output
Materials
Value Added-Energy
Agriculture
Services
Energy
Value-Added
se-ene
International Domestic
Primary Energy
Electricity Labor Capital
Refined
Oil
Natural
Gas
Intra-national Local
Figure 3: General Production Structure
8

-------
Within each region, SAGE also models five representative households based on their pre-tax money
income level in the initial year of the model (Table 2).9 The income groups are selected to match current
U.S. income quintiles at a national level as closely as our underlying data source allows. Each
representative household is assumed to maximize inter-temporal per capita welfare subject to a budget
constraint and conditional on initial endowments of capital, fixed factor resources, and time. The inter-
temporal welfare function is an isoelastic utility function (i.e., constant relative risk aversion), while infra-
temporal preferences are modeled as a nested CES function (Figure 4).10
The nested structure of the intra-temporal utility function treats energy and materials in a similar fashion
to the standard production function. Households choose their relative consumption of primary energy
sources before selecting the ratio of primary energy to electricity. The energy bundle is then traded off
against non-transportation final consumption goods, a bundle that is then traded off against
transportation. At the top level of the intra-temporal utility function the ratio of consumption to leisure
is selected.
The inter-temporal connection between periods in the model occurs through the capital stock carried
over from one period to the next. The growth of the capital stock is a function of the depreciation rate
and endogenously determined investment. We assume a partial putty-clay specification for capital to
more appropriately represent the mobility of extant capital across sectors. Production associated with
existing capital at the start of the model's time horizon is modeled as Leontief based on the initial year's
cost shares, while production with new capital has the substitution possibilities afforded in the nested CES
structure presented in Figure 3. New capital stock is considered perfectly mobile across sectors, while
existing capital has limited and costly mobility as captured by a CET function that supplies extant capital
9	Note that money income includes cash based transfer payments, such as Social Security and the Supplemental
Nutrition Assistance Program, but does not include non-cash based transfer payments, such as Medicare and
Medicaid, which are included in consumption.
10	Households are differentiated based on income sources and consumption expenditures. However, the
substitution elasticities within the households' utility functions are consistent across the representative households.
Table 2: SAGE Households
Household Benchmark Year Income [2016$]
hhl
hh2
hh3
hh4
hh5
< $30,000
$30,000 - $50,000
$50,000 - $70,000
$70,000 -$150,000
> $150,000
9

-------
across sectors. The exception is any sector associated with a fixed factor, such as the resource extraction
or agriculture sectors. In those sectors, we do not model production from extant capital, and instead
directly calibrate the own-price supply elasticity to empirical estimates through the substitution elasticity
between the KLEM composite and the fixed factor.
Household
Utility
Consumption
Leisure
se_c
Other Consumption
se-cem
Non-Energy
se-cm
Transportation
se-cene
Agriculture
Services
Primary Energy Electricity
se-cen
International Domestic
Coal
Natural Refined
Gas
Oil
se-dn
Intra-national Local
Figure 4: Household Preferences
SAGE has a single government agent representing all jurisdictions. The government raises revenue
through ad valorem taxes on capital, labor, production, and consumption. Real government expenditures
are assumed to grow at the balanced growth rate, based on population and productivity growth. The
government balances its budget through lump sum transfers.
10

-------
In this paper, we extend the model of Marten and Garbaccio (2018) to account for changes in marginal
and average tax rates across households with different incomes. The SAGE model matches observations
of final good consumption, factor income, and personal taxes across households by using transfers from
the government to balance the household budget constraints in the benchmark social accounting matrix.
The transfer payments are indexed to the consumer price index (CPI) in the model, as is common in
practice (Fullerton et al., 2011). However, in the model developed by Marten and Garbaccio (2018) all
households in a region face the same effective marginal personal income tax rate for capital and labor
income. This means that government transfers to households include a recycling of tax revenue to make
up the difference between the value of personal income taxes collected at the marginal rate compared to
average rate. This is of concern for measuring incidence since a portion of household income that would
be valued with factor prices in reality, are instead being valued at final good prices (i.e., the CPI) in the
model. Therefore, we extend the model to better represent average personal income tax rates to improve
the model's representation of the share of household income due to transfers, while also improving the
representation of effective marginal tax rates across households.
In the model, we represent effective marginal personal income tax rates separately for capital and labor
income to better capture the average marginal taxes paid on an additional dollar of wage income versus
investment income for the population underlying the representative household. We also model separate
effective Federal Insurance Contribution Act (FICA) marginal tax rates by household to capture the
contribution limit on old age survivor insurance that reduces the marginal FICA tax rate for higher income
households. We estimate region and income specific earnings weighted average effective marginal tax
rates using the National Bureau of Economic Research's TAXSIM model (Feenberg and Coutts, 1993). We
generate a representative sample of regional tax units using the U.S. Census Bureau's Current Population
Survey (CPS) March Supplement and compute marginal FICA and labor income tax rates based on
perturbing wage income. To compute the regional effective marginal tax rates used in the model we take
a weighted average based on wage income and the CPS March Supplement person weights. To compute
the marginal tax rate on capital income we perturb long-term capital gains income and take a weighted
average based on investment income and the CPS March Supplement person weights. A detailed
description of the methodology is presented in Appendix B.
To match the observed average tax rate in the IMPLAN dataset we refund the difference between taxes
that would be paid according to the marginal tax rate and the observed benchmark tax payments and
index these transfers at factor prices instead of the CPI used for other transfers. The price used to index
11

-------
this tax refund is a weighted average of factor prices with weights based on benchmark year income
shares. This approach allows us to use marginal tax rates to model behavior, while ensuring that the after-
tax factor income share matches the observations in the benchmark dataset. A detailed description of the
methodology is presented in Appendix B.
There are three main types of inputs to the model: (1) the social accounting matrix describing the state of
the economy in the initial year; (2) substitution elasticities that define opportunities to move away from
the structure observed in the initial year; and (3) parameters defining the expected evolution of the
economy in the baseline. These inputs are described in more detail in Appendix A.
We solve the model as a mixed complementarity problem (MCP) following the approach of Mathiesen
(1985) and Rutherford (1995). The MCP approach represents the model as a series of zero-profit
conditions, market clearance conditions, budget constraints, household first-order conditions, and closure
rules. The problem is formulated in the General Algebraic Modeling System (GAMS).11 The MCP is solved
using the PATH solver (Ferris and Munson, 2000).
2.2 Modeling Regulations
While there are some notable exceptions, environmental regulations rarely rely on market-based
incentives in practice. Instead, it is common for environmental regulations to resemble an emissions rate
standard, specify the use of certain types of pollution control equipment, and/or require the alteration of
production processes. While modifying input use to reduce emissions is often incentivized by regulation,
the output channel does not aid facilities in meeting regulatory requirements. Thus, regulatory
requirements can often be interpreted as technology mandates that a sector use more inputs to produce
the same amount of output. Therefore, we follow Marten et al. (2018) and focus our analysis on the
additional inputs to production required for compliance.
Building on prior work, we model the additional inputs required to comply with environmental regulations
as productivity shocks in the regulated industry (e.g., Hazilla and Kopp, 1990; Pizer and Kopp, 2005; Pizer
et al., 2006). One potential pitfall of this approach is that the substitution possibilities across inputs to
pollution abatement activities are the same as across inputs to production activities in the regulated
sector. The alternative is to model a separate pollution abatement sector with unique substitution
elasticities. Since pollution abatement is not a well-defined activity within the national accounts, and there
is a dearth of available information regarding the inputs to abatement activities and how they respond to
11 GAMS Development Corporation. General Algebraic Modeling System (GAMS) Release 24.2.3. Washington, DC.
12

-------
changes in relative prices, we do not pursue this strategy. An advantage of this approach is that it is
possible to move away from a Hicks neutral shock to examine the potential GE impacts of regulations
requiring a different but more expensive composition of inputs (i.e., process changes).
In most cases, analysts engaged in a rulemaking process have an engineering-based cost estimate
available that indicates what additional inputs are required based on baseline levels of production valued
at baseline prices. Such an estimate can also be used to inform how to introduce a regulation into a CGE
model. Given the exploratory nature of our analysis, we don't have the luxury of detailed engineering
estimates. Therefore, we consider a range of potential input requirements for compliance activities. First,
we use the input requirements associated with past compliance activities for U.S. environmental
regulations. Nestor and Pasurka (U.S. EPA, 1995) established input values for pollution abatement
activities to comply with U.S. air pollution regulations. Since air pollution regulations make up a large
proportion of regulations, in terms of volume and costs, this provides a reasonable starting point.12
However, it has been shown that the results of CGE analyses of regulations can be sensitive to this
assumption (e.g., Nestor and Pasurka, 1995), so we also consider the case of a Hicks'-neutral abatement
requirements, along with capital- and labor-only cases.
The social cost of environmental regulation is measured using the equivalent variation (i.e., the maximum
amount of money a representative agent is willing to pay to forego the burden of the regulation). We
compute this household-specific value numerically as the difference between the present value of
baseline expenditures and those associated with the optimal path of consumption and leisure that would
lead to the same level of inter-temporal welfare as the regulatory case but with prices fixed at their
baseline values.13
2.3 Measuring the Average Progressivity of the Regulatory Cost Distribution
It is useful to have a quantitative measure of the average progressivity of the distribution of regulatory
costs for at least two reasons. First, if the distribution of costs as a share of income is not everywhere
either increasing or decreasing with income the distribution will not be unambiguously progressive or
regressive, respectively. Second, it is useful to have a single quantitative measure that describes the
12	Appendix C provides a mapping of the Nestor and Pasurka (1995a) cost shares to the commodities in our model.
13	The environmental regulations considered in this paper are relatively marginal changes, such that computing
household-specific willingness-to-pay as the change in full consumption (consumption plus leisure) evaluated at
benchmark prices produces the same results as using EV that would also take into account the curvature of the utility
function and therefore, the differences in baseline income levels across households.
13

-------
distribution of costs to facilitate comparisons across the regulatory scenarios studied. Therefore, we use
the Suits (1977) index, which is commonly used to measure the average progressivity of taxes to facilitate
comparison, and can easily be adapted to the cases of environmental regulations.14 Consider the Lorenz
curve mapping the cumulative distribution of income across households to the cumulative distribution of
social costs across households, as depicted in Figure 5. The linear curve OB represents that case where
the household costs are proportional to income. The convex curve OCB represents the case where the
regulatory costs are progressively distributed throughout the entire distribution - costs initially increase
slower than income but then increase faster than income at the top of the income distribution. The dashed
line represents the case where the costs are increasing slower than income for both the low- and high-
income households, but faster than income for middle-income households. In this case, which is common
in our results, middle-income households bear a disproportionate share of the regulatory costs relative to
the initial income distribution. For cases, such as the dashed line in Figure 5, the Suits index tests whether
the distribution of costs in these cases is progressive on average.
Specifically, the Suits index measures the average progressivity as
Y
5 = 1-- ;	(1)
where Z is the area of the triangle OAB and V is the area between the Lorenz curve and the axis OA. For
the solid curve OCB, Y is the area OABC between the Lorenz curve and the axis. If the regulatory costs
are proportional to income, Y = Z and S = 0. If the regulatory costs are distributed progressively on
average relative to income then S>0 as Y Z.
14 This approach is closely related to the suggestion of Robinson et al. (2016) to use Gini coefficients to quantitatively
study the distribution of regulatory costs.
14

-------
100
50-
"O
o
50
100
Accumulated Percent of Income
Figure 5: Lorenz Curve of Regulatory Costs
3 Results
We consider a suite of 168 illustrative regulatory scenarios, across all combinations of the twenty one
regulated sectors, four input assumptions for the compliance technology (Hick's neutral, capital only,
labor only, or Nestor and Pasurka shares), and two vintage differentiation assumptions (all sources and
new source only). In Section 3.1 we begin by examining the share of social costs borne by each income
quintile when all emission sources are affected. In Section 3.2 we consider the impact of these costs
relative to income and decompose the use and source-side effects. In Section 3.3 we consider the
distributional impact of vintage differentiation.
3.1 Share of Costs by Household Income
We begin by considering the distribution of social costs across income quintiles. The approximate
percentage of aggregate social costs borne by each income quintile are presented in Figure 6 for the
regulatory scenarios in which all sources are affected by the policy.15 As expected the level of costs is
15 The household definitions in Table 2 are the best approximation of income quintiles we can model due to
limitations in the underlying benchmark dataset. However, the share of households in each income bin is not equal
to exactly 20%. To approximate the share of costs borne by income quintiles we assume that per household costs
15

-------
increasing with income. Households in the bottom income quintile are expected to bear around 3% to
8% of the overall social costs, while the top income quintile is expected to bear around 35% to 53% of
the overall social costs. The distribution is relatively linear across the first four income quintiles, but
overall the distribution across income quintiles is convex.
c/j 35.
O
S 30'
o
h-
O 25'
"E
O 20
®
CL
Input Bias
-* Capital Only
» Labor Only
• Neutral
11 Nestor & Pasurka
Income Quintile
Figure 6: Approximate Percentage of Costs Borne by Income Quintiles
Note: The dispersion of the points along the x-axis within a household is only intended to
improve the readability of the figure. However, the dispersion is the same for each sector,
providing comparability within and across households.
Consumption increases with income and therefore, it is expected that higher income households will bear
a larger share of the costs. However, these shares are substantially different from those quintiles' overall
share of consumption. Under the assumption of full price pass through (i.e., no source-side effects) the
share of costs borne by each quintile would be consistent with their share of consumption. However, the
bottom income quintile consumes 11% of all goods and services compared to their 3% to 8% share of the
regulatory costs under the illustrative scenarios. While the top income quintile consumes 31% of all goods
experienced by each of the five representative households is representative of the per household costs borne by the
associated quintile and apply those estimates to the population in each quintile. The largest effect of this scaling is
on the share of costs in quintiles 4 and 5, as the income bracket for hh4 covers more than 20% of the population
while the income bracket for hh5 covers less than 20% of the population.
16

-------
and services compared to their 35% to 53% share of the regulatory costs. If one were to instead compare
the share of final goods consumption for the directly regulated to share of regulatory costs a similar
pattern emerges, with the lowest quintiles share of costs below their consumption share and the highest
quintiles share of costs typically above their consumption share.
The difference between households' consumption shares and their regulatory cost shares is due to the
presence of source-side effects. In the near- to medium-run, production exhibits decreasing returns to
scale due to rigidities on the supply side. For example, limited mobility of workers or capital across space
and sectors, which are partially captured in the model through the putty-clay representation of capital
and regional representation of labor markets. The upward sloping supply curves due to limited factor
mobility cause a portion of the regulatory costs to be passed through returns to primary factors instead
of final goods prices. Sectors, such as fossil fuel extraction and agriculture, that are reliant on fixed factors
of production will also see some costs passed on through lower returns to those fixed factors, for similar
reasons. Since high income households own a larger share of effective primary factor income they are
more susceptible to costs from the source side. Furthermore, a notable share of consumption for
households at the bottom of the distribution is funded through transfer payments, which are
predominantly (over 90%) indexed for inflation (Fullerton et al., 2011). Therefore, the purchasing power
for a substantial portion of those households' income is protected from increasing in final goods prices as
a result of the regulation. These source-side effects cause a larger share of the costs to be borne by higher
income households than would be estimated by the use side alone.
3.2 Relative Burden of Regulatory Costs
As is well known, the bottom income quintile accounts for notably less than 20% of aggregate income and
the top income quintile accounts for notably more than 20% of aggregate income. Specifically, the bottom
quintile receives 7% of overall income after taxes and transfers compared to 46% for the top quintile.16
Therefore, it is important to consider the distribution of regulatory costs in context of the existing income
16 We note that these values, based on the benchmark data in our model, are consistent recent U.S. Congressional
Budget Office estimates for shares of income after transfers and federal taxes by quintile (CBO, 2018). Slight
differences are because CBO only accounts for federal income taxes while our model also includes state income
taxes.
17

-------
Capital Only	Labor Only
100
in 40
t/>
o
O
.<5 20
o
o
w
o
c
0)
o
Neutral
Nestor & Pasurka
100
ro
3 80
E
o
o
60
40
20
80 100 0	20
Accumulated Percent of Income
20
o
40
60
100
Figure 7: Lorenz Curves for Regulatory Costs
Note: Each panel represents a different assumption about the input composition of the required
compliance activity. Within each panel the 21 curves represent the regulatory scenarios varying
the directly regulated sector.
distribution to better understand the distribution of burden. For our set of regulatory scenarios in which
all sources of pollution are covered, Figure 7 presents the Lorenz curves mapping the cumulative
distribution of income after transfers and taxes to the cumulative distribution of regulatory costs. If a
curve follows the 45-degree line that indicates that the costs are distributed proportionally to income. In
scenarios where the curve always remains below the 45-degree line, regulatory costs as a share of income
18

-------
are always increasing with income. For scenarios where the curve starts below but then crosses the 45-
degree line, costs as a share of income are higher for middle-income households than for low- and high-
income households.
In general, the distribution of regulatory costs remains fairly close to the solid 45-degree line representing
a neutral distribution (within the context of the existing income distribution). However, for most scenarios
the Lorenz curve starts out below but then crosses the 45-degree line. In these cases, the regulatory costs
are initially distributed progressively, but eventually costs as a share of income begin to fall as income
increases. In all but one of the 84 simulations the household costs as a share of benchmark income is
increasing with income for the first three quintiles. However, between the third and fourth income
quintiles, costs as a share of income increase in only 71% of the simulations and in only 14% of the
simulations do costs as a share of income increase between the fourth and fifth quintile. Therefore, only
a small share of the illustrative regulatory scenarios have a cost distribution that is progressive throughout
the entire income distribution. In most cases, the highest income quintile has costs as a share of income
that are also less than the third (86%) and second (74%) income quintiles. In 11% of the simulations the
highest income quintile has costs as a share of income that are lower than the lowest (first) income
quintile. Therefore, while the source-side effects increase the amount of regulatory costs borne by high
income households, their share of costs remains less than their share of overall income.
To more easily compare the incidence across the illustrative regulatory scenarios we turn to the average
progressivity of the cost distribution, as measured by the Suits index described in Section 2.3. The Suits
index for the scenarios in which all pollution sources in the regulated sector are subject to the regulation
is presented in Figure 8. A positive value for the index indicates a distribution of costs that is progressive
on average, with a higher index value indicating a more progressive distribution on average. A negative
index value indicates that the distribution of costs under that scenario are regressive on average. In nearly
all cases, the distribution of regulatory burden is slightly progressive on average, with a mean index of
0.01 and a standard deviation of 0.04.
19

-------
bom -
cam ¦
chm -
con ¦
cpu -
fbm -
fmm -
pmm -
prm -
q tem -
wpm ¦
col -
cru ¦
ele -
gas-
ref-
agf-
min -
trn -
ttn -
wsu -
-0.10	-0.
Figure 8: Suits Index
When examining the estimates of the Suits index three themes emerge. First, the distribution of
regulatory costs is more sensitive to the regulated sector than the input composition of the abatement
activity. Characteristics of the sector being regulated have an important influence on the strength and
direction of the use-side effects and in turn the overall cost distribution. The simulations that exhibit a
relatively more regressive distribution of costs are associated with sectors where a large share of output
is consumed as a final good and the per capita consumption of the commodity are relatively similar across
households (i.e., low elasticity of demand with respect to income). For example, water, sewage, and waste
services (wsu), food and beverage manufacturing (fbm), electricity (ele), and refined petroleum (ref). For
these commodities, final consumption scales less than proportionally with income leading the use-side
incidence to be fairly regressive. This effect works in the opposite direction as well. Regulations affecting
sectors where final good consumption is disproportionately concentrated with high income households
tend to be more progressive on average. For example, while gasoline consumption rises only modestly
with income leading regulations directly affecting the refined petroleum (ref) sector to be relatively
regressive, the fact that new cars are disproportionately purchases by higher income households causes
(U
O
Z3
(Q
m
(Q
o
0
Input Bias
Capital Only
Labor Only
Neutral
Nestor & Pasurka
0.00
Suits Index
0.05
0.10
20

-------
regulations directly affecting the transportation equipment manufacturing (tern) to have a relatively
progressive distribution of costs.17
Second, for most sectors, the distribution of costs is relatively similar across the different input
composition scenarios. The input composition affects the distribution of costs by influencing the degree
to which regulatory costs are passed through different factor prices. However, as will be demonstrated in
more detail in Section 3.2.1, we find that the source-side incidence is dominated by the role of the existing
capital stock. When all sources of pollution are affected an important avenue through which regulatory
costs are distributed is the returns to existing capital in the regulated sector. Due to its limited mobility,
regulations that lower the productivity of existing capital lead the owners of that capital stock to bear a
notable share of the incidence. This effect tends to dominate the source-side distribution of regulatory
costs leading the input composition of compliance to have a small effect. Exceptions are regulations in
sectors producing water, sewage, and waste services (wsu), electricity (ele), and refined petroleum (ref)
where the input composition of abatement can have a notable effect on the distribution of costs. In these
sectors, a significant share of output is associated with final good consumption, demand is relatively
inelastic, and foreign imports are small. In these situations, the owners of existing capital remain able to
pass on a significant share of the regulatory costs to consumers, such that impacts on the return to existing
capital are no longer the dominant effect on the source side.
Third, regulations that fall on sectors associated with sector-specific fixed factors (i.e., natural resources
or land) are more progressive in general, particularly regulations in the natural gas extraction (gas), crude
oil extraction (cru), and agriculture and forestry (agf) sectors. While all production exhibits decreasing
returns to scale in the short- to medium-run due to the limited adaptability of the existing capital stock,
in most sectors it is assumed to approach constant returns to scale in the long-run. However, sectors
associated with a sector-specific fixed factor continue to exhibit decreasing returns to scale. Therefore,
some of the regulatory costs will continue to be passed through in the form of lower returns on the fixed
factor, leading to more progressive and impactful source-side effects from regulations in these sectors.
The few Lorenz curves in Figure 7 that remain below the 45-degree line for the entire income distribution
are cases where the regulated sector is associated with a sector-specific fixed factor.
17 It is important to note our model does not include a detailed representation of the used car market, which can be
important for evaluating the incidence of regulations affecting the production of automobiles (Jacobsen, 2013).
21

-------
3.2.1 Use- and Source-Side Incidence
To better understand the role of the use and source side in determining the distribution of costs we
conduct two decomposition runs, similar to the approach of Rausch et al. (2011). To estimate the
distribution of costs through the use side we recalibrate the model so that the after-tax shares of income
from capital, labor, and resources are equal across households.18 When the income source shares are
recalibrated to be equivalent across households the distribution of costs will be determined by differences
in consumption shares, thereby identifying the incidence of the use side. To estimate the distribution of
costs through the source side we recalibrate the model so that the consumption shares are equal across
households.19 When the consumption shares are equal the distribution of costs relative to income will be
primarily determined by differences in income sources, thereby identifying the incidence on the source
side. Figure 9 presents the Suits index estimates for the two recalibrations. For ease of comparison the
hollow points represent the Suits index estimate from the default model, as presented in Figure 8.
Figure 9a is based on the recalibration in which after-tax factor income shares are set equal across
households, thereby highlighting the distribution of costs through the use side. As expected, the
distribution of costs on the use side is notably regressive. This result is more pronounced for scenarios
where the per capita consumption of the regulated sector's product is similar across households (e.g.,
water, sewer, and waste services (wsu) and electricity (ele)). Some of the costs are still passed on to factors
of production through the source side, but those costs will be proportional to income under the
recalibration. Therefore, the differentiation in the Suits index across input composition scenarios is
indicative of the impact that assumption has on the share of costs estimated to be passed through the
source side. Within a given sector's results, a Suits index closer to zero suggests a greater share of the
costs are passed through the source side for that input composition. For a capital-intensive abatement
technology, a greater share of the cost is distributed through the source side than for a labor-intensive
abatement technology, as indicated by the Suits index generally being closer to zero for capital intensive
technologies in Figure 9a. This is because of the increased mobility of labor relative to capital (which is
inclusive of extant capital).
18	The shares are recalibrated to be equal across households within a region, leaving some small variation in shares
across regions. Recalibrating shares to be equal across regions would require adjusting regional trade and production
patterns, which would complicate the interpretation of the results.
19	Ibid.
22

-------
-0.10
-0.05	0.00
Suits Index
0.05
m
-------
Figure 9b is based on the recalibration in which consumption shares are set equal across households,
thereby highlighting the distribution of costs through the source side. As expected, the distribution of
costs on the source side is more progressive than the use side, independent of the scenario. Across the
regulatory scenarios, the source-side Suits index estimates are roughly consistent with the estimates from
the default calibration presented in Figure 8. These results suggest that regulatory costs are largely passed
through to factor prices such that incidence estimates based on the assumption of full output-price pass
through will be significantly biased, even for sectors with high levels of final good consumption that is
shared almost equally across the income distribution. There are multiple reasons that source-side effects
will be important, including limited sectoral and spatial mobility of capital, fixed factors in production,
imperfect substitution between imports and exports, and indexed transfer payments.
The results of this decomposition also provide insight into the importance of the input composition of
compliance and the immobility of extant capital on the distribution of regulatory costs. As seen in Figure
9a, the share of costs passed on to consumers through the use side is relatively sensitive to the input
composition of compliance. However, the source-side distribution in Figure 9b is not notably affected by
the input composition. This suggests that while the input composition's effect on the share of costs passed
through output prices is large relative to the overall use-side effect, it is small relative to the source-side
effect, in most cases. This is because the source-side incidence is driven in large part by the role of extant
capital. The limited mobility of extant capital causes a majority of the source-side incidence to be passed
through lower returns to existing capital. Under the extreme assumption that extant capital is fully
malleable, the input composition has a greater effect on the overall incidence, however the source-side
effects still dominate the overall incidence (see Appendix D). Therefore, while the low mobility of extant
capital is a strong force in defining the incidence of regulatory costs in our central results, without this
assumption other source-side drivers remain stronger than those on the use side.
We note that in a few cases, the estimates of the Suits index under the case of equal consumption shares
are lower than the default estimates. This occurs for sectors where high income households have a
notably larger share of final good consumption for the sector (e.g., balance of manufacturing (bom),
transportation equipment manufacturing (tem), and transportation (trn), where the later includes air
travel). This is because recalibrating the consumption shares to be equal alters the incidence of policy
induced changes to transfer payments, which biases estimates of source-side Suits index downwards. In
the model, the consumer price index (CPI) used to index transfer payments is defined as the post-tax
consumption weighted national price index in the benchmark year. This is similar to the general approach
24
-------
used by the U.S. Bureau of Labor Statistics to create the U.S. CPI that is used to index many transfer
payments in practice.20 When a regulation is imposed on a sector where high income households have a
disproportionately high share of final consumption, the CPI will increase more than the unit price of
consumption for lower income households. As a result, under some regulatory scenarios the change in
transfer payments to low income households may more than compensate for any price changes they face
in the default model. When the model is recalibrated to set the consumption shares equal across
households such an outcome is no longer possible. Therefore, in the cases where higher income
households disproportionately consume the directly regulated sector's product the recalibrated model
with equal consumption shares will underestimate the average progressivity of the cost distribution on
the source side.
3.3 Vintage Differentiation and Cost Incidence
It is not uncommon for compliance obligations under environmental regulations to be differentiated by
the vintage of the affected source. One of the most common cases is that of new source standards, where
the regulation only applies to sources constructed after the regulation is promulgated (sometimes
referred to as the grandfathering of existing sources).21 Therefore, we consider the sensitivity of cost
distribution to vintage differentiation of the standards. We re-run all of the illustrative regulatory
scenarios varying the regulated sector and the input composition of the compliance activities but only
apply the regulation to production associated with new capital to approximate a new source standard.
The per unit output cost of the regulation is held constant at the level used in the prior experiments, but
only production associated with new capital is subject to the productivity shock associated with the
regulation. The Suits index values for the vintage differentiated regulations is presented in Figure 10. Note
that we do not plot results for the resource extraction sectors (col, cru, gas, min) and the agriculture and
forestry sector (agf) because the model does not differentiate between new and extant capital in those
sectors, as discussed in Section 2.1.
20	In practice, over time the BLS updates the basket of goods used to calculate the U.S. CPI though it always lags
changes in consumption patterns by a couple of years. If there is any substitution across consumption goods due to
a regulation this will be picked up through changes in the U.S. CPI.
21	In some cases, such as New Source Performance Standards promulgated under Section 111(b) of the Clean Air Act
new sources are defined as those constructed after the regulation is proposed instead of the final promulgation
date.
25
-------
bom
cem
chm
con
cpu
fbm
fmm
pmm
u wprri
cru
ele
mm
trn
-0.10
-0.05
0.00
Suits Index
0.05
m
0.10
Input Bias
Capital Only
Labor Only
Neutral
Nestor & Pasurka
Affected Sources
0 New Sources
O All Sources
Figure 10: Suits Index for New Source Regulations
Under the new source regulations, the estimates of the Suits index are generally lower than the case
where all sources face the same compliance costs, signaling that the distribution of costs for technology
mandates is generally more regressive for vintage differentiated regulations. When the productivity of
new sources falls, less new capital is deployed in the regulated sector and the real returns paid to existing
capital see upward pressure. This represents scarcity rents being created by the regulatory design and
captured by the owners of existing capital. Since returns on capital represent a significant portion of
income for the top quintiles, these scarcity rents mitigate some of the regulatory impact for high income
households resulting in a more regressive distribution of costs. In Appendix E, we replicate the use and
source-side decomposition of Section 3.2.1 for the case of the new source regulations. Both the use and
source-side distributions are more regressive for new source regulations than for the case where all
sources are regulated. This highlights how vintage differentiation increases the regulatory burden passed
through output prices and lowers the burden on owners of capital.
An exception are capital intensive abatement technologies, where in many cases, a new source standard
has a more progressive distribution of costs than under a regulation affecting all sources. The new source
regulations lower the real returns paid to new capital, relative to the baseline, due to the increased
production costs imposed by the regulation. This effect on the average returns paid to new capital is most
pronounced for regulations that are capital intensive, since the average marginal product of new capital
26
-------
is most effected under this scenario. In many of these cases, this decreases in the returns to new capital
dominate any scarcity rents captured by the owners of the existing capital stock leading to a more
progressive distribution of regulatory costs on average under a new source only standard. However, for
most of the scenarios in which abatement is not capital intensive, any reductions in the average returns
to new capital are dominated by the increased returns on the fixed existing capital stock. In sectors where
demand for domestically produced goods is highly inelastic (e.g., food and beverage manufacturing (fbm),
refined petroleum (ref), electricity (ele), and water, sewer, and waste (wsu)) there is ample opportunity
to pass abatement costs on through prices, leading to small source-side effects. In these cases, new source
regulations have a more regressive distribution of costs, even for capital intensive technologies.
For similar reasons, labor-intensive new source standards have a larger impact on the average returns
paid to labor. Therefore, under these scenarios, relative to the other scenarios, a larger share of the
source-side costs are passed through to labor, which is a relatively greater source of income for lower
income households. Combined with the scarcity rents generated by the owners of existing capital, this
causes the Suits index to be substantially lower under labor intensive new source standards than for the
case where all sources are affected by the regulation. Suggesting that the distribution of costs may be
significantly more regressive for a labor intensive new source standard than an equivalent sector-wide
standard, all else equal.
4 Discussion
We consider the general distribution of social costs for single-sector command-and-control environmental
regulations. We evaluate the incidence of costs for a suite of illustrative regulatory scenarios that vary
across the affected sector, compliance input requirements, and affected vintage of sources. We find that
for sector specific environmental regulations with relatively uniform effects across sources in a sector, the
costs generally increase with income independent of the regulated sector or input requirements for
compliance. In our regulatory scenarios, we found the share of social costs borne by low income
households to be less than their share of aggregate consumption and the share of social costs borne by
high income households to be greater than their share of aggregate consumption. Household costs are
not proportional to consumption because not all regulatory costs are passed on through final output
prices. This finding highlights the importance of accounting for source-side effects when evaluating the
distribution of regulatory costs for command-and-control policies in practice.
27
-------
The growth of regulatory costs when moving up the income distribution is estimated to be faster than
income growth over the first four quintiles in most of the regulatory scenarios considered. This results in
a distribution of regulatory burden that is progressive over the first four income quintiles. However, in
most (86%) of the scenarios the costs as a share of income are lower for the highest income quintile than
for at least one of the preceding quintiles. Notable exceptions are regulations in fossil fuel extraction
sectors where the source-side incidence is estimated to be strong enough to result in a progressive
distribution of costs across the entire income distribution due to a significant share of the costs being
passed on to owners of the affected natural resources.
Given that the distribution of regulatory costs relative to income is not increasing across the entire income
distribution, we use the Suits index to make quantitative comparisons regarding the average
progressivity/regressivity of the incidence of regulatory costs. Based on the Suits index, we find that, on
average, the distribution of regulatory costs is slightly progressive for around two thirds of the regulatory
scenarios considered. This is because we find that a substantial share of regulatory costs is passed on to
households through lower returns to primary factors of production, which disproportionately affect high-
income households. Meanwhile low-income households after-tax and transfer income is partially
protected from regulatory costs passed through on the use side due to a compensating effect of indexed
transfer payments on the source side. Together, these effects dominate the distribution of regulatory
costs. For the scenarios considered, we find distribution of regulatory costs through the source side is
substantially more important for determining the overall incidence of costs compared to the distribution
of regulatory costs through the use side. This result has important implications for the common empirical
approach to estimating distributional impacts that focuses primarily on the distribution of costs through
consumption prices and suggests that more attention should be paid to impacts of regulations on sources
of income.
However, we do find that in some cases the use-side incidence has a notable effect on the overall
distribution of costs. These cases include regulations affecting utilities that produce and distribute
electricity, waste, water, and sewage services, refined petroleum, and food and beverage manufacturing.
These are sectors with high shares of production being final consumption and where demand is both
highly inelastic and relatively similar across households leading to most of the costs being passed through
a regressive use side. When focusing on the use side we find the distribution of costs for regulations in
these sectors to be notably regressive, consistent with prior studies (e.g., Burtraw et al., 2010; Hasset et
28
-------
al2009). Though, we find that even for policies in these sectors the source-side effects are of first-order
importance for determining the overall distribution of regulatory costs.
A common characteristic of many federal environmental regulations is that they are only applicable to
new sources of emissions (e.g., grandfathering). Under this type of vintage differentiation, the regulatory
costs for new sources act as a new barrier to entry on which owners of existing sources can extract rents.
Since these rents are predominately captured by high-income households, in most cases vintage
differentiation leads to a notably more regressive distribution of costs relative to a regulation that affects
all sources equally. An exception is the case of capital-intensive abatement technologies, where the
scarcity rents are dominated by reduced returns to new capital that is now less productive under the
regulation. The regressive effect of vintage differentiation is greatest for regulations requiring labor-
intensive abatement activities where the source-side costs fall predominantly on labor, which is a
relatively greater source of income for middle-income households and does not offset the scarcity rents
captured by owners of existing sources.
It should be noted that, based on the illustrative nature of our investigation, we apply the simplifying
assumption that the cost of abatement is the same for new and existing sources. While this is the case for
some regulations, one stated motivation for vintage differentiation is that pollution abatement is cheaper
for new sources that can build their production facility with the environmental regulation in mind instead
of retrofitting an existing facility. For a regulation affecting all sources with a common stringency, this
would lead to a case where the costs are differentiated across capital vintage, with higher costs falling on
production associated with existing capital. The limited mobility of existing capital means that the effect
of the cost differential will likely be to increase the share of costs borne by capital owners, thereby
increasing the regressivity of a regulation affecting all sources with a common stringency above what we
estimate. If the use of vintage differentiation works to equate the abatement costs between sources the
incidence may be more in line with our results for the all source regulations.
In this paper, we focus on the vertical incidence of regulatory costs across income quintiles and the role
of the source side in determining that incidence. When interpreting our results, it is important to note
there is a great deal of heterogeneity within income quintiles. For example, a significant percentage of
the households in the lowest income quintile are in college or retired and would be expected to currently
have low levels of money income (Crain and Wilson, 2017). These households are combined in a single
quintile with the perennially poor even though lifetime wealth across those two groups may be
substantially different. Furthermore, household size and location are important for determining the
29
-------
lifestyle afforded to a household based on a given level of income. Future research examining whether
and how the source-side effects impact the horizontal incidence within quintiles is warranted. This could
be accomplished by further disaggregating the representative households within the quintiles (e.g.,
Rausch et al., 2011; Cronin et al., 2017) or through using different classifications of households, such as
on the basis of lifetime income or consumption rather than contemporaneous annual income (Fullerton
and Metcalf, 2002; Fullerton et al., 2011).
Our study is intended to be a broad look at the incidence of typical command-and-control environmental
regulations and therefore, some simplifying assumption were made that should be revisited in a detailed
policy analysis. First, we consider regulations imposed on relatively aggregate sectors of the economy.
Implicit in this assumption is that all commodities produced within an aggregate sector are perfect
compliments. In cases where a regulation only affects a segment of a sector and for which the sector also
produces close substitutes, such characteristics may have important implications for the distribution of
costs. The presence of close substitutes is likely to limit price increases in the regulated sector leading the
source-side incidence to have even greater dominance, but this is worth further investigation. Second, we
consider regulations whose cost of compliance is equal across space. In practice, environmental
regulations often have compliance costs that are differentiated across space. These regional differences
may have important implications for the incidence of costs.
Finally, in this paper we focus on the distribution of regulatory costs and the strength of source-side
effects in defining the incidence for technology mandates. However, it is important to note that a
complete analysis of the incidence of environmental regulations should consider both the distribution of
costs and benefits over the populations of interest. Some benefits may be separable in households' utility
functions, as we implicitly assume. Though, in some cases the beneficial impacts of pollution abatement
could affect equilibrium in the economy through multiple channels (Williams, 2002; Carbone and Smith,
2008; Marten and Newbold, 2017). The degree to which the beneficial impacts interact with regulatory
costs in equilibrium and affect the incidence of environmental policies remains an important question for
future research.
30
-------
References
Armington, P. 1969. "A Theory of Demand for Products Distinguished by Place of Production." Staff
Papers-International Monetary Fund: 159-178.
Arora, V. 2014. "Estimates of the Price Elasticities of Natural Gas Supply and Demand in the United
States." Technical Report. University Library of Munich, Germany.
Balistreri, E., and T. Rutherford. 2001. "Dynamic General Equilibrium Analysis at the State Level:
Assessing the Economic Implications of the Kyoto Protocol." Working paper. Printed in the proceedings
of the Fourth Annual Conference on Global Economic Analysis, June 27-29, Purdue University Vol. 2,
Session 1, Regional Economics I.
Beckman, J., T. Hertel, and W. Tyner. 2011. "Validating Energy-Oriented CGE Models." Energy Economics
33(5): 799-806.
Bento, A. 2013. "Equity Impacts of Environmental Policy." Annual Reviews Resource Economics 5: 181-
96.
Bento, A. and Jacobsen, M. 2007. "Ricardian Rents, Environmental Policy and the 'Double-Dividend'
Hypothesis." Journal of Environmental Economics and Management 53: 17-31.
Blonz, J., Burtraw, D., and Walls, M. 2012. "Social Safety Nets and U.S. Climate Policy Costs." Climate
Policy 12: 474-490.
Burtraw, D., Sweeny, R., and Walls, M. 2010. "The incidence of U.S. climate policy: alternative uses of
revenues from a cap-and-trade auction." National Tax Journal 62(3): 497-518.
Carbone, J. and Smith, V. 2008. "Evaluating policy interventions with general equilibrium externalities."
Journal of Public Economics 92(5-6): 1254-1274.
Caron, J. and Rausch, S. 2013. "A Global General Equilibrium Model with US State-Level Detail for Trade
and Environmental Policy Analysis-Technical Notes." Joint Program Technical Note TN #13.
Caron, J., Cole, J., Goettle, R., Onda, C., McFarland, J., and Woollacott, J. 2018. "Distributional
implications of a national co2 tax in the U.S. across income classes and regions: a multi-model
overview." Climate Change Economics 9(1): 1-32.
Casler, S. and Rafiqui, A. 1993. "Evaluating fuel tax equity: direct and indirect distributional effects."
National Tax Journal 46(2): 197-205.
CBO. 2017. International Comparisons of Corporate Income Tax Rates.
https://www.cbo.gov/publication/52419
31
-------
Chateau, J., Dellink, R. and Lanzi, E. 2014. "An Overview of the OECD ENV-Linkages Model: Version 3."
OECD Environment Working Papers No. 65.
Crain, V. and Taylor, W. 2017. "Use With Caution: Interpreting Consumer Expenditure Income Group
Data." Prices and Spending 6(5).
Cronin, J., Fullerton, D., and Sexton, S. 2017. "Vertical and horizontal redistributions from a carbon tax
and rebate." NBER Working Paper 23250.
Dinan, T. and Rogers, D. 2002. "Distributional Effects of Carbon Allowance Trading: How Government
Decisions Determine Winners and Losers." National Tax Journal 55(2): 199-221.
Feenberg, D. and Coutts, E. 1993. "An Introduction to the TAXSIM Model." Journal of Policy Analysis and
Management 12(1): 189-194.
Ferris, M. C., Munson, T. S. 2000. "Complementarity Problems in GAMS and the PTAH Solver." Journal of
Economic Dynamics and Control 24(2): 165-188.
Fullerton, D. 2008. "Distributional Effects of Environmental and Energy Policy: An Introduction." NBER
Working Paper 14241.
Fullerton, D. 2011. "Six Distributional Effects of Environmental Policy." Risk Analysis 31(6): 923-929.
Fullerton, D. and Heutel, G. 2007. "The general equilibrium incidence of environmental taxes." Journal of
Public Economics 91: 571-591.
Fullerton, D. and Heutel, G. 2010. "The General Equilibrium Incidence of Environmental Mandates."
American Economic Journal: Economic Policy 2(3): 64-89.
Fullerton, D., and G. Metcalf. 2001. "Environmental Controls, Scarcity Rents, and Pre-Existing
Distortions." Journal of Public Economics 80: 249-267.
Fullerton, D. and Metcalf G. 2002. "Tax Incidence." Handbook of Public Economics 4: 1787-1872.
Fullerton, D. and Muehlegger, E. 2019. "Who Bears the Economic Costs of Environmental Regulations?"
Review of Environmental Economics and Policy 13(1): 62-82.
Fullerton, D., Heutel, G., and Metcalf, G. 2011. "Does the Indexing of Government Transfers Make
Carbon Pricing Progressive?" American Journal of Agricultural Economics 94(2): 347-353.
Gianessi, L., Peskin, H. and Wolff, E. 1979. "The Distributional Effects of Uniform Air Pollution Policy in
the United States." The Quarterly Journal of Economics 93(2): 281-301.
Gravelle, J. and Smetters, K. 2006. "Does the open economy assumption really mean that labor bears
the burden of a capital income tax?" Advances in Economic Analysis and Policy 6(1): 1-44.
32
-------
Hasset, K., Mathur, A., and Metcalf, G. 2009. "The Incidence of a U.S. Carbon Tax: A Lifetime and
Regional Analysis." The Energy Journal 30(2): 155-177.
Hazilla, M. and Kopp, R. 1990. "Social Costs of Environmental Quality Regulations: A General Equilibrium
Analysis." Journal of Political Economy 98(4):853-873.
Hertel, T., R. McDougall, B. Narayanan, and A. Aguiar. 2008. "Behavioral Parameters." Global Trade,
Assistance, and Production: The GTAP 7 Data Base.
Jacobsen, M. 2013. "Evaluating US Fuel Economy Standards in a Model with Producer and Household
Heterogeneity." American Economic Journal: Economic Policy 5(2): 148-187.
Jorgenson, D., Goettle, R., Ho, M., and Wilcoxen, P. 2013. Double Dividend: Environmental Taxes and
Fiscal Reform in the United States. MIT Press, Cambridge, Massachusetts.
Koesler, S., and M. Schymura. 2015. "Substitution Elasticities in a Constant Elasticity of Substitution
Framework: Empirical Estimates Using Non-Linear Least Squares." Economic Systems Research 27(1):
101-121.
Marten, A., and R. Garbaccio. 2018. "The SAGE Applied General Equilibrium Model vl.0: Technical
Documentation." NCEE Working Paper 2018-05.
Marten, A., Richard, G., and Wolverton, A. 2018. "Exploring the General Equilibrium Costs of Sector-
Specific Environmental Regulations." NCEE Working Paper 2018-06.
Marten, A., and S. Newbold. 2017. "Economy-Wide Effects of Mortality Risk Reductions from
Environmental Policies." NCEE Working Paper 2017-03.
Mathiesen, L. 1985. "Computation of Economic Equilibrium by a Sequence of Linear Complementarity
Problems." Mathematical Programming Study 23: 144-162.
McClelland, R., and Mok, S. 2012. "A Review of Recent Research on Labor Supply Elasticities."
Congressional Budget Office Working Paper.
Murray, B. C., Keeler, A., and Thurman, W. N. 2005. "Tax Interaction Effects, Environmental Regulation,
and "Rule of Thumb" Adjustments to Social Cost." Environmental & Resource Economics 30: 73-92.
Nestor, D., and Pasurka, C. 1995. "Alternative specifications for environmental control costs in a general
equilibrium framework." Economics Letters 48(3-4): 273-280.
Parry, I. 2004. "Are emissions permits regressive?" Journal of Environmental Economics and
Management 47: 364-387.
Pizer, W., and Kopp, R. 2005. "Calculating the Costs of Environmental Regulation" in K-G. Maler and J.
Vincent (Eds.). The Handbook of Environmental Economics (pp. 1307-1351). Elsevier. Vol. 3.
33
-------
Pizer, W., Burtraw, D., Harrington, W., Newell, R., and Sanchirico, J. 2006. "Modeling Economy-Wide
versus Sectoral Climate Policies Using Combined Aggregate-Sectoral Policies." The Energy Journal 27(3):
135-168.
Rausch S., Metcalf, G., Reilly, J., and Paltsev, S. 2010. "Distributional Implications of Alternative U.S.
Greenhouse Gas Control Measures." The B.E. Journal of Economic Analysis and Policy 10(2): Article 1.
Rausch S., Metcalf, G. and, Reilly, J. 2011. "Distributional Impacts of Carbon Pricing: A General
Equilibrium Approach with Micro-Data for Households." Energy Economics 33: S20-S33.
Rausch, S. and Mowers, M. 2014. "Distributional and Efficiency Impacts of Clean and Renewable Energy
Standards for Electricity." Resource and Energy Economics 36: 556-585.
Robinson, L. Hammitt, J., and Zeckhauser, R. 2016. "Attention to Distribution in U.S. Regulatory
Analyses." Review of Environmental Economics and Policy 10(2): 308-328.
Robison, H. 1985. "Who Pays for Industrial Pollution Abatement?" The Review of Economics and
Statistics 67(4): 702-706.
Rose, A. and Oladosu, G. 2002. "Greenhouse Gas Reduction Policy in the United States: Identifying
Winners and Losers in an Expanded Permit Trading System." The Energy Journal 23(1): 1-18.
Ross, M. 2014, "Structure of the Dynamic Integrated Economy/Energy/Emissions Model: Computable
General Equilibrium Component, DIEM-CGE." Nicholas Institute for Environmental Policy Solutions
Working Paper 14-12.
Rutherford, T.F. 1995. "Extension of GAMS for Complementarity Problems Arising in Applied Economic
Analysis." Journal of Economic Dynamics and Control 19(8): 1299-1324.
Serletis, A., Timilsina, G., and Vasetsky, O. 2010. "Interfuel Substitution in the United States." Energy
Economics 32(3):737-745.
Suits, D. 1977. "Measurement of Tax Progressivity." American Economic Review 67(4): 747-752.
U.S. EPA. 1995. "The U.S. Environmental Protection Industry: A Proposed Framework for Assessment."
U.S. EPA. 2010. Guidelines for Preparing Economic Analyses. Report Number EPA 240-R-10-001.
Available at: http://yosemite.epa.gov/ee/epa/eerm.nsf/vwAN/EE-0568-50.pdf/$file/EE-0568-50.pdf
U.S. EPA. 2015. "Economy-Wide Modeling: Social Cost and Welfare White Paper." Available at:
https://vosemite.epa.gov/sab/sabproduct.nsf/0/07E67CF77B54734285257BB0004F87ED/$File/Social+C
ost+and+Welfare+White+Paper+9-22-15.pdf
Williams, R. 2002. "Environmental tax interactions when pollution affects health or productivity."
Journal of Environmental Economics and Management 44(2): 261-270.
34
-------
35
-------
Appendix
A Model Calibration
The social accounting matrix is built from the 2016 state level accounts in the IMPLAN dataset.22 The
IMPLAN dataset is extended in three ways. First, ad valorem taxes for labor and capital income are added
to the dataset (see Appendix B). Second, oil and gas extraction is disaggregated into separate sectors for
crude oil extraction and natural gas extraction using state level data on production and consumption by
sector from the U.S. Energy Information Administration and trade data from the U.S. Census Bureau.
Third, we use population estimates for each representative household by region from the U.S. Census
Bureau's Current Population Survey.
The substitution elasticities for the production functions and Armington trade specification are adopted
from recent empirical studies. The three KLEM substitution elasticities (se_klem, se_kle, and se_kl) are
adopted from Koesler and Schymura (2015), while the substitution elasticities for the energy bundle
(se_ene and se_en) are adopted from Serletis, et al. (2010). The Armington elasticities between the local-
intra-national composite and intra-national imports (se_nf) are adopted from Hertel et al. (2008). To
calibrate the Armington elasticity between local and intra-national imports (se_dn) and the
transformation elasticity between output destinations (te_dx) we follow Caron and Rausch (2013). The
price elasticities of supply used to calibrate the substitution between the KLEM composite and fixed
factors in resource extraction and agriculture sectors (se_rklem) are adopted from additional sources. For
natural gas extraction, crude oil extraction, and coal mining we follow Arora (2014), Beckman et al. (2011)
and Balistreri and Rutherford (2001), respectively. For agriculture and forestry, we follow the Hertel et al.
(2002). In the intra-temporal utility function the substitution elasticity between consumption and leisure
(se_cl), along with the benchmark time endowment, are calibrated to match the midpoint of the ranges
for the compensated and uncompensated labor supply elasticities in the review of McClelland and Mok
(2012).23 We adopt the substitution elasticities in the intra-temporal utility function's energy bundle
(se_cene, se_cen) from Serletis et al. (2010). The remaining substitution elasticities in the intra-temporal
utility function (se_c, se_cm, and se_cem) are adopted from Caron and Rausch (2013), who use the same
nested CES specification. The inter-temporal substitution elasticity of full consumption is adopted from
22	IMPLAN Group, LLC, 16740 Birkdale Commons Parkway, Suite 206, Huntersville, NC 28078; www.IMPLAN.com .
23	The calibrated compensated labor supply elasticity is 0.2 and the calibrated uncompensated labor supply elasticity
is 0.5 based on the midpoints in McClelland and Mok (2012).
36
-------
Goulder and Hafstead (2018). Additional details and specific parameter values are presented in Marten
and Garbaccio (2018).
The exogenous parameters defining expectations about the growth and structure of the economy in the
baseline are derived from U.S. Energy Information Administration's 2018 Annual Energy Outlook (AEO).
Economic growth is driven primarily by population growth and Harrod neutral (i.e., labor embodied)
productivity growth. Both of these parameters are set to the average growth rates over the time horizon
of the most recent AEO. Energy intensity improvements are assumed to be capital embodied and
calibrated by shifting the future cost shares in the nested CES production functions to match the sector
specific average growth rates of energy intensity of production reported in the most recent AEO.
Consumption shares in the intra-temporal utility function are similarly shifted away from energy goods to
approximate the average reduction in the share of real consumption expenditures on specific energy
types as reported in AEO. Finally, the share of coal in electricity production is shifted towards capital and
labor, to match the shift from coal fired generation to renewables in AEO (noting that the share of
electricity generation from natural gas is expected to remain relatively constant in AEO thereby not
requiring additional calibration).
B Specification of Marginal and Average Personal Income Tax Rates
The model we use is the same as detailed in Marten and Garbaccio (2018), expect for improvements in
the representation of taxes to more accurately represent the share of income index by labor and capital
prices and an update to more recent estimates of state and local sales tax rates. In this section we describe
the tax structure of the updated model. The model explicitly includes consumption, tCri personal labor
income tax, tlr h , personal capital income tax, trr h, other business taxes/subsidies, tyr s, and corporate
income tax, tkr h . Production taxes net of any subsidies, tyr s, are based on the average rate observed in
the IMPLAN database. The corporate income tax, tkr h , is assumed to be constant across the U.S. and is
based on an assessment of the average effective marginal corporate income tax rate by the U.S.
Congressional Budget Office (CBO, 2017). Consumption taxes are based on estimates of the combined
local and state consumption tax rates from the Tax Foundation.24 The tax rates on corporate income and
consumption are presented in Table 3.
24 https://taxfoundation.org/state-and-local-sales-tax-rates-2018/
37
-------
Table 3: Tax Rates on Corporate Income and Consumption
Region ^r,h	^Cr
nen
0.19
0.06
mat
0.19
0.07
enc
0.19
0.07
wnc
0.19
0.08
sat
0.19
0.07
esc
0.19
0.08
wsc
0.19
0.09
mnt
0.19
0.07
pac
0.19
0.08
Personal income taxes on labor and capital incomes are differentiated across regions and households. We
represent effective marginal personal income tax rates separately for capital and labor income to better
capture the average marginal taxes paid on an additional dollar of wage income versus investment income
for the population underlying the representative household. Effective marginal Federal Insurance
Contribution Act (FICA) taxes are also differentiated across regions and households. This allows the payroll
tax rates to capture the annual limit on Old Age and Survivor's Insurance (OASI) taxes, which would not
be possible if the payroll taxes were collected on the firm side based on the model's structure. Data from
the U.S. Census Bureau's Current Population Survey (CPS) Annual Social and Economic Supplement (ASEC)
is used to create a representative sample of tax returns. These sample returns are then run through NBER's
Taxsim model version 27 to estimate marginal tax rates for wage income, FICA, and long-term capital gains
income for each sample return (Feenberg and Coutts, 1993).25 For each region and household we compute
the weighted average effective marginal tax rate from the sample returns by weighting the Taxsim results
by the CPS ASEC earned income and applying the supplement weights.
From the CPS, the filing status variable (filestat) and the dependent status variable (dep_stat) are used to
distinguish between single/head of household taxpayers and dependent taxpayers. All married taxpayers
are assumed to file jointly, and the person records for each couple are identified using the a_spouse
variable. The dep_row variable in the CPS is used to assign non-filing dependents to taxpayers, along with
the ages of the dependents. This information is used to populate the Taxsim variables used to assess
personal exemptions, the Dependent Care Credit, the Child Credit, and the Earned Income Tax Credit.
25 http://users.nber.org/~taxsim/taxsim27/
38
-------
The income variables in the CPS ASEC are mapped to the Taxsim variables as described in Table 4. For
married couples, all income values entered into Taxsim are the joint earnings, except in the case of wage
and salary income, which are kept separate. The CPS no longer included imputed capital gains and
therefore, they are omitted from the submission to Taxsim. This limitation may bias the weighted average
effective marginal tax rates downwards for the household representing the top income quintile (where
nearly all capital gains accrue) if the inclusion would cause some household to be in a higher tax bracket.
Table 4: CPS to NBER Taxsim Income Mapping
Taxsim


Variable
Description
CPS Variable(s)26
pwage
Wage and salary income of primary taxpayer
ws_val, semp_val, frse_val
swage
Wage and salary income of spouse
ws_val, semp_val, frse_val
dividends
Qualified dividend income
div_val
stcg
Short term capital gains or losses
NA27
Itcg
Long term capital gains or losses
NA28
otherprop
Other property income
rnt_val
nonprop
Other non-property income
oi_val, ed_val
pensions
Taxable pensions and IRA distributions
rtm_val
gssi
Gross social security benefits
ss_val, ssi_val, srvs_val, dsab_val
ui
Unemployment compensation
uc_val
transfers
Other non-taxable transfer income
paw_val, wc_val, vet_val, csp_val, fin_val
The implicit deductions for each filer are computed as the difference between adjusted gross income (agi)
and taxable income (tax_inc) as reported in the CPS minus personal exemption deductions accounting for
the phase out. From this value we subtract property and state taxes and submit the higher of this value
or zero to Taxsim as potential sources of itemized deductions. Property taxes in the CPS ASEC (prop_tax)
are associated with household records so we divide those taxes equally amongst all tax filing units in a
household.
For each representative filer, Taxsim returns the effective marginal tax rate for primary earner wage
income. Using the CPS ASEC person weights and primary earner wages, a weighted average of the
effective marginal tax rates for wage income are computed for each region and representative household
26	Except for the primary and spouse wage and salary income each Taxsim variable is the sum of the CPS variables
for both the primary taxpayer and their spouse for married taxpayers.
27	The CPS ASEC does not include information on imputed capital gains after 2010.
28	Ibid.
39
-------
in the model. A similar exercise is conducted for long-term capital gains taxes, though the weighted
average uses dividend and interest income as a proxy for capital income.
The personal labor income tax rates by region and household are presented in Figure 11, the FICA tax
rates are presented in Figure 12, and the personal capital tax rates bare presented in Figure 13. The
crossbars represent the national income weighted average effective marginal tax rate, for the specific
income category.
hh1	hh2	hh3	hh4	hh5
Household
Figure 11: Labor Income Effective Marginal Tax Rate by Household and Region
40
-------
0.14
<0
O
I 0.12
-------
personal income tax payment is split between labor and capital sources based on their share of pre-tax
factor income, such that
tl-refundt,h=tlr J0 ,
1 —
max (income _ tax 0r h, o)
tiJOr* +trrA re_exOrh+reOrh +^reseOrsh
(2)
and
tr _ refund, r h = trrjl re _ exOr h + re0r h + £ reseOr s h
1 —
max (income _ tax 0r h,0)
+K,J re_exOrih+reOrih +^rese0r^h
(3)
where 10r h is benchmark year labor income, re_exOr h is benchmark year income from extant capital,
reOrh is benchmark year income from new capital, and reseOrsh is income from fixed factor resources
employed in sector 5. The transfers are assumed to grow at the balanced growth rate and The transfer
tl_refundt r h is index at the wage rate plt r and the transfer tr _refundt r h is indexed by an index of
capital prices
Prt,rr0r,h +Pr_ex_aggtrkh_ex0rh +^prestr5res/?Or5
prh_indtrh =	
rOrh+kh_exOrh +^reshOr5
where pr_ex_Oggt r is the return to extant capital, prt r is the return to new capital, and prestrs is the
return to fixed factors used in sector 5.
42
-------
C Regulation Input Bias Specification
Table 5: Alternative Input Shares for Abatement Technology
Input
Nestor and
Pasurka
Capital Only
Labor Only
agf
-
-
-
cru
-
-
-
col
-
-
-
min
-
-
-
ele
0.270
-
-
gas
-
-
-
wsu
-
-
-
con
0.060
-
-
fbm
-
-
-
wpm
0.010
-
-
ref
0.010
-
-
chm
0.010
-
-
prm
0.025
-
-
cem
0.025
-
-
pmm
-
-
-
fmm
-
-
-
cpu
0.006
-
-
tern
0.001
-
-
bom
0.003
-
-
trn
0.010
-
-
ttn
0.010
-
-
srv
0.200
-
-
hit
1
0.160
--
1.000
K
0.200
1.000
-
43
-------
D Sensitivity to Putty-Clay Capital Specification
Figure 14 presents the use and source-side decompositions for the case were existing capital is fully
malleable (i.e., no modeling of partial putty-clay capital). The hollow points in the figure are denoted as
"Default" and in this case refer to the Suits index for the overall distribution of regulatory costs when the
model is run without the partial putty clay structure. These results pertain to the scenarios in which all
sources of production are subject to the regulation.
In general, the results are similar to the case where extant capital is assumed to have very limited mobility
(Figure 9). The source-side effects still dominate the overall distribution of costs as may be seen from the
similarity between the source-side sensitivity in Figure 14b and the hollow points representing the Suits
index for the overall incidence. The main difference is that the presence of extant capital with very limited
mobility is no longer the dominant factor in determining the share of regulatory costs that are passed on
through factor prices. Under the extreme assumption that all capital is malleable, even in the near term,
the input composition of compliance activities has a greater impact on the share of costs passed on
through the source side. However, the source-side effects still dominate the shape of the overall
distribution.
44
-------
4X10
0.05
Suits Index
(a) Equal After-Tax Factor Shares



••




Q)


3


C
/ -•

qT


o
- .

c


—%


Z5'


(O


m


3
/ •••

CD





(Q
"Y •;

<<


o





-------
E Use- and Source-Side Decomposition for New Source Regulations
m
Input Bias
Capital Only
Labor Only
Neutral
Nestor & Pasurka
Version
Equal
0 Factor
Shares
O Default
-0.1	0.0
Suits Index
0.1
(a) Equal After-Tax Factor Shares
01
c
5T
o
Input Bias
Capital Only
Labor Only
Neutral
Nestor & Pasurka
-0.1
0.0
Suits Index
m
CD
0,1
Version
Equal
9 Consumption
Shares
O Default
(b) Equal Consumption Shares
Figure 15: Use- and Source-Side Suits Index for New Source Regulations
46
-------