Access to this full-text is provided by IOP Publishing.
Content available from Environmental Research Letters
This content is subject to copyright. Terms and conditions apply.
Environ. Res. Lett. 17 (2022) 064011 https://doi.org/10.1088/1748-9326/ac6cfa
OPEN ACCESS
RECEIVED
25 October 2021
REVISED
14 April 2022
ACC EPT ED FOR PUB LICATI ON
5 May 2022
PUBLISHED
19 May 2022
Original content from
this work may be used
under the terms of the
Creative Commons
Attribution 4.0 licence.
Any further distribution
of this work must
maintain attribution to
the author(s) and the title
of the work, journal
citation and DOI.
LETTER
Distributional health impacts of electricity imports in the United
States
Eleanor M Hennessy∗, Jacques A de Chalendar, Sally M Bensonand Ineˆs M L Azevedo
Energy Resources Engineering, Stanford University, Stanford, CA, United States of America
∗Author to whom any correspondence should be addressed.
E-mail: emh@stanford.edu
Keywords: electricity imports, particulate matter, air pollution, carbon emissions, distributional effects, environmental justice
Supplementary material for this article is available online
Abstract
The electric grid is evolving rapidly in response to climate change. As renewables are incorporated,
more interconnection of the grid is expected. Exposure to fine particulate matter (PM2.5) from
fossil-fuel generation causes adverse health impacts, including thousands of premature deaths each
year in the United States. It is well understood that PM2.5 exposure can occur at great distances
from pollutant sources, but insufficient work has been done to understand the role of grid
interconnection and trade in causing pollution-related mortality. Regions with clean generation
can import electricity from regions with highly polluting generation sources, allowing them to
benefit from the electricity consumption while people in other regions suffer the associated health
damages. We use flow tracing and consumption-based accounting to characterize the health
damages from exposure to PM2.5 from electricity imports. We find that 8% of our estimated
premature deaths from electricity consumption in the United States are due to electricity imports.
There is large geographic heterogeneity, with the most impacts occurring in the Midwest. While
the West Coast has much cleaner generation and lower impacts overall, in many West Coast
Balancing Areas, more than 50% of the estimated premature mortality associated with electricity
consumption is caused by electricity imports, with some groups experiencing larger impacts than
others.
1. Introduction
The United States electricity grid is a highly intercon-
nected system. The grid is divided into six regions
that are overseen by the North American Electric
Reliability Corporation (NERC) to ensure reliabil-
ity and security of the grid. Within each NERC
region, Independent System Operators (ISOs) or
Regional Transmission Organizations control the grid
and electricity markets in specific geographical areas.
Balancing areas (BAs) are smaller areas controlled
by balancing authorities, who balance power within
their region. Electricity is exchanged between BAs
to maintain balance between supply and demand.
While some BAs generate the majority of the power
they use, others import a large portion. Some small
BAs have very little of their own generation and
rely nearly entirely on imports. In 2019, of the 66
BAs included in this study, 56 imported at least
10% of their electricity, and ten imported more
than 75%. In the SI, section 2.1 (available online at
stacks.iop.org/ERL/17/064011/mmedia), we provide
a table on annual net imports for each BA in 2019.
Climate mitigation plans and decreasing invest-
ment costs have spurred a wave of renewable energy
development, and as renewable energy penetra-
tion increases, electricity exchanges are expected to
increase to enable geographic aggregation of diverse
resources [1,2]. While electricity exchanges between
regions may be beneficial for carbon reduction,
embodied emissions of SO2, NOx, and PM2.5 asso-
ciated with them impact populations living far from
the populations consuming the electricity. The trans-
port of air pollutants is fairly well understood, but the
link to consumption of electricity and the associated
impacts of premature mortality that arise from elec-
tricity generation from fossil fuel sources requires fur-
ther exploration. In this work we use flow tracing to
© 2022 The Author(s). Published by IOP Publishing Ltd
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
do consumption-based accounting of health damages
caused by air pollution from electricity consumption
and to assess the magnitude and distribution of air
pollution-related premature mortality due to electri-
city imports for each US region.
Air pollution exposure is a major public health
risk. Each year, millions of premature deaths are
caused by poor ambient air quality [3]. A pollut-
ant of primary concern is fine particulate matter
(PM2.5), which causes 90% of air pollution-related
deaths [4]. In the US, exposure to PM2.5 is respons-
ible for thousands of deaths annually [5]. PM2.5 is
especially dangerous due to its small size. Consist-
ing of particles with diameters 2.5 µm or less, it can
penetrate deep into the lungs, damaging respirat-
ory and cardiovascular systems, and causing ailments
including ischemic heart disease, lung cancer, and
strokes [6,7].
PM2.5 is emitted directly, in the form of black car-
bon, and is formed from precursors including nitro-
gen oxides and sulfur oxides, which interact with
other molecules, such as ammonia in the atmosphere
to form secondary PM2.5[8]. There are many sources
of PM2.5, one of which is electricity generation with
fossil fuels [8]. Fossil fuel power plants (especially
coal plants) emit SO2, NOxand primary PM2.5. Their
stacks loft emissions high into the atmosphere, which
contributes to long-range transport of the pollutants.
As a result, emissions of PM2.5 from electricity gen-
erating units cause health impacts both near a plant,
and far from it [9].
Thind et al track net transfer of air pollution mor-
tality from electricity generation in each state, finding
that in 39 states, more than 50% of mortality associ-
ated with air pollution from the power sector is due to
emissions in another state [10]. Similarly, Dedoussi
et al show that between 41% and 53% of prema-
ture mortality caused by exposure to PM2.5 from all
sources is due to emissions generated in a different
state [5]. In the power sector, more than 70% of the
estimated premature mortality is caused by out of
state emissions [5].
In addition to the burden due to transport of
pollutants themselves, there is also a burden associ-
ated with the transfer of electricity between differ-
ent locations. De Chalendar et al [11] track embodied
emissions in electricity transfers between BAs in the
United States, demonstrating that some BAs are net
importers of pollution, while some are net exporters.
However, de Chalendar et al do not track the physical
transport and exposure pathways of pollutants [11].
In this work, we use the same underlying data as
de Chalendar et al [11] to track emissions associated
with electricity imports and the transport of those
emissions. We use consumption-based accounting to
identify where premature mortality occurs due to
emissions associated with electricity imports. Tes-
sum et al consider a similar approach in looking
at emissions associated with consumption of goods,
finding a mismatch between socioeconomic groups
that consume the most goods and those exposed
to the most pollution [12]. To our knowledge, we
are the first to apply consumption-based accounting
specifically to electricity imports while accounting for
the transport and ultimate exposure to pollutants.
This is of key relevance to understanding the implic-
ations of future climate mitigation and air pollution
policy interactions for the power sector as regional
electric grids in the United States become more inter-
connected and dependent on each other.
We use a source receptor matrix called ISRM [13]
created by Goodkind et al based on InMAP [14],
a reduced complexity air quality model (RCM), to
determine the location of premature mortality associ-
ated with imports and within-BA generation in each
BA in the Contiguous US. InMAP was developed by
Tessum et al [14], and has since been updated and
compared with other air quality models [15]. In the
methods and materials section, we explain in more
detail how we use InMAP and its underlying data. A
description of InMAP and the ISRM can be found in
SI section 1.2. We perform a BA-level, county-level
and census block group-level analysis to identify areas
that are disproportionately impacted, and to compare
the public health burden caused by electricity imports
to each BA.
This work provides insights on the consequences
of increasing interconnections between different
regions. As climate change mitigation goals rely on
increasing renewable energy sources, there have been
pushes to increase the interconnection between dif-
ferent areas of the grid. One example is the Califor-
nia ISO’s Energy Imbalance Market, which allows
for real-time trading between BAs in the Western
Interconnect [16]. Another is the suggestion to fully
integrate the Western Grid, though this has been
unsuccessful thus far. While studies have shown that
increasing grid interconnection allows for greater
integration of renewables [17], the impact on overall
emissions of air pollutants and their resulting health
damages has not been assessed. We find there is great
geographic heterogeneity in the impacts of electricity
imports on both carbon emissions and air pollution
mortality. This indicates there may be a need for
local policies to account for these differences. Over-
all, we find that imports are responsible for roughly
8% of the estimated premature mortality related to
electricity consumption. This figure may grow if
BAs expand their connections. Questions must be
addressed about the implications of responsibility for
premature deaths caused by consumption in differ-
ent BAs. At the very least, citizens deserve to have the
information made available to them, so they under-
stand the health consequences of electricity exports
taking place from generating stations near them. This
is particularly true for coal-fired generating stations
2
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
Figure 1. Conceptual flows of electricity and health damages. In this simplified example, BA 1 is a producer and consumer of
electricity, while BA 2 is a producer of electricity. BA 1 generates some of the electricity it consumes (A: ‘self-generation’), and
imports the electricity produced in BA 2 (B: ‘imports’). BA 1 causes health damages to itself through emissions from its own
electricity production (C), and from pollutants that are transported from the emissions caused in BA 2 by the electricity imported
to BA 1 (G). Electricity consumption in BA 1 also causes health damages in BA 2 due to local emissions in BA 2 from production
of electricity exported to BA 1 (F), and through transport of pollutants emitted in BA 1 (D). Electricity production in both BA 1
and BA 2 caused by consumption in BA 1 also causes health damages outside of the two BAs (E) and (H). Total damage caused by
self-generation in BA 1 is equal to C +D+E. Total damage caused by imports to BA 1 is equal to F +G+H. Note that while BA
2 is assumed not to consume electricity for simplicity, this is unlikely in real BAs, most of which would consume some of the
electricity they produce, which would result in additional flows of health damages associated with self-generation.
which have much higher air pollution emissions than
natural gas plants.
In figure 1, we show an illustrative example with
two regions, identifying flows associated with electri-
city production and consumption, and sources and
receptors of pollution.
2. Methods and materials
To assess the significance of electricity imports on
human health due to air pollutant emissions, we
analyze exposure to fine particulate matter formed
from emissions related to electricity consumption
in each BA in the contiguous United States. While
many policy decisions are made at the state and local
level, BAs correspond to the physical structure of the
electricity grid and the electricity exchanges that are
occurring, and thus they are a more natural choice
for studying electricity exchanges. However, we note
that our analysis of the air pollution impacts from
electricity consumption is conducted at the county
and block group level. Our approach consists of three
steps: (a) identifying where electricity is consumed;
(b) tracing flows of electricity to determine where
electricity was generated; and (c) tracking formation
of and exposure to PM2.5 leading to health damages
caused by emissions at the generation source. We use
the following publicly available data sources in our
analysis.
•Hourly electricity transfer data from EIA-930 con-
tains data on electricity generation and demand
in each BA, and electricity imports and exports
between each BA in 2019 [18].
•The Environmental Protection Agency (EPA)’s
Continuous Emissions Monitoring System data
provides hourly emissions of SO2, NOx, and CO2
from power plants in the United States [19].
•The National Emissions Inventory provides annual
data on primary PM2.5, NH3, and Volatile Organic
Compounds (VOC) emissions [20].
•The Energy Information Administration (EIA)
form 861 plant data provides a database of the
power plants used within each BA [21].
In our analysis we follow a similar approach to
de Chalendar et al [11] to assess the impacts of elec-
tricity consumption in each BA separately. First, we
determine the fraction of electricity inputs (genera-
tion and imports) consumed within the BA, using the
relationship described in equation (1), where Drep-
resents total demand in the BA, Xis the sum of all
3
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
exports from the BA, and fBA is the fraction of elec-
tricity consumed within the BA:
fBA =D
D+X.(1)
For each BA, we split the total electricity demand in
each hour into imports, and ‘self-generation’, or elec-
tricity generated within the BA that is used within
the BA. We assume that the fraction of electricity
used within the BA is the same for both imports and
self-generation. The total demand in each BA is met
by a combination of self-generation and imports, as
described in equation (2), where Irepresents electri-
city imports from each other BA, and Grepresents
generation. Self-generation, then is G×fBA:
D=(∑I+G)×fBA.(2)
We split our analysis into two parts, separately mod-
eling the emissions of CO2, SO2, and NOxassociated
with self-generation and imports in each BA. Begin-
ning with self-generation, for each BA we identify
the fraction of generation in each hour that is going
towards meeting demand within the BA. Given the
physically interconnected nature of the grid, we
assume that the generation fraction can be applied
to all power plants operating in the BA during the
given hour. Based on this assumption, we calculate
the hourly emissions associated with self-generation
for each plant, as in equation (3). Here ESG is the
hourly emissions of each plant associated with self-
generation in the BA, fBA is the fraction of genera-
tion going towards self-generation, and Epis the total
hourly emissions of the plant. Scaling each plant’s
emissions by the generation fraction requires the
assumption that emissions of each plant are propor-
tional to generation at the same plant:
ESG =fBA ×Ep.(3)
For imports, we follow a similar procedure. For each
BA, we identify all BAs that electricity is impor-
ted from. For each BA providing imports, we cal-
culate the fraction of generation in that BA going
towards providing imports to the initial BA, as shown
in equation (4). Ii,j represents the hourly electricity
imports from the ith BA to the jth BA, and Girep-
resents the total hourly generation in the ith BA.
Using the same assumptions as for self-generation,
we use the fraction of generation in the BA provid-
ing imports to calculate the hourly emissions of each
power plant in the BA, as in equation (5), where
Ei,j represents the emissions associated with imports
from the ith BA to the jth BA:
fimports =Ii,j
Gi
(4)
Ei,j=fimports ×Ep.(5)
This process is repeated until hourly plant emis-
sions have been calculated for plants in all BAs from
which the initial BA imports electricity. To model
the fate and transport of the pollutants, we use the
InMAP Source Receptor Matrix (ISRM) [13]. The
ISRM describes the marginal increase in PM2.5 con-
centration in each grid cell due to 1 ton emissions of
each pollutant in each grid cell assuming stack height
emissions. As health damages from exposure to PM2.5
are associated with annual average concentration, the
input to the ISRM is the annual total emissions in each
location. Thus, to use our plant-level emissions data
with the ISRM, we sum the hourly data to achieve the
annual total for each plant. The output of the model is
the increase in concentration of fine particulate mat-
ter across the US. While there are a number of reduced
complexity air quality models available, we choose to
use InMAP due to the ability to model emissions on
a fine scale. Previous work has demonstrated that a
finer grid-scale results in more accurate estimates of
exposure to PM2.5 [9,22].
To measure the health impacts of PM2.5 expos-
ure, we calculate the resulting increase in premature
mortality, using equation (6) [23,24], where ∆Mxis
the change in premature mortality in each grid cell,
Mx0is the all-cause mortality rate, βis the PM2.5
coefficient, ∆PM2.5 is the change in annual average
PM2.5 concentration, and Pxis the population in each
grid cell:
∆Mx=M0
x(eln(β)
10 ∆PM2.5−1)×Px.(6)
We aggregate results to the BA and county-level by
summing the premature mortality and population in
each grid cell within the boundaries of the geography.
Grid cells that span multiple BAs or counties are split
between them. We assess the impacts in three ways:
(a) total premature mortality caused by electricity
imports and self-generation in each BA; (b) prema-
ture mortality per TWh caused by electricity imports
and self-generation in each BA, and (c) premature
mortality per capita occurring in each BA and each
county.
In addition to tracking health impacts, we
track CO2emissions caused by imports and self-
generation. To do this, we determine CO2emis-
sions using the same method used to determ-
ine criteria pollutant emissions, as described by
equations (1)–(3). We then sum the emissions Ei,j
from all of the plants in BAs that the initial BA
imports from to determine the total CO2emissions
associated with imports. We sum emissions ESG from
all plants in the BA to determine the CO2emissions
associated with self-generation.
There is uncertainty in each stage of our modeling
and data, including uncertainty regarding emissions,
the resulting PM2.5 concentrations as determined by
InMAP, and the health impacts determined from the
dose-response function. Uncertainty in the emissions
4
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
is present in two forms: first, from measurement
errors in the underlying data source, which has been
estimated to be <5% [25], and second in variab-
ility in emissions and electricity transfers between
years. To address the latter uncertainty, we perform
the same baseline analysis for another year (2016),
with different generation and consumption patterns
(see SI section 4.2). To assess uncertainty in the RCM
modeling, we repeat our analysis using APSCA [26],
another source receptor matrix built using EASIUR
[27]. Overall, InMAP results in ∼10% more deaths
than APSCA (see SI section 4.1). While RCMs are less
accurate than chemical transport models (CTMs),
they have been shown to perform well for population-
weighted exposure analyses [28]. The largest source of
uncertainty is in the relationship between PM2.5 con-
centration and mortality. To assess this uncertainty,
we use two different dose-response functions based
on distinct cohort studies, one by Krewski et al [24]
and one by Lepeule et al [23], which we present as a
high and low estimate of mortality. These estimates
vary by a factor of nearly 2.5. Further details on the
basis for these functions and a comparison of the res-
ults can be found in the SI in table S3.
3. Results
3.1. Overall premature mortality
While traditionally policy on air quality and health
has focused on where emissions occur, we find
that where electricity is consumed is an equally
important issue warranting further policy attention.
Importantly, we find that electricity consumption in
the contiguous United States caused between ∼4000
[1664–6505] (low estimate, using [24]) and ∼9000
[4716–13896] (high estimate, using [23]) premature
deaths in 2019. Electricity consumption in a BA can
result in damage in a different BA in two ways: by
importing electricity generated in another BA, or by
producing electricity itself that leads to transport of
pollution and exposure in another BA. In the US, 92%
of health damages are associated with electricity pro-
duced in the same BA it is consumed in (hereafter
called ‘self-generation’), though nearly half of these
damages occur outside of the BA due to transport of
pollution. The remaining 8% are caused by electricity
imported from a different BA than it is consumed in.
In contrast to self-generation damages, 86% of mor-
tality caused by imports occurs in a different BA than
where the electricity is consumed. A future grid with a
larger role for power exchanges between regions could
lead to imports causing a larger portion of damages.
Our estimates are in line with mortality estimates
from electricity generation in the United States from
previous studies, which range from 8500 to 16 400 for
various years of analysis [5,9,10,22].
These results are geographically heterogeneous.
Figures 2(g) and (h) highlight the variation in pre-
mature mortality caused by electricity consumption
in each BA. Consumption of electricity in large BAs
in the Midwest and Eastern US cause far more
damage than consumption of electricity in BAs in
the Western, Southeastern, and Northeastern US.
This is true for both imported electricity consump-
tion and self-generation. Consumption of power
in Midcontinent ISO (MISO) causes more deaths
than any other BA through both self-generation and
imports. Consumption of electricity in PJM Inter-
connect (PJM), Electric Reliability Council of Texas
(ERCO) and Southwest Power Pool (SWPP) leads to
significant premature mortality from self-generation,
while California ISO (CISO) and Tennessee Val-
ley Authority cause significant mortality from their
electricity imports.
BAs causing the most mortality correspond to
those consuming the most electricity. Figure 2(a) and
(b) show the electricity consumption of each BA as
self-consumption and imports. In self-generation, the
BAs consuming the most electricity are PJM, MISO,
ERCO, and SWPP. These are the same four BAs
responsible for the most deaths from self-generation,
together accounting for 74% of premature mortal-
ity associated with self-generation. For electricity
imports, CISO and MISO are the two BAs with the
most electricity imports and the BAs causing the most
deaths from their electricity imports, but overall there
is less alignment between BAs with the highest electri-
city consumption and the highest damages.
BAs vary greatly in size and population, which
has a direct impact on the total damages each BA is
responsible for. We normalize the consequences of air
pollution as deaths caused/TWh consumed, and deaths
caused/million customers served. We include the met-
ric of deaths caused per million customers served as
a proxy for normalizing by the population density of
each BA, given that normalizing by the actual popu-
lation would be misleading due to a large portion of
mortality occurring outside of each BA. When deaths
are normalized by electricity consumed, the patterns
in mortality caused are largely the same as for total
mortality, though less pronounced. BAs in the Mid-
west cause the most damage per TWh through both
imports and self-generation. BAs in the Intermoun-
tain West cause more significant deaths/TWh through
their imports than they do on a total mortality basis.
In the SI we illustrate these results with figure S1.
Based on deaths caused per customer served, the pat-
tern for self-generation is similar, but a somewhat dif-
ferent pattern emerges for imports. By this metric, the
most damaging BAs are in the Intermountain West
and the Midwest. This suggests that while imports to
BAs in the Intermountain West are not as damaging
on a per-energy basis, they cause a disproportionate
amount of damage relative to the service they provide.
BAs in the West Coast, on the other hand, have a lower
energy per customer footprint, and as a result cause
fewer damages per customer served (see SI figure S2).
We explore the impacts of various emissions scen-
arios in SI section 3.
5
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
Figure 2. Electricity consumption, emissions, and mortality caused by imports and self-generation. (a) Electricity imported by
each BA (‘electricity imports’) (in TWh); (b) electricity produced in the same BA where it is consumed (‘self-generation);
(c) NOxemission intensity (in tonnes NOx/TWh) associated with electricity imported by a BA; (1d) NOxemission intensity (in
tonnes NOx/TWh) from self-generation; (e) SO2emission intensity (in tonnes SO2/TWh) associated with electricity imported by
a BA; (f) SO2emission intensity (in tonnes SO2/TWh) from self-generation; (g) annual premature mortality (deaths/year) from
air pollution from electricity imports in a BA (regardless of where the premature mortality occurs); (h) annual premature
mortality (deaths/year) from air pollution from electricity self-generation in a BA (regardless of where the premature mortality
occurs). The high estimate of mortality is shown here. The geographic spread is the same for the low estimate, but the magnitude
is roughly half.
3.2. Premature mortality transfers
In figure 3, we link the BAs responsible for premature
mortality to the regions where mortality occurs. For
both electricity imports and self-generation, the Mid-
west is the most impacted region, while New England,
the West Coast, and the Rocky Mountain regions
are the least impacted. For self-generation, prema-
ture mortality from air pollution primarily occurs
within the BA consuming electricity and in nearby
BAs, with mortality occurring outside of the BA gen-
erally occurring to the East of the BA. It is intuit-
ive that most of the effects are within the BA or in
nearby BAs as self-generation is the electricity gener-
ated within the BA, but pollutants are also transpor-
ted away from the source, causing damages at further
distances [9,10]. Damages from electricity imports
do not follow the same pattern, and have more vari-
ation. While some BAs, such as Duke Energy Florida,
Inc. (FPC) and Florida Power & Light Co. (FPL), both
located in Florida, primarily impact the region they
are located in, others such as CISO have significant
impacts in a variety of regions.
3.3. Environmental justice and distributional
effects
To gain greater insight into how different com-
munities are impacted by air pollution-related health
damages associated with electricity consumption, we
estimate the impacts specific to each race and ethni-
city. To do so, we overlay the gridded PM2.5 concen-
trations from InMAP with US Census block groups
to calculate the race/ethnicity-specific mortality in
each block group. We illustrate this approach in the
SI in figure S3, which shows an example of this cal-
culation for Black and White populations in an area
around Atlanta, Georgia. Further methodological
details are shown in SI section 2.3. Our fine-grained
results for race/ethnicity impacts are then aggreg-
ated to the BA level and to the national level. Spe-
cifically, we compute aggregate race-specific impacts
from a given BA as the ratio of the total number
of deaths caused by that BA for that race/ethnicity
to the total population of that race/ethnicity, where
total deaths and population are summed at the
block group level. Similarly, national impacts for a
6
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
Figure 3. Distribution and transfer of premature mortality caused by imports and self-generation in the most damaging BAs. The
left node contains the ten most-damaging BAs, by premature mortality they are responsible for. The right node contains the
region where the mortality occurs. The high mortality estimate is shown here.
race/ethnicity are computed as the ratio of the total
number of deaths for that race/ethnicity to the total
population for that race/ethnicity. On a national
level, Blacks have the highest race-specific mortality
rate (∼3.3 Black deaths/100 000 Blacks) followed by
Whites (∼2.8 White deaths/100 000 Whites), which
is slightly higher than the overall national aver-
age (∼2.5 deaths/100 000). Additionally, individu-
als below the federal poverty line have a mortal-
ity rate of ∼2.7 deaths/100 000, slightly above the
average. All other groups have somewhat lower elec-
tricity consumption-related mortality rates. We are
not the first to suggest that exposure to air pollu-
tion is correlated to race and income. Numerous
studies have demonstrated disparities in air pollution
exposure [29–31]. Thind et al also find that Blacks
are disproportionately impacted by electricity-related
PM2.5 exposure [10].
However, as with other metrics, these results are
geographically heterogeneous. Figure 4shows the
percent difference between the race/ethnicity-specific
mortality caused by each BA and the average mor-
tality for all races caused by each BA. This includes
both deaths caused within the BA and deaths caused
outside of it. No BAs cause White deaths/capita
that are more than 14% greater than the average
deaths/capita. BAs in the Southeast and Texas cause
disproportionately high Black deaths/capita, while
BAs in the West cause fewer Black deaths/capita than
average. BAs in the Pacific Northwest and Intermoun-
tain West cause disproportionately high Native Amer-
ican deaths/capita. BAs in the West, specifically CISO,
cause high Asian, Pacific Islander, Hispanic, and
Other deaths/capita. New York ISO likewise causes
disproportionately high Asian, Hispanic, and other
deaths/capita.
3.4. Carbon emissions
In addition to assessing health impacts of electri-
city imports, we assess climate impacts, by calculat-
ing CO2emissions attributed to imports. Overall, we
find CO2emissions from imports represent 9.5% of
CO2emissions from electricity consumption. Emis-
sions are once again geographically heterogeneous, as
shown in the SI in figure S4, with BAs in the Midwest
and Eastern US causing more emissions than BAs in
the Western US. This is true for both self-generation
and imports, with the exception of CISO, with over
50% of its CO2emissions caused by imports. This
can be explained by CISO’s relatively high total elec-
tricity imports (27%), and that it imports from BAs
with relatively high CO2emissions/TWh. While CO2
is not responsible for the health impacts of electri-
city consumption, CO2emissions are correlated with
other air pollutant emissions, and the BAs causing the
7
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
Figure 4. Percent difference between race/ethnicity-specific deaths/capita caused by each BA’s electricity consumption and average
deaths/capita caused by each BA’s electricity consumption, regardless of where deaths occur. Red indicates the specified race has
higher deaths/capita caused by the BA than the average deaths/capita caused by the BA. Blue indicates the specified race has lower
deaths/capita caused by the BA than the average deaths/capita caused by the BA.
most deaths also cause the most CO2emissions. See
SI figure S5 for details.
3.5. Significance of imports
In most BAs, particularly those in the Midwest and
Eastern US, self-generation causes more overall dam-
ages. However, this is not uniformly true across the
country. Figure 5shows the percentage of mortality
caused by the power consumption in each BA attrib-
utable to electricity imports. Notably, the West Coast
stands out. In nearly all Western BAs, more than 50%
of mortality attributed to consumption in each BA
is caused by electricity imports. This is in part due
to Western BAs importing a greater fraction of their
electricity. Additionally, Western BAs tend to have
cleaner grids [11], meaning that self-generation in
those BAs produces lower emissions. Importing from
BAs further East with higher emissions factors results
in electricity imports contributing a larger portion
of damages.
While the total impacts of each BA are highly
dependent on the amount of electricity the BA gen-
erates and imports, we can gain further understand-
ing by comparing the damages per TWh. In addition
to relying on electricity imports for a larger fraction
of their electricity consumption, Western BAs’ elec-
tricity imports have higher NOxand SO2emissions
Figure 5. Percent of premature mortality caused by each BA
due to imported electricity.
intensities than their self-generation (see SI figure
S6 for details). This results in imports causing more
damages per unit of electricity consumed than self-
generation. In contrast, BAs in the Rocky Mountain
region import power from Western BAs with cleaner
grids, and as a result, cause fewer emissions and mor-
tality per unit of electricity imported than per unit of
electricity generated within the Rocky Mountain BAs.
In the Midwest, there is some variation, but overall
emissions intensities and deaths/TWh are similar for
imports and self-generation.
4. Discussion and conclusions
Our results demonstrate that imports are responsible
for a small, but significant percentage of premature
8
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
mortality. Furthermore, diferent demographic
groups are not impacted equally, with Blacks exper-
iencing the highest impacts. While the total amount
of premature mortality caused by imports is relatively
small (about 700 deaths), this is likely to increase if
grid interconnections are expanded in grids with large
amounts of coal-fired power generation. While BAs
in the Midwest and Eastern US cause the most prema-
ture mortality in total, electricity imports represent
a much higher fraction of premature mortality in
Western BAs. BAs such as CISO cause more than half
of their air pollution damages through imports, most
of them significant distances from the BA. CISO is of
particular concern, as the BA has supported increas-
ing grid interconnection throughout the West. The
Energy Imbalance Market allows for increased trad-
ing, which could result in higher levels of electricity
imports. As we have shown, in 2019 the results of
this trading were primarily seen in the four corners
region, where populations are not benefiting from
CISO’s electricity. Notably, the Navajo Generating
Station, a coal plant responsible for some of these
impacts, shut down in late 2019 [32], suggesting that
in future years the impacts of CISO’s imports will be
lower.
Our results are highly dependent on the state of
the electricity system. A rapid shift away from coal
and towards renewables and/or natural gas could
drastically change the distribution of impacts due to
imports (SI section 3). A number of states in the West
have set ambitious renewable energy goals and net-
zero emissions targets. California’s SB100 mandates
60% renewable power by 2030, and carbon-free elec-
tricity by 2045 [33]. Legislation in Washington, SB
5116, calls for the elimination of coal power by 2025
and 100% carbon-free electricity by 2045 [34]. Legis-
lation in Nevada (SB 358), which requires 50% renew-
able electricity by 2030 [35], is less stringent, but
would still represent significant increases in renew-
able energy. A system-wide reduction in coal would
reduce the impacts to distant communities of Western
BAs’ imports. However, given different state policies
and renewable portfolio standards, it is likely that dif-
ferent regions of the country will follow different tra-
jectories for reducing emissions. As such, it is essen-
tial that future assessments of the impact of electricity
imports account for changing grid conditions.
Imports not only affect air pollution emissions,
but also carbon emissions. While generally, increasing
electricity exchange increases the viability of renew-
able energy and associated benefits of reduced emis-
sions, this may not be the case everywhere. On
average, electricity imports have a lower carbon emis-
sions factor than within-BA generation. However, as
with premature mortality, this is highly heterogen-
eous. In some BAs, imports have higher carbon emis-
sions factors as is the case in much of the West.
This suggests that carbon reduction programs should
treat imports differently in different regions. While
BAs in the Midwest would see a carbon benefit from
increasing the share of imports, Western BAs with
relatively clean grids, such as CISO would actually
see an increase in carbon by increasing the share of
imports. As the grid grows progressively cleaner, the
distributional effects of power imports are expected
to decrease, but it is essential that impacts of electri-
city imports are monitored and considered during the
energy transition.
Data availability statement
The data that support the findings of this study are
available upon reasonable request from the authors.
Acknowledgments
This work was supported by the Energy Resources
Engineering Department at Stanford University. The
authors declare no conflicting interests.
ORCID iDs
Eleanor M Hennessy https://orcid.org/0000-0002-
9471-5765
Jacques A de Chalendar https://orcid.org/0000-
0002-3586-199X
Sally M Benson https://orcid.org/0000-0002-
3733-4296
Inˆ
es M L Azevedo https://orcid.org/0000-0002-
4755-8656
References
[1] National Renewable Energy Laboratory 2012 Executive
Summary Renewable Electricity Futures Study (Golden, CO:
National Renewable Energy Laboratory) (https://doi.org/
10.2172/1063076)
[2] Frew B, Becker S, Dvorak M, Anderson G and Jacobson M
2016 Flexibility mechanisms and pathways to a highly
renewable US electricity future Energy 101 65–78
[3] World Health Organization 2018 Ambient (outdoor) air
pollution (available at: www.who. int/news-room/fact-
sheets/detail/ambient-(outdoor)-air-quality-and-health)
(Accessed 5 May 2020)
[4] Lim S S et al 2012 A comparative risk assessment of burden
of disease and injury attributable to 67 risk factors and risk
factor clusters in 21 regions, 1990–2010: a systematic analysis
for the global burden of disease study 2010 Lancet
380 2224–60
[5] Dedoussi I, Eastham S, Monier E and Barrett S 2020
Premature mortality related to United States cross-state air
pollution Nature 578 261–5
[6] Pope C A, Thun M J, Namboodiri M M, Dockery D W,
Evans J S, Speizer F E and Heath C W 1995 Particulate air
pollution as a predictor of mortality in a prospective study of
US adults Am. J. Respir. Crit. Care Med. 151 669–74
[7] Dockery D, Pope C A, Xu X, Spengler J D, Ware J H,
Fay M E, Ferris B G and Speizer F E 1993 An association
between air pollution and mortality in six US cities New
Engl. J. Med. 329 1753–9
[8] US Environmental Protection Agency Health and
environmental effects of particulate matter
(PM)—particulate matter (PM) pollution
9
Environ. Res. Lett. 17 (2022) 064011 E Hennessy et al
(available at: www.epa.gov/pm-pollution/health-and-
environmental-effects-particulate-matter-pm) (Accessed 5
May 2020)
[9] Goodkind A L, Tessum C W, Coggins J S, Hill J D and
Marshall J D 2019 Fine-scale damage estimates of particulate
matter air pollution reveal opportunities for
location-specific mitigation of emissions Proc. Natl Acad. Sci.
USA 116 8775–80
[10] Thind M P S, Tessum C W, Azevedo L and Marshall J D 2019
Fine particulate air pollution from electricity generation in
the US: health impacts by race, income, and geography
Environ. Sci. Technol. 53 14010–9
[11] de Chalendar J A, Taggart J and Benson S M 2019 Tracking
emissions in the US electricity system Proc. Natl Acad. Sci.
USA 116 25497–502
[12] Tessum C W et al 2019 Inequity in consumption of goods
and services adds to racial-ethnic disparities in air pollution
exposure Proc. Natl Acad. Sci. USA 116 6001–6
[13] Goodkind A, Tessum C, Coggins J, Hill J and Marshall J
InMAP source-receptor matrix (ISRM) dataset Zenodo
(available at: https://zenodo.org/record/2589760#.
YO9b25NKgV) (Accessed 11 March 2019)
[14] Tessum C W, Hill J D and Marshall J D 2017 InMAP: a
model for air pollution interventions PLoS One 12 e0176131
[15] Industrial Economics, Incorporated 2019 Evaluating
reduced-form tools for estimating air quality benefits
Technical Report (US: EPA)
[16] California Independent System Operator Western EIM
(available at: www.westerneim.com/Documents/
WesternEIMFactSheet.pdf) (Accessed 10 June 2021)
[17] MacDonald A, Clack C T M, Alexander A, Dunbar A,
Wilczak J and Xie Y 2016 Future cost-competitive electricity
systems and their impact on us CO2emissions Nat. Clim.
Change 6526–31
[18] US Energy Information Administration 2019 (available at:
www.eia.gov/electricity/gridmonitor/about) (Accessed 5
March 2020)
[19] US Environmental Protection Agency 2019 (available at:
https://ampd.epa.gov/ampd/) (Accessed 18 October 2020)
[20] US Environmental Projection Agency 2017 (available at:
www.epa.gov/air-emissions-inventories/2017-national-
emissions-inventory-nei-data (Accessed 18 October
2020)
[21] US Energy Information Administration 2019 (available at:
www.eia.gov/electricity/data/eia861/) (Accessed 23 January
2020)
[22] Paolella D A, Tessum C W, Adams P J, Apte J S, Chambliss S,
Hill J, Muller N Z and Marshall J D 2018 Effect of model
spatial resolution on estimates of fine particulate matter
exposure and exposure disparities in the United States
Environ. Sci. Technol. Lett. 5436–41
[23] Lepeule J, Laden F, Dockery D and Schwartz J 2012 Chronic
exposure to fine particles and mortality: an extended
follow-up of the Harvard Six Cities study from 1974 to 2009
Environ. Health Perspect. 120 965–70
[24] Krewski D et al 2009 Extended follow-up and spatial analysis
of the American Cancer Society study linking particulate air
pollution and mortality Res. Rep. Health Eff. Inst. 140 5–114
[25] Gluck S, Glenn C, Logan T, Vu B, Walsh M and Williams P
2012 Evaluation of NOxflue gas analyzers for accuracy and
their applicability for low-concentration measurements J. Air
Waste Manage. Assoc. 53 749–58
[26] Heo J, Adams P and Gao H 2017 Public health costs
accounting of inorganic PM2.5 pollution in metropolitan
areas of the United States using a risk-based source-receptor
model Environ. Int. 106 119–26
[27] Heo J, Adams P J and Gao H Oliver 2016 Reduced-form
modeling of public health impacts of inorganic PM 2.5 and
precursor emissions Atmospheric Environ. 137 80–89
[28] Gilmore E A et al 2019 An inter-comparison of the social
costs of air quality from reduced-complexity models
Environ. Res. Lett. 14 070416
[29] Bell M and Ebidu K 2012 Environmental inequality in
exposures to airborne particulate matter components in the
United States Environ. Health Perspect. 120 1699–704
[30] Clark L, Millet D and Marshall J 2014 National patterns in
environmental injustices and inequality: outdoor NO2air
pollution in the United States PLoS One 9e94431
[31] Miranda M, Edwards S, Keating M and Paul C 2011 Making
the environmental justice grade: the relative burden of air
pollution exposure in the United States Int. J. Environ. Res.
Public Health 81755–71
[32] Greentech Media The Navajo generating station coal plant
officially powers down. Will renewables replace it? (available
at: www.greentechmedia.com/articles/read/navajo-gene
rating-station-coal-plant-closes-renewables) (Accessed 15
December 2020)
[33] California Legislative Information Sb-100 California
renewables portfolio standard program: emissions of
greenhouse gases (available at: https://leginfo.legislature.ca.
gov/faces/billNavClient.xhtml?bill_id=201720180SB100)
(Accessed 20 August 2020)
[34] Washington State Legislature Sb 5116-2019-20 (available at:
https://app.leg.wa.gov/billsummary?BillNumber=5116&
Initiative=false&Year=2019) (Accessed 20 August 2020)
[35] Nevada State Legislature SB358 (available at: www.leg.
state.nv.us/App/NELIS/REL/80th2019/Bill/6651/Overview)
(Accessed 20 August 2020)
10
