Still Toxic After All These Years

Still Toxic After All
These Years
Air Quality and Environmental Justice in the
San Francisco Bay Area
February 2007
Manuel Pastor, UC Santa Cruz
James Sadd, Occidental College
Rachel Morello-Frosch, Brown University
Prepared for the Bay Area Environmental Health Collaborative by the
Center for Justice, Tolerance & Community, University of California, Santa Cruz

Foreword
In the last several decades, policy makers have paid increasing attention to issues of
environmental justice – the idea that the health burdens and risks of poor air quality and
proximity to hazards are inequitably distributed by race and income. In 1994, for example,
the Clinton administration adopted an executive order that made environmental justice a lens
through which all federal environmental policy was to be assessed. The state of California took up
the banner in 1999, with legislation that directed the state’s environmental agencies to develop
environmental justice policies and strategies.
This shift in public sector attention has not been due to a sudden burst of goodwill on the part of
regulators. Rather, public pressure by vibrant community-based groups coupled with mounting
evidence compiled by academic researchers have both made the case clear and changed the
political calculus. This winning combination of community voice and careful research has created
examples of change across the country, and more groups have sought to develop partnerships
that would generate both the scientiﬁc platform and the organizing energy to protect community
health.
This report emerges from one such partnership: the Bay Area Environmental Health
Collaborative (BAEHC). With support from the San Francisco Foundation, some of the Bay
Area’s leading environmental justice and community health organizations came together with
researchers from the Center for Justice, Tolerance and Community of UC Santa Cruz (CJTC) to
help document the state of environmental disparity in the Bay Area. The results, as detailed in the
report, are clear: environmental inequality is unfortunately alive and well, a fact that threatens the
wellness of the most aﬀected communities.
The issue, of course, is what should be done to reduce disparities and improve environmental
quality for everyone in the Bay Area. At the end of this report, we oﬀer some possible policy
directions but we realize that these are merely a start to a longer conversation between
community, business and regulatory leaders. Our only insistence is that such a dialogue be
initiated: as we witnessed in the Hurricane Katrina disaster of 2005, leaving some of us less
protected ultimately poses environmental risks and costs for everyone.
For supporting this project, including its community outreach component, we thank The
California Endowment, the California Wellness Foundation, The San Francisco Foundation, and
the W. K. Kellogg Foundation. For providing careful data analysis, we thank Justin Scoggins, Bill
Jesdale, and for inspiring us with their energy, enthusiasm and commitment to this important
work, we thank the activists and leaders involved in the Bay Area Environmental Health
Collaborative.
• Manuel Pastor, James Sadd, Rachel Morello-Frosch, February 2007

Contents
Introduction
1
..........................................................................................
Understanding the Stakes ...................................................................2
Understanding the Data ......................................................................4
Toxic Releases and Geographic Proximity
6
..........................................
Ambient Air Toxics and Estimated Health Risks
8
..................................
Potential Policy Implications ..............................................................10
Looking Forward ................................................................................14
Technical Appendix ............................................................................15
References ........................................................................................17
Selected Web Resources ..................................................................18

Introduction
On May 15, 2006, residents in
Bayview Hunters Point, a mostly
minority low-income community
nestled at the bottom edge of San
Francisco, celebrated a stunning
victory: after nearly a quarter
century of organizing, protests, and
civil disobedience, they convinced
Paciﬁc Gas and Electric to ﬁnal y
shutter one of California’s oldest
and dirtiest power plants. In an
area already saturated with toxic
sites and high pol ution emissions,
community members had long
blamed the plant for elevated levels
of asthma, cancer, and other health
ailments – and local resident Tessie
Ester seemed to express the general
sentiment when she commented:
On April 11, 2006, Bayview Hunters Point community residents and members of
“When I look over at those stacks,
Greenaction for Health and Environmental Justice blockaded the gates of PG&E’s
and there is nothing coming out, I
Hunter’s Point power plant. State and PG&E officials announced the closure of the
plant weeks later.
can’t help but cry.”
of idling. This, too, has been long in the making:
Just a month earlier, activists from Richmond, a
activists in West Oakland have been complaining
largely minority inner-ring suburb dotted with
for years that they face a per capita level of toxic
petrochemical facilities, managed to persuade
diesel particulates that is seven times that for the
the Bay Area Air Quality Management District
rest of Alameda County.
(BAAQMD) to tighten up regulations on a
practice cal ed “ﬂaring.” Flaring occurs when
While these recent environmental justice victories
reﬁnery operators attempt burn oﬀ gas build-
are heartening, the length of time it has taken
ups; while reﬁneries argue that this is safer than
to address these problems is not. Moreover, the
releasing the gases, those living along the fencelines
commonality of disparity – in which communities
of such operations have long worried about the
with lower incomes and higher proportions of
health impacts. Once again, the victory was sweet
minority residents are more often subject to
but long in coming: activists in Richmond have
environmental hazards – has led some to argue
been ﬁghting for ﬂare control for over twenty years.
for a more comprehensive and precautionary
And they know that they wil have to continue to
approach, one that would both prevent or reduce
monitor progress – the devil is ultimately in the
exposures and health risks before they occur and
details of the BAAQMD’s implementation of the
reduce persistent inequalities.
new ﬂare rules.
This report seeks to contribute to that task by
Meanwhile, the Oakland-based “Ditching Dirty
both documenting environmental disparity in
Diesel” col aborative has, over the past few years,
the Bay Area and oﬀering some principles for
been recognized by both foes and friends for its
new strategies. We begin by discussing the data
eﬀorts to curtail truck and school bus emissions
we use to look at disproportionate environmental
in minority and low-income neighborhoods. The
exposures, and then highlight the patterns revealed
community-based campaign, in which activists
by several types of quantitative analysis. The main
reached out to both business owners and aﬀected
point we make is simple: even after control ing
residents, has been credited with inﬂuencing the
for income, land use and other variables that
state’s Air Resources Board to adopt a rule that wil
are frequently used to explain away disparate
require al diesel trucks to have a device that wil
patterns of exposure, we stil ﬁnd a separate and
automatical y shut engines oﬀ after ﬁve minutes
independent eﬀect of race on estimated pol ution
–1–

burdens. Communities of color, who often feel they
due to poverty and psycho-social stressors. And it
may be disparately impacted by undesirable land
wil require turning rhetoric into reality in terms
uses, indeed have reason to be concerned.
of community engagement and neighborhood
This presents an opportunity as wel as a chal enge:
protection.
surely Bay Area leaders, who often pride themselves
The focus in this report might seem speciﬁc
on their devotion to environmental preservation
– how best to insure that current disparities wil
and protection, can develop an environmental
be reduced. But such an eﬀort can serve a broader
justice policy which leads the state and its regions.
purpose. Researchers are beginning to ﬁnd that
To do so wil require both adding real force to
where health and environmental disparities
the current set of rules as wel as developing new
are greatest, overal health and environmental
regulatory strategies. It wil require a cumulative
outcomes are worse for everyone. Protecting the
approach that considers multiple sources of air
most vulnerable populations can lead to better
pol ution, a precautionary strategy that puts health
environmental protection for al of us – and it can
before economic interests, and a social overlay
only be done if we examine the patterns of disparity
that accounts for disparities and vulnerabilities
honestly and work together for a better future.
Understanding the Stakes
Despite the history of activism and action around
Most of this literature, however, has not
environmental justice issues in the Bay Area,
oﬀered a quantitative assessment of the overal
the literature shows a surprising gap: there have
environmental disparity in the region. This
been no published analyses of the overal state of
may seem a minor gap: after al , pol s from the
environmental disparity in the region.
Public Policy Institute of California suggest
This is not to say that there has not been a wealth
that Californians from al ethnic groups agree
of studies focused on particular pol utants or
that environmental “bads,” like toxic wastes,
particular communities. Recent exemplary studies
are more likely to be in minority communities
include the Paciﬁc Institute’s 2003 publication
while environmental “goods”, such as open space
Clearing the Air (which focused on diesel pol ution
and parks, are less likely to be found in those
in West Oakland) as wel as its 2006 publication
neighborhoods.
Paying With Our Health (which focused on the

impacts of the goods transport
industry on several Bay Area
and California neighborhoods);
also both important and path-
breaking was Breathing Fire,
a 2005 publication about
ﬂaring released by the West
County Toxics Coalition and
Global Community Monitor.
Academics have also been busy,
with a slew of excel ent academic
articles and books produced
by distinguished researchers
such as Dara O’Rourke from
UC Berkeley, Andrew Szasz
from UC Santa Cruz, David
Pel ow from UC San Diego,
and Michael Lipsett from
the Environmental Health
Investigations Branch of the
Department of Health Services,
Organizing for a healthy environment, great jobs, and healthy foods - members of People
among others.
Organizing to Demand Environmental and Economic Rights (PODER) and the Chinese
Progressive Association (CPA) at a community press event.
–2–

But public perceptions
of inequality and social
science research are not
always in line. After a
pioneering study on
environmental disparity
issued by the United
Church of Christ twenty
years ago helped to
launch a wave of activism
and policy concern, a
subsequent set of studies
conducted by researchers
at the University of
Massachusetts in the
In July of 2005, PODER and the Mission Anti Displacement Coalition organized a grassroots
forum to bring together young people, families, and elders to learn how to leverage strong
mid-1990s argued that
community benefits from the proposed land use changes in their neighborhoods.
environmental inequality
did not hold across the country and further
tend to be where income levels and property values
suggested that income diﬀerentials explained
are low, and co-location of the poor and toxics
most observed racial disparities in the siting of
simply reﬂects the normal workings of the market
environmental hazards.
system. In both perspectives, while health impacts
could remain signiﬁcant, environmental disparities
Since then, the empirical debate has raged on,
are basical y benign in intent – the association of
with methodological disputes centered on both
particular neighborhoods and particular hazards is
statistical techniques and the scale of analysis.
seen as a matter of accident or choice.
On the scale issue, new studies have indicated
that disparities should be assessed on a regional
The power perspective suggests that hazard location
basis since this reﬂects the reality of economic
and poor air quality depends on a community’s
geographies – furniture factories in Los Angeles
ability – or inability – to resist placement of
are not likely to move to Seattle and software
undesirable land uses in their neighborhood. In
developers in Seattle were not like to head south,
this view, discriminatory practices and/or power
and so inequality has to be considered in the
diﬀerentials are largely responsible for the patterns
context of the industrial clusters, economic
of environmental disparity that are frequently
development, and traﬃc patterns that exist in any
observed. And since race and power are so highly
particular metro area. When this approach is taken,
intertwined in our society, patterns of diﬀerence
disparities by race show up consistently and while
by race are suggestive of patterns of diﬀerence by
income does seem to matter, control ing for it does
power.
not always eliminate the racial gap.
In the real world, al three of these factors – land
The stress in the research literature on the role
use, income and power– are inextricably linked.
of race is not simply a function of civil rights
Communities with less political voice may be less
concerns. Rather it is deeply connected to
able to contest incompatible land uses, and income
understanding and weighing the merits of three
is not just a reﬂection of a market system but
strands of explanation for the location of both
also a marker of inﬂuence in the decision-making
hazards and emissions: those analyses that focus
process. Yet, if race stil matters once land use and
on land use, those that emphasize the importance
income levels are accounted for, this suggests that
of income, and those that lift up the dynamics of
diﬀerential access to political power and policy
power.
voice may be important to consider and address in
the regulatory process.
The land use perspective suggests that hazards are
located where complementary land uses, such as
Of course, another reason to be worried about
industrial facilities or traﬃc arteries, are clustered;
racial diﬀerence in hazard location is simple
therefore, any correlation of environmental “bads”
because of the potential impacts on the health
with race is viewed as an unfortunate byproduct of
and wel -being of diﬀerent ethnic communities.
economic geography. The income view sees the role
At stake in the empirical debate, then, are
of property values as key: more hazardous land uses
both theories of causation and strategies for
–3–

improving public health. In this light, a broad
on school children, both in terms of disparate
empirical study of the Bay Area is essential for
exposures and the potential eﬀect on asthma
understanding whether the voiced concerns of
hospitalizations and academic performance.
diverse communities are speciﬁc and isolated cases
or part of a broader regional pattern that regulatory
Along every dimension, there is persistent and
agencies should address.
strong empirical evidence of environmental
inequality in Southern California. It is, therefore,
Such studies have been done in other parts of
no surprise that many eﬀective community-
California. In a series of analyses, we examined the
based groups have emerged, making the region a
situation in the Los Angeles metropolitan area from
hotspot of environmental justice organizing and
several diﬀerent vantage points: the distribution
the origin of groundbreaking state legislation on
of treatment, storage, and disposal facilities and
environmental equity sponsored by L.A.-area State
transfer sites, the allocation of large industrial
Senators and Assembly Members.
facilities that are known to release large amounts
of hazardous air pollutants, and the distribution
What about the Bay Area? Are there also general
of cancer and non-cancer health risks associated
patterns of environmental inequality? Are these
with air toxics emissions from mobile and
patterns related to land use, income, or race – or all
stationary emission sources. We have also assessed
three? And if disparities exist, are there any nuances
environmental inequalities and their impacts
in the pattern that can help us understand how best
to protect communities and their health?
Understanding the Data
To get at these issues, we considered several
This report focuses on study results from our
diﬀerent databases on toxic air emissions and
analysis of the TRI and NATA. The decision to use
concentrations, and combined these with
these two federal data sets is partly because they
neighborhood demographic characteristics available
have been used more widely in the academic and
from the 2000 Census, including income levels,
popular literature and this facilitates comparability
ethnicity, and language ﬂuency. The environmental
to other regions and other studies; we would
databases included:
also note that the California EPA’s air toxics risk
• the U.S. EPA’s Toxic Release Inventory (TRI)
data is only available to the public as a set of
for 2003, a col ection of self-reported toxic air
web-based images of grids that translate poorly
emissions data from large industrial facilities;
into the neighborhood-level detail needed for
this exercise. Future research could and should
• the 2001 Community Health Air Pol ution
use the California-generated data, as wel as
Information System (CHAPIS) from the
incorporate community-level estimates of health
California Air Resources Board, an emissions
risks from exposure to outdoor toxic air that are
inventory from both mobile and stationary
now becoming available under the BAAQMD’s
sources, based on emissions inventory
Community Air Risk Evaluation (CARE) project.
information from both the state and some
Improvements in data accuracy and availability for
regional air boards;
future research assessments are among the many
• the 1999 National Air Toxics Assessment
things we cal for at the end of this report.
(NATA), a data set developed by U.S. EPA that
What is in the TRI? Mandated under the
estimates annual average ambient air toxics
Emergency Planning and Community Right-
exposures from both mobile and stationary
to-Know (EPCRA) provisions of the Superfund
emission sources that can be utilized to estimate
Amendments and Reauthorization Act (SARA)
potential cancer risk and respiratory hazard at
of 1986, the Toxic Release Inventory requires
the neighborhood level; and
certain industrial and commercial facilities, as wel
• a set of ambient air toxics concentration and
as federal facilities, to report to the U.S. EPA on
health risk estimates generated by the California
annual releases and transfers of nearly 650 toxic
Air Resources Board using information from
compounds. There are inherent limitations to the
a statewide emissions inventory cal ed the
TRI data: emissions are self-reported estimates and
California Emission Inventory Development
not actual measures of releases; smal area emissions
and Reporting System (CEIDARS).
sources, such as chrome platers, auto body paint
–4–

shops and dry cleaners are not required to report;
signiﬁcant eﬀect of diesel. Speciﬁc details on how
and the TRI does not include releases from mobile
we calculated cumulative risk estimates appear in
sources which are known to signiﬁcantly contribute
the Appendix.
to pol ution levels and health risks.
A few caveats about these cancer and respiratory
Despite its limitations, most of the literature on
hazard estimates are in order. First, these risks are
environmental justice has taken this database as
calculated based on assumptions about ambient
a starting point in asking questions about the
exposures and toxicity and do not represent actual
proximity of certain communities to potential
cancer or respiratory cases. The latter are typical of
hazards. We fol ow that strategy here, ﬁrst
epidemiological studies; the risk estimates we derive
calculating proximity using a complicated process
are instead ecological measures that characterize
of geocoding and double-checking facility location,
a census tract and essential y ask what would be
and then drawing a circle around each facility to
the cumulative impact if a resident lived in the
assess whether a signiﬁcant percent of nearby or
neighborhood for their whole life. In reality, people
adjoining neighborhoods fal within a speciﬁed
constantly move across diverse environments in
distance range. This approach is superior to
a single day – traveling from where they live to
simply asking whether a facility is located within a
where they work to where they go to school to
neighborhood or census tract, since many facilities
where they worship, etc. – and they also move
tend to be on major thoroughfares that border
their households over time from neighborhood to
communities and thus can aﬀect more than one
neighborhood. Despite this, these risk estimates are
area.
useful for comparing the overal pol ution burdens
Our second set of environmental health indicators,
between neighborhoods and clarifying what the
the U.S. EPA’s National Air Toxics Assessment
implications may be for residents’ health.
(NATA) for 1999, is built upon an underlying
A second caveat is that these modeled estimates
inventory of air toxics emissions that includes both
account for ongoing and sustained exposures for
stationary and mobile sources. The inventory is
both stationary and mobile sources – but they
derived from ﬁve primary sources, including state
do not capture what might be termed “episodic”
and local air quality regulatory agencies, EPA’s own
incidents. Such “episodes” could range from
air toxics regulatory program and its TRI database,
occasional ﬂaring at reﬁneries to longer-term air
mobile source emissions estimates developed by
toxic concentrations that can be generated by,
EPA’s Oﬃce of Transportation and Air Quality,
say, oﬀ-road diesel equipment during a multi-
and other emission estimates generated from
year construction project. These exposures are
activity data (such as oﬀ-road sources). Using the
important to the local communities but cannot be
emissions data as inputs, an air dispersion “fate
ful y captured or modeled by the approach taken
and transport” model that accounts for movement
here. Future studies might seek to model these
and atmospheric chemistry of pol utants (due to
eﬀects, particularly to guide remediation activities.
the eﬀect of winds, temperature, and atmospheric
stability) is used to estimate the concentration
Final y, we also utilize land use information
of each air pol utant for each census tract in the
from the 2001 U.S. Geologic Survey (USGS)
continental United States.
Land Cover Characterization Program, an eﬀort
that makes use of aerial photo and satel ite
The NATA data generated by this process includes
imagery interpretation to classify land use at a
tract-level concentration estimates for diesel
spatial resolution of 30 meters. Unfortunately,
particulates and 177 of the 187 air toxics listed
the results combine industrial, commercial, and
under the 1990 Clean Air Act Amendments. The
transportation land uses together; while this is
U.S. EPA also reports ﬁgures on cancer risk and
appropriate for a broad emissions database like
respiratory hazard but these risk estimates do
NATA, it is less clear that it is appropriate for the
not include diesel and some other air toxics. In
TRI’s facilities-based estimates – for which industry
our analysis, we combined cancer potency values
is the driving land use. In the TRI analysis, we
and respiratory hazards values from U.S. EPA
therefore use a proxy based on more readily
as wel as from the California EPA to estimate
available census data, the percent of area employees
cumulative lifetime cancer and respiratory risks
in manufacturing; for further technical details on
associated with ambient air toxics exposure. This
choices regarding variables, techniques, and other
process enabled us to include contributions from
matters, please see the Appendix to this report.
as many pol utants as possible, including the
–5–

Toxic Releases and Geographic Proximity
The easiest way to examine disparities
Figure 1: Locations of Facilities with Air Releases (as Recorded in the Toxic
is to assess visual patterns – that is,
Release Inventory or TRI) Relative to Neighborhood Demographics in the 9-County
Bay Area
to map the locations of TRI facilities
in comparison to underlying
neighborhood demographics. We
do this in Figure 1, comparing the
locations of facilities with active air
releases as recorded in the TRI relative
to 2000 census tracts in the Bay Area
ranked by percent people of color.
The Bay Area in this case refers to the
nine counties covered by both the
Association of Bay Area Governments
(ABAG) and the Bay Area AQMD;
for ease, we focus the map on the
more populated sections of the region
which include the larger cities of San
Francisco, Oakland, and San Jose.
The visual correlation between the
percentage people of color and TRI
Using the same geographic comparison, Latinos
location is striking. But since appearances might
are more than twice as likely and Asians are slightly
be deceiving, Figure 2 provides a demographic
more likely to be living with one mile of a TRI.
breakdown of the populations by proximity to
Is this pattern just a function of income, land use,
a TRI release. The three population groups we
or other factors? Table 1 shows the breakdown
examine are: those communities within one mile of
for our three proximity categories of a number of
a TRI facility, those between one and two and a half
variables, including the poverty rate, per capita
miles from a TRI facility, and those located further
income, the level of home ownership, the percent
than two and a half miles from a TRI facility with
of land devoted to industrial, commercial, and
active releases.
transportation uses, and population density. We
As can be seen, the percent Anglo in a tract declines
also include two other demographic variables:
at closer proximity to a TRI facility. By contrast,
the percent of local employees employed
African Americans are three times more likely to
in manufacturing and the percent of recent
live within one mile of a TRI as they are to live
immigrants.
more than two and a half miles away from a TRI.
As can be discerned from the
table, there is an income gradient,
Figure 2: Population by Race/Ethnicity (2000) and Proximity to a TRI Facility with Air
Releases (2003) in the 9-County Bay Area
with increased proximity and lower
100%
4%
4%
4%
Other
incomes highly correlated. Likewise,
17%
20%
21%
home ownership, a standard measure
80%
4%
of wealth, is lower in the more
Asian/Pacific Islander
12%
8%
12%
proximate neighborhoods. Nearer to
60%
TRIs, a greater percentage of land
21%
African American
tends to be devoted to commercial,
30%
industrial, and transportation uses
40%
63%
Latino
and the percent of the local labor
Percentage of population
force engaged in manufacturing,
45%
20%
33%
an indirect indicator of industrial
Non-Hispanic White
land use frequently employed in the
0%
research literature, tends to be higher
within 1 mile
1 to 2.5 miles
more than 2.5
miles away
as well. Population density is lower
Proximity to an active TRI
in neighborhoods that are closer to
–6–

TRI facilities, something that is partly a
Table 1: Demographic and Land Use Characteristics of Tracts (2000) in
function of the fact that in neighborhoods
Relation to Proximity to an Active TRI Facility (2003) in the 9-County Bay Area
hosting the type of industrial facilities that
TRI Proximity
report to the TRI, some land is devoted
to non-residential uses. Final y, ﬁgures for
Between 1 More than
immigrants who arrived in the 1980s and
Less than 1 mile and 2.5 2.5 miles
mile
miles
away
1990s indicate that they are twice as likely
Percent persons in poverty
12%
9%
6%
to live within one mile of a TRI as to live
Median per capita income
$19,702
$25,140
$34,187
more than 2.5 miles away, suggesting that
Percent home owner
52%
57%
61%
part of the ethnic disparity for Latinos and
Percent industrial, commercial and
Asians might be connected to immigration
transportation land use
17%
9%
5%
status, language ﬂuency, and other factors.
Population density (persons per square mile)
9,202
10,107
9,748
Percent employed in manufacturing
19%
16%
12%
The major question addressed in the
Percent recent immigrants (1980s and later)
26%
21%
15%
research literature, however, is whether
the racial disparities depicted in Figure
2 disappear once researchers control for income.
uses with industrial uses, and does not distinguish
Figure 3 shows that they do not: using the one mile
their relative percentages. While the inclusion
break and plotting the income levels of groups,
of commercial and transportation work wel for
we see that the likelihood of being near a TRI
pol ution burden measures that include mobile
facility declines as income rises (and so does the
and stationary sources, it is problematic for the
disparity between groups). However, there is a
more industrial uses associated with TRI’s. In our
racial disparity in proximity at each and every level
Los Angeles studies, we were able to utilize more
of income.
precise measures of land use and so could separate
and test industrial land use on its own; here, we are
One way to consider the separate impacts
forced instead to use a standard indirect measure,
of the various factors – income, land use,
the percent of the local labor force that is employed
population density, race, etc. – is through what
in manufacturing.
is termed multivariate statistical analysis. Such
an approach helps to isolate whether increases
The results are shown in Table 2. To simplify
in one measure, holding al the others constant,
matters, the table displays the sign of the
aﬀects the probability of a neighborhood being
relationship between the demographic and land use
proximate to a TRI. We do this, taking into
variables on one hand and TRI facility proximity
account homeownership, income, percentage of
on the other. The asterisks indicate whether results
manufacturing employees, population density, and
are statistical y signiﬁcant (more asterisks indicate
racial composition of the neighborhood.
higher signiﬁcance and statistical signiﬁcance,
as usual, is measured as the likelihood that the
One caveat is in order. The land use measure we
reported signs are in error – hence, the “lower”
have available for the Bay Area is imperfect as it
the signiﬁcance threshold, the better). As can be
combines commercial and transportation land
seen in the ﬁrst column, the pattern revealed by
the simple comparisons in
Figure 3: Percentage Households within One Mile of an Active TRI (2003) by Income and
Figure 2 and Table 1 general y
Race/Ethnicity in the 9-County Bay Area
50%
Asian/Pacific Islander
hold in a multivariate analysis:
home ownership, income,
African American
and population density are
40%
Latino
negatively correlated with
Non-Hispanic White
proximity to a TRI facility,
30%
while our proxy for industrial
land use is positively associated
20%
with proximity to such a facility.
Percentage of Households
Even control ing for al these
factors, African Americans and
10% <$10K $10K- $15K- $25K- $35K- $50K- $75K- >$100K
Latinos are signiﬁcantly more
$15K $25K $35K $50K $75K $100K
likely to be near a TRI; Asians
Household Income
are not, although the result is
not statistical y signiﬁcant.
–7–

Given the patterns on immigration observed
Table 2: Multivariate Correlates of Neighborhood Proximity to an Active
earlier, we were curious whether more recent
Toxic Release in the 9-County Bay Area (Proximate = Within 1 Mile)
migrants were more or less likely to be proximate
Coeff. Stat.
Coeff. Stat.
to TRI releases once we control ed for other
Model variables
Sign
Sig.
Sign
Sig.
factors. Since the percent immigrant is highly
% owner occupied housing units
-
**
-
correlated with percent Latino and Asian, we
ln(per capita income)
- *** - ***
decided to separate out what might be a more
ln(population density)
-
**
-
**
meaningful measure in terms of communication:
% manufacturing employment
+ *** + ***
“linguistic isolation.” Linguistic isolation is a
% African American
+ *** + ***
measure developed by the Census by which a
% Latino
+ *** + **
household is designated isolated if no household
% Asian/Pacific Islander
-
-
member older than 14 speaks English “very
% linguistical y isolated households
+
*
wel .” In the second column of Table 2, we see
* indicates significance at the .10 level;
that linguistic isolation does matter – that is,
** indicates significance at the .05 level;
*** indicates significance at the .01 level
N = 1,403
N = 1,403
there is an eﬀect of limited English language
capacity even when control ing for al other
households. Stil , this analysis only covers the
variables. Moreover, the statistical signiﬁcance of
sort of stationary sources recorded in the TRI
percent Latino fal s somewhat when control ing for
– ones that have been important to community
language, suggesting that outreach to this group in
groups like the West County Toxics Coalition,
a native language might be especial y important for
Communities for a Better Environment and others
both conveying information and al owing groups
which have struggled to clean up the activities of
to mobilize to have their concerns expressed.
local reﬁneries. What about the truck and other
However, the bottom line is that the race eﬀect
traﬃc emission sources that have given rise to deep
does not disappear – accounting for income
concern in neighborhoods like West Oakland, San
and other factors does not eliminate the pattern
Leandro, the Mission District of San Francisco,
of disparity for African American and Latino
and elsewhere in the Bay?
Ambient Air Toxics and Estimated Health Risks
To look at this, we turn to the National Air Toxics
excess cancer risks from ambient air toxics that can
Assessment (NATA). As noted above, NATA
be compared to our earlier map of demographics
includes ambient air toxics concentration estimates
(Figure 4).
from large industrial facilities as wel as smal er area
and mobile emission sources. This is important
The observant reader wil note that air toxic risk
because the largest proportion of estimated cancer
does not seem to be an equal opportunity aﬀair:
risk from ambient air toxics – over 70 percent in
there are higher levels of risk in Richmond and
the Bay Area – are related to mobile emissions.
West Oakland, as wel as parts of San Francisco
Of course, this fact does not diminish the need to
and East San Jose, al heavily minority areas. Stil ,
address stationary sources: as the previous section
air quality is a chal enge confronting the whole Bay
shows, facility-based emissions are unevenly
Area: even the so-cal ed “low” risk areas have an
distributed and can be the main environmental
average level of estimated cancer risk which is an
health concern in certain communities. Stil , the
order of magnitude above the goal of ten cancers
NATA data on underlying emissions al ows us to
in a mil ion used by BAAQMD in regulating new
oﬀer a more complete picture of both cancer risks
facilities. This suggests the general importance of a
and respiratory hazard.
cumulative approach: a little risk here, a little risk
there, and soon you have health risks that are wel
Visual y understanding the pattern of this data
above those benchmarks that trigger regulatory
with respect to race is more chal enging than with
concern.
the TRI – we cannot plot single facility points
against demographic breaks since we are also
Once again, we dig deeper into the distribution of
breaking the neighborhoods up in groups ranging
burdens by breaking up Bay Area neighborhoods
from those least-aﬀected to those most-aﬀected.
(or census tracts) by their degree of both cancer
Because of this, we simply show a Bay Area map of
risk and respiratory hazard; fol owing the general
breaks in the map, we designate “least risk” areas as
–8–

those that are more than one standard
Figure 4: 1999 NATA Estimated Cancer Risk (All Sources) by 2000 Census Tracts,
deviation lower than the Bay Area
9-County Bay Area
average, and “most risk” areas as those
that are more than a standard deviation
above this average. We then compare
demographic, income, and other
characteristics, for the communities
that fal into each of these categories in
Table 3.
The table indicates that the higher
risk areas have higher proportions of
minority and immigrant residents.
They have a higher percentage of land
devoted to industrial, commercial, and
transportation land uses, and have a
lower level of home ownership. Poverty
rises slightly as we go from the “least
risk” to the middle range areas and then
doubles in the “most risk” areas; median
per capita income fol ows a similar
trajectory and is much lower in the “most risk”
hazard from ambient air toxics wil also rise,
areas. This suggests that the “least risk” areas are
mostly because it is associated with more economic
not necessarily the most aﬄuent in the Bay Area,
activity; and a subsequent negative eﬀect in which
but one should note that they are more stable, with
higher incomes eventual y provide a defense, either
relatively high rates of home ownership.
economical y or political y against higher levels of
pol ution (this is the pattern indicated in the broad
As before, a ful consideration of the importance of
breaks in Table 3 and it technical y means that the
various measures requires that we do a multivariate
second variable is entered as a square of the ﬁrst).
analysis. The speciﬁcation for this is quite similar
Second, because ambient air toxics are related
to what we used when examining the TRI pattern,
to al of the various industrial, commercial, and
but with a few modiﬁcations. First, we enter
transportation uses, we can utilize that land use
income with two eﬀects: an initial positive eﬀect at
variable in this exercise. Third, because we assume
the very lowest levels of income in which we expect
that since transit uses and commercial activity rise
that as income rises, cancer risk and respiratory
with population, population density is assumed to
be positively correlated with ambient
Table 3: Demographic and Land Use Characteristics of Census Tracts by Estimated
Cancer and Non-Cancer Risk Category
air toxics.
Cancer Risk
Respiratory Hazard
The results are shown in Table 4.
Again, even after control ing for
Lowest
Highest
income and other factors, race seems
Middle
hazard
Middle
hazard
Least risk
range
Most risk
ratio
range
ratio
to matter both for our measures
Percent Anglo
68%
48%
39%
66%
49%
33%
of cancer risk and our measure of
Percent African American
4%
7%
16%
5%
6%
16%
respiratory hazard. In the analysis in
Percent Latino
17%
20%
17%
18%
19%
24%
which we add linguistic isolation, we
Percent Asian Pacific Islander
7%
21%
24%
7%
22%
23%
ﬁnd it to be important for the cancer
Percent Other
4%
4%
4%
4%
4%
4%
risk variable, but very insigniﬁcant
Percent home owner
70%
61%
28%
71%
59%
34%
– indeed, the eﬀect is as nearly close
Median per capita income
$28,231
$28,187
$22,973
$27,137
$29,329
$20,487
to zero as is possible in such tests
Percent persons in poverty
7%
8%
15%
7%
8%
15%
– for respiratory hazards, partly
Population density (persons
because it is so highly correlated with
per square mile)
2,929
8,175
24,194
2,603
9,346
19,425
percent Latino and percent Asian.
Percent industrial,
commercial and
transportation land use
3%
8%
17%
4%
8%
20%
The bottom line is simple.
Percent recent immigrants
Considering either the hazard
(1980s and later)
10%
21%
24%
10%
21%
26%
from exposure to nearby TRI
–9–

toxic air releases, or the
Table 4: Multivariate Correlates of Estimated Cancer and Non-Cancer Risk from Air Toxics in the
level of estimated cancer
9-County Bay Area
risk and respiratory
Cancer Risk
Respiratory Hazard
hazard from air toxics
concentrations estimated
Coeff. Stat. Coeff. Stat.
Coeff. Stat. Coeff. Stat.
in the NATA data, there
Model variables
Sign
Sig.
Sign
Sig.
Sign
Sig.
Sign
Sig.
is a general pattern of
% owner occupied housing units
- *** - ***
- *** - ***
environmental inequity
relative per capita income (tract/state)
+ *** + ***
+ *** + ***
in the Bay Area: densely
relative per capita income squared
- *** - ***
- *** - ***
populated communities
ln(population density)
+ *** + ***
+ *** + ***
of color characterized
% industrial/commercial/transportation
land use
+
by relatively low wealth
*** + ***
+ *** + ***
% African American
+
and income and a larger
*** + ***
+ *** + ***
% Latino
+
share of immigrants,
*** + **
+ *** + ***
% Asian/Pacific Islander
+
disproportionately bear the
*** + ***
+ *** + ***
% linguistical y isolated households
+ ***
-
hazard and risk burden for
* indicates significance at the .10 level;
the region.
** indicates significance at the .05 level;
*** indicates significance at the .01 level
N = 1,402
N = 1,402
N = 1,402
N = 1,402
Potential Policy Implications
Although it is important to examine and document
impacted communities, and localized interventions
environmental disparities, the true chal enge facing
to reduce emissions from problematic large and
the Bay Area is developing and implementing
smal area sources.
strategies to minimize inequalities and decrease
exposures to potential environmental hazards for
In thinking through additional new approaches for
al residents.
addressing environmental inequalities in the Bay
Area, we would suggest several guiding principles.
Fortunately, there are numerous community groups
engaged in discussions and debates with regulators,
The ﬁrst is the need to consider cumulative impacts.
business leaders, and others about adopting more
In the analysis above, we ﬁrst overlaid one set of
health-protective approaches. As noted earlier,
hazards, the location of the Toxic Release Inventory
persistent advocacy eﬀorts of several Bay Area
facilities, and found inequities by race and income.
environmental justice organizations persuaded
We then took a more comprehensive database that
the Bay Area Air Quality Management District
includes mobile sources and health risk estimates,
to adopt a new ﬂare control rule for reﬁnery
the National Air Toxics Assessment, and found
operations. This rule prohibits routine ﬂaring
a similar pattern. Further analysis with state of
by reﬁneries and requires them to prepare Flare
California data charting the location of chrome
Management Plans for each ﬂare at their facilities
platers, hazardous waste sites, and other local y
that speciﬁcal y outlines steps they have taken and
undesirable land uses, although not shown in this
can take to reduce the frequency and duration of
report, conveys the same message: environmental
ﬂaring events. Reduction of air toxics emissions
inequity is alive and wel in the Bay Area.
from these large reﬁneries due to ﬂaring wil
Such analyses also reveal that communities are not
directly beneﬁt the region, particularly fenceline
simply impacted by one set of air releases or one
communities living next to these facilities.
type of hazard, but by several, and some of these
Similarly, advocacy and organizing eﬀorts have
may accumulate and interact to impact community
been successful at drawing attention and support
health in ways that are poorly understood. While
from business, unions, and the public about
our traditional approach to regulating air quality
the need to reduce diesel emissions from trucks
stil tends to be site-by-site and source-by-source,
idling at the Port of Oakland. In San Francisco,
mounting evidence strongly suggests that it is
adoption of the precautionary principle has helped
time for a more comprehensive neighborhood-
to encourage initiatives that promote toxics use
based approach that considers al pol ution sources
reduction, environmental health assessments in
when permitting new facilities or deciding where
–10–

to focus resources
on environmental
remediation and
emissions reduction
eﬀorts. The failure to
consider the cumulative
impacts from multiple
sources and the factors
that enhance community
vulnerability to the
adverse eﬀects of
pol ution exposures
may undermine the
fundamental regulatory
mission to protect public
health.
While a cumulative
approach to regulatory
decision-making
might seem abstract
On October 18, 2005, the Bay Area Ditching Dirty Diesel Collaborative distributed more than 8,000
and untried, it is
informational anti-idling fliers to diesel truck and bus drivers, as well as local residents.The California
Air Resources Board (CARB) held a hearing two days later in which they closed a loophole that
quite relevant to
allowed truckers with sleeper cabs to idle their vehicles overnight, effective 2008.
the contemporary
chal enges facing the
impediments are ﬁnding themselves confronted
state. Both Southern and Northern California
with the worst environmental conditions in the
are facing environmental pressures as a result of
region. And the health consequences are quite
state and regional decisions to pursue economic
real: in a separate analysis, we have found that the
revitalization through the rapid expansion of the
respiratory hazards we have modeled are highly
goods movement industry – that is, the logistics
correlated with hospitalization for asthma, even
of shipping products to and from our ports and
after control ing for the other factors.
sending them via rail and truck to the rest of the
Taking social vulnerability and cumulative
country. While this business strategy might ensure
exposures into account could mean developing
the Bay Area’s preeminent position in a globalized
strategies for determining which neighborhoods
economy and the role of Los Angeles as an entry
might need special outreach, regulatory protection,
for U.S. trade, it creates the real possibility
or engagement in the policy and rule-making
for worsening air quality and community
process. For example, regulatory strategies such
environmental health risks in places like West
as targeted air quality monitoring could be
Oakland in the north and the Alameda Corridor in
enhanced in segregated neighborhoods where
the south. Regulating as though al neighborhoods
poor air quality is a particular concern. Similarly,
mattered would encourage a search for alternatives
this information could help communities and
to diesel fuels, stricter regulations on truck and
local agencies understand how to target their
train idling, and remediation of local point sources
eﬀorts to reduce emissions from major sources.
as major transportation corridors experience
These targeted monitoring and emission source
increases in truck traﬃc volume.
reduction strategies should be done in partnership
A second guiding principle involves the need
with communities who could play a critical role in
to consider social vulnerability. The analysis
helping to identify smal er emissions sources that
above suggests that environmental hazards have
typical y fal below the regulatory radar screen but
over time tended to gravitate to places with the
that may be located near sensitive receptors (e.g.,
least economic, social and political power. This
residential communities or schools). Communities
is the most striking of contradictions from a
can also help agencies balance the need for
health perspective: those residents least likely to
more eﬀective regulation with the promotion of
have access to adequate health care because of
economic opportunities within a region. Previous
income shortfal s, language barriers, and other
agency/community col aborations of this sort
–11–

include monitoring and source reduction eﬀorts
The distance between the stated intentions of
conducted by the California Air Resources Board
the BAAQMD and the views of activist residents
and the communities of Barrio Logan in San Diego
points to a gulf that needs to be overcome. Getting
and Wilmington in the Los Angeles area.
to common ground wil require more discussions
Some might worry that paying attention to
and more col aborations. One way that might help
both cumulative exposures and social conditions
would be the development of real community-
would essential y become a red light for economic
based participatory research projects. While the
activity in the region. But this view oversimpliﬁes
BAAQMD has been crafting new databases of
perceived tradeoﬀs between environmental
modeled emissions, communities have been
integrity and business vitality, and promotes an
out trying to col ect ground-level data on the
outdated understanding of the sophistication of
conditions in their neighborhoods. Surely these
community-based organizations. Indeed, special
eﬀorts can be brought together, a process that
attention to environmental and social justice can
would generate more trust in both the data and the
actual y help the economy: a landmark community
good wil of policy makers.
beneﬁts agreement around the expansion of the
A ﬁnal guiding principle should be meaningful
Los Angeles International Airport set aside $500
action. Perhaps the most important frustration
mil ion for noise abatement for homes and schools
expressed by community members is simple:
and provided job training for local residents even as
discussions of community engagement and the
it facilitated a go-ahead for the project. The point
need for better data col ection drag on while
of addressing environmental inequities is not to
their children remain at risk from toxic air and
stop economic growth but rather to balance costs
local health hazards. There is certainly a need for
and beneﬁts in ways that work best for al aﬀected
better data, better science, and better methods
communities within a region.
to document disparities – and we hope that this
A third guiding principle should be promoting
report wil contribute to a discussion about these
meaningful community participation. Such
issues. How the regulatory community should
participation means involving communities
address fundamental socioeconomic drivers
and their representatives at appropriate points
of environmental health also remains an open
in research and decision-making. It means
question. Final y, the capacity of environmental
understanding that capacities are uneven, and that
and public health agencies to proactively engage
community groups may need additional technical
with these issues is somewhat constrained by
training, information, and partners to insure that
legislative mandates that structure the priorities
their views on environmental health issues are
of their research, regulatory, and enforcement
articulated and understood. And it means reaching
activities.
out in languages that are accessible: not only
Yet agencies that conduct research can begin
Spanish, Chinese, Laotian, and other tongues, but
to grapple with how to integrate place-based
also by translating and communicating scientiﬁc
inequality measures and neighborhood-level
information and research results in formats
indicators of socioeconomic status (SES) with
that community organizations can leverage and
the individual-level factors that have traditional y
disseminate to their constituents.
commanded regulatory attention. Moreover,
although causal y linking the presence of
The Bay Area Air Quality Management District
environmental pol ution with potential y adverse
(BAAQMD) has stated their commitment to
health eﬀects is an ongoing chal enge, particularly
community participation and environmental
in situations where diverse populations are
justice but community activists remain skeptical.
chronical y exposed to complex chemical mixtures
They argue that information and outreach in
from various sources, the ongoing quest for better
multiple languages is available only under political
data and unequivocal proof of cause and eﬀect
pressure and not as a matter of course. And they
should not make us lose sight of a basic public
suggest that the recent modeling of air toxics under
health principle— namely, the importance of
the District’s Community Air Risk Evaluation
disease prevention. This requires regulatory and
(CARE) program wil not provide enough details
land use planning agencies to work proactively
for neighborhood-level analysis of environmental
with communities to constantly seek and develop
exposures and inequities.
opportunities for emissions source reduction that
–12–

can improve air quality for speciﬁc neighborhoods
are broader than traditional indicators of “sensitive
and the region as a whole.
receptors” (i.e. children and the elderly). These
BAAQMD could make signiﬁcant steps toward
measures should include socioeconomic status,
embracing the four principles we outline above by
access to health services, community capacity for
implementing some very speciﬁc short- and long-
civic engagement, and information on the incidence
term regulatory initiatives. In the short-term, the air
of health outcomes that are linked to both the
district can proactively facilitate more open access
social and physical environment of neighborhoods.
to critical data sets and emissions inventories that
Community participation in this process wil
enable communities to assess environmental health
be critical to the development of policy-relevant
issues as wel as identify potential y problematic
and transparent indicators of vulnerability and
emissions sources that require community and
cumulative impact.
regulatory intervention. Even more important,
Eﬀective regulatory and policy initiatives that
the BAAQMD could col aborate more with cities,
advance environmental justice wil require
counties and public stakeholders to expand its
combining local and regional approaches to
inventories to include unregulated sources that may
data col ection, air quality monitoring, analysis,
be contributing signiﬁcantly to local cumulative
dissemination of results, and, most important,
environmental health risks.
regulatory intervention. In some instances regional
Critical to this process wil be eﬀorts to identify
activities aimed at emission source reduction
the advantages and limitations of al emissions
can result in signiﬁcant decreases in certain
inventories and data sets that the BAAQMD
pol utant burdens. But this regional work must be
develops and disseminates. For example, most
complemented by local initiatives that combine
inventories are based on estimates of emissions and
emissions reduction incentives and in some cases
not actual measurements, and they tend to only
outright caps in those neighborhoods that are
capture pol ution emissions from “normal operating”
already severely aﬀected by high pol ution levels
conditions and not episodic events that may occur
from myriad mobile and stationary sources. These
due to industrial accidents, or an unusual ramping
localized regulatory activities can be leveraged
up of activity due to construction or the expansion
through col aboration with other agencies to
of an existing facility. Moreover, it wil be important
problem-solve environmental health issues associated
to consider emissions from “magnet” sources (e.g.
with land use, industrial development and zoning
ports and terminals), grandfathered sources, and
decisions.
uncontrol ed emissions sources that could be either
Final y, as the Bay Area undergoes an impressive
permanent or temporary (such as construction
development and construction boom in what
activities). In this way, communities can work more
were once low income communities, the issue
closely with the air district to identify activities
of gentriﬁcation wil have to be forthrightly
that may fal below the “regulatory radar” and not
addressed. On the one hand, environmental
get captured in existing inventories but that may
clean-up and exposure reduction activities should
actual y contribute signiﬁcantly to
local pol ution burdens.
Over the longer term the
air district should develop a
cumulative impact approach
in its regulatory activities and
permitting decisions. This wil
require rethinking traditional risk
assessment in ways that take into
account emissions from multiple
pol utants coming from multiple
sources that can have both
localized and regional impacts.
Similarly, the air district wil have
to develop a more open, and
deliberative process to develop
Youth from southeast San Francisco work with PODER and the Chinese Progressive
Association to organize a community media event highlighting the results of a community
measures of vulnerability that
survey and grassroots community planning process.
–13–

not simply be targeted toward so-cal ed “up and
– does not wind up displacing the residents whose
coming” neighborhoods, but aimed at improving
concern and activism prompted the clean-up. Those
environmental quality for al residents in the Bay
who have suﬀered through the toxic soup for many
Area. On the other hand, proactive measures wil
years should be among those to reap the rewards
need to be taken to insure that any increase in area
from a new commitment to the environmental
attractiveness because of environmental remediation
quality in general and environmental justice in
– which is likely through this to raise property values
particular.
Looking Forward
The Bay Area has often prided itself on leading
continue to press for cleaner air for all the region’s
the state on environmental issues. When the
residents. We are host to some of the country’s
state passed a Global Warming Solutions Act in
most dynamic companies, ﬁrms rooted in the new
September 2006, the signing ceremony was staged
economy and therefore aware that improving the
on Treasure Island. When the state authorized a
environment represents not a business drag but a
new biomonitoring program in the same month
business opportunity. And we have among us some
– an approach that will allow us to know rather
of the country’s best researchers and scientists,
than guess about pollutant exposures – Bay Area
people who can bring data and tools from our
legislators were at the forefront.
academic and public sector institutions to analyze
Yet this study demonstrates that the Bay Area is
problems and suggest solutions.
also characterized by an unequal distribution of our
The task is to marshal these resources for a new
environmental burdens and opportunities – and
direction and new partnerships that will take social
no amount of sophisticated statistical attempts to
equity as a serious prism for understanding and
control for other factors seems to erase the stain of
improving the environment. The challenge is to
racial and economic inequality.
consider cumulative impacts in a new regulatory
We can and should do better.
approach, building on the wisdom of communities
who know that it is not just one sort of hazard or
We enjoy, after all, the presence of some of the
release that threatens their health. And while we
country’s most vibrant and creative community-
fully understand the need to deepen the research
based environmental organizations, groups that
and consider the complexities, the time for action
have generated their own research eﬀorts and
is now.
–14–

Technical Appendix
In order to have the bulk of this report as accessible
cancer risk from pollutant j in census tract i, C ij
as possible, we have chosen to conﬁne certain
is the concentration in micrograms of pollutant
technical details to this appendix. Here, we discuss
per cubic meter of air (µg/m3) of the air toxic j
in more detail data sets and variables as well as
in census tract i, and IUR is the inhalation unit
j
techniques.
risk estimate for pollutant j. In accordance with
In our consideration of facilities listed in the Toxic
California’s AB2588 “Hot Spots” Guidelines and
Release Inventory (TRI), we examined only those
EPA’s cancer risk guidelines, cancer risks of each
with active air releases; the TRI itself includes
pollutant were assumed to be additive and were
numerous facilities that are not currently generating
summed together in each tract to derive a total
air emissions and we exclude these. Active facilities
individual lifetime cancer risk.
were located using address-matching (geocoding)
Respiratory hazard was derived by comparing
of the street address reported to EPA against the
each pollutant concentration estimate by its
address ranges in high quality spatial data sets
corresponding Reference Concentration (RfC)
recording roads and street statewide. To check for
to derive a hazard ratio. An RfC for chronic
location accuracy and possible errors, each facility
respiratory eﬀects is deﬁned as the amount of
address was located using current versions of two
toxicant below which long-term exposure to
diﬀerent street databases, one from TeleAtlas and
the general population of humans, including
one from Geographic Data Technology, two of the
sensitive subgroups, is not anticipated to result in
most reliable data providers available.
any adverse eﬀects. The actual respiratory hazard
The basic unit for neighborhood analysis for both
ratios for each pollutant in each census tract were
the TRI and NATA analysis (see below) was the
calculated using the following formula:
census tract, a standard in both demographic and
HR = C /RfC
ij
ij
j
environmental analysis. To determine which tracts
were proximate, we drew, as noted in the text,
where HR is the hazard ratio for pollutant j
ij
various radii; if half of a tract’s population fell
in tract i, C is the concentration in ug/m3 of
ij
within each resulting circle, as measured by the
pollutant j in census tract i, and RfC is the
j
populations of the census blocks that fell within
reference concentration for pollutant j in ug/m3. An
the circle, we considered that tract to be aﬀected.
indicator of total respiratory hazard was calculated
This procedure is a slight modiﬁcation of the best
by summing together the hazard ratios for each
practice procedure in the case of stationary sources
pollutant in order to derive a total respiratory
as laid out in a recent article by Mohai and Saha
hazard index:
(2006). The radii utilized were one mile and two
HI = Σ HR
i
j
ij
and a half miles; we also tightened the focus to look
where HI is the sum of the hazard ratios for all
at the area within one half mile of a TRI but the
i
pollutants (j) in census tract i.
results were quite similar to the demographics for
the one mile radius, and so we present just the three
For all of the analysis presented above, we
breaks (with one mile, between one and two and a
used estimates of cancer risk and respiratory
half miles, and beyond) in the text.
hazard that were based on the Hazardous Air
Pollutant Exposure Model (HAPEM). HAPEM
The National Air Toxics Assessment data is brieﬂy
integrates ambient concentration estimates with
explained in the text and more on the 1999
information on indoor/outdoor microenvironment
iteration can be found at http://www.epa.gov/
concentrations, penetration of outdoor pollutants
ttn/atw/nata1999/. To calculate cancer risk, we
into indoor environments, local populations,
combined air toxics concentration estimates with
and individual-level activity patterns to generate
inhalation unit risk estimates for each carcinogenic
an expected range of inhalation exposure
compound to estimate overall cancer risks.
concentrations for each census tract before applying
Estimated cancer risks for each pollutant in each
the inhalation unit risk estimates and reference
census tract were derived with the formula
concentrations for each pollutant to obtain the
R = C x IUR ,
ﬁnal estimates. HAPEM tries to take into account
ij
ij
j
where R is the estimate of individual lifetime
more realistic scenarios of people’s day-to-day
ij
–15–

activities and the fact that they tend to move
which unweighted means are reported. This is done
across various locations (e.g., from home to work,
because the variation in land area of census tracts
or home to school) and spend a majority of their
(which would be the appropriate weight for the two
time in indoor environments (e.g., the home,
variables mentioned above) is so great that applying
workplace, school, or commuting in a car). Our
such a weight would greatly distort the view of a
analytical results are nearly identical when the
typical census tract falling into each category. Such
basic concentration-based estimates are used, but
a distortion by the use of population or household
the HAPEM-based exposure results are generally
weights does not occur for the other variables.
considered better at accounting for diﬀerences in
For the multivariate analysis of TRI location,
estimated population exposures.
we used a logit regression; this is an appropriate
As noted in the text, these are modeled cancer
strategy for a case where the dependent variable
risks and respiratory hazards. Still, we should stress
consists of two possibilities, either being proximate
that in a set of multivariate regressions separate
(say, less than one mile away) or not being
from the current analysis, a signiﬁcant relationship
proximate (say, further than one mile away) to a
was found between the respiratory hazard
TRI. For the NATA analysis, we utilized ordinary
ratio described above and age-adjusted asthma
least squares regressions techniques.
hospitalization rates (taken as three year averages
Since for clarity of presentation, we report just
over the years 1998 through 2000) that were made
signs and coeﬃcient levels, some might wonder just
available to us by Community Action to Fight
how close to signiﬁcance some of those variables
Asthma (CAFA). Although the analysis was carried
that do not obtain traditional signiﬁcance levels
out at the Zip Code Tabulation Area (ZCTA)
might be. In the TRI logistic regression in the ﬁrst
– the level of geography at which the asthma data
column of Table 2, the negative sign for the percent
was made available and one that is far less detailed
Asian and Paciﬁc Islander has a signiﬁcance level
than the census tract – the signiﬁcance of the
of .864, meaning that it is virtually certain that the
relationship, even when controlling for several
real value is zero. In the second column, it appears
measures that would seem to explain both the
that housing ownership has slipped in signiﬁcance
incidence of asthma and the event of hospitalization
but only to the .125 level. The inclusion of the
for the condition, lends some degree of conﬁdence
statistically signiﬁcant measure for linguistic
to the respiratory hazard ratio we have developed as
isolation raises the signiﬁcance of the negative sign
a measure of health impacts.
for Asian Paciﬁc Islanders, implying that for this
While most of the data comparisons are quite clear
population, linguistic isolation may be especially
in the text, it is useful to mention several things.
important. Finally, in the NATA regressions in
First, in the simple comparisons of tracts by degree
Table 4, readers might be struck by the negative
of cancer risk or respiratory hazard, we talk about
sign for linguistic isolation. However, with a
breaking the data by standard deviations from
signiﬁcance level of .904, it is virtually certain
the mean. The cancer risk and respiratory hazard
that the real eﬀect is null rather than negative.
measures, however, are not normally distributed;
While this is still of interest, a better reading and
by contrast, the natural logs of these values seem
maybe even presentation of that sign would be as a
to be close to a normal distribution. Because of
question mark.
this, we use the logged value as the dependent in
Though also not reported in the tables, the model
our regressions and also in determining the means
ﬁt as measured by the reported Nagelkerke R
and standard deviations that categorize “most”
Square values for the TRI regressions ranged from
and “least” aﬀected tracts for the comparison in
0.2581 to 0.2616; these might seem to be low but
Table 3, as well as in the map that comprises Figure
the ﬁgures are actually quite good for this type
4. This is an approach we have used before in
of regression. The ﬁt for the NATA regressions
several diﬀerent pieces, including most recently in
as measured by the reported Adjusted R Square
Pastor et al. (2005). We should also note that the
values ranged from 0.5632 to 0.5721, which is
statistics reported for the comparisons drawn in
a very good performance. In both the TRI and
Tables 1 and 3 are weighted means of each variable
NATA analysis, population density is entered as
across all tracts falling into each category, applying
a log on the grounds that its eﬀect diminishes
the appropriate weight, with the exception of
at higher values; in the TRI analysis, we entered
population density and the share of land devoted
per capita income as a log for the same reason. In
industrial, commercial and transportation uses, for
the NATA analysis, however, we utilize per capita
–16–

income in a U shaped formula – that is, we enter
be in a tract in order to generate a risk estimate.
the normalized value of per capita income (relative
Similarly, for the same tract in the Figure 2, we
to the state value) and the square of that value
assigned to it the percent people of color category
– because of our assumptions about the shape
that best reﬂected the surrounding neighborhoods.
of the relationship at lower and higher levels of
Also in Figure 2, the demographic breaks were
income. There was no attempt to control for spatial
derived by ranking all Bay Area tracts into thirds
autocorrelation although we intend to do that in
according to the percentage people of color. Thus,
future analyses.
the numbers reported in the legend are tertiles,
Finally, regarding the two maps presented, we
and they have been rounded to the nearest whole
should note that because HAPEM estimates of
percentage point for convenience.
cancer risk and respiratory hazard are only made
For those interested in more detail on these various
for census tracts in which people reside, and there
techniques, we recommend our various analyses of
is one tract in the Bay Area that contains no people
Southern California, especially Sadd, et al. (1999)
(the San Francisco International Airport) and hence
and Morello-Frosch, et al. (2002, 2001). For those
has no available cancer risk estimate to include in
interested in the general empirical debate about
Figure 4, rather than dropping it from the map we
environmental justice disparities, we suggest United
assigned to it the distributional category it would
Church of Christ (1987), Anderton, et al. (1994),
receive under the basic concentration-based cancer
Lester et al. (2001), and Ash and Fetter (2004).
risk estimate, which does not require people to
References
Anderton, Douglas L., Anderson, Andy B.,
J. 2002. “Environmental Justice and Regional
Oakes, John M., and Fraser, Michael R. 1994.
Inequality in Southern California: Implications for
“Environmental equity: The demographics of
Future Research.” Environmental Health Perspectives,
dumping.” Demography 31 (2):229-248
2002, 110 (Supplement 2): 149-154.
Ash, Michael and T. Robert Fetter. 2004. “Who
O’Rourke, Dara and Gregg Macey. 2003.
Lives on the Wrong Side of the Environmental
“Community Environmental Policing: Assessing
Tracks? Evidence from the EPA’s Risk-Screening
New Strategies of Public Participation in
Environmental Indicators Model.” Social Science
Environmental Regulation.” Journal of Policy
Quarterly 85(2): 441-462.
Analysis and Management, vol. 22, no. 3, summer,
Denny Larson. 2005. Breathing Fire, in their own
pp. 383-414
words. San Francisco, CA: Global Community
Palaniappan, Meena, Diana Wu, and Jacki
Monitor. http://www.reﬁneryreform.org/
Kohleriter. 2003. Clearing the Air: Reducing Diesel
downloads/GCMFlareReport.pdf
Pollution in West Oakland. Oakland, CA: Paciﬁc
Lester, James P., David W. Allen, and Kelly M. Hill.
Institute. http://www.pacinst.org/reports/diesel/
2001. Environmental Justice in the United States:
clearing_the_air_ﬁnal.pdf
Myths and Realities. Boulder, Colorado: Westview
Palaniappan, Meena, Swati Prakash, and Diane
Press.
Bailey. 2006. Paying With Our Health. The Real Cost
Mohai, Paul and Robin Saha. 2006. “Reassessing
of Freight Transport in California. Oakland, CA:
Race and Socioeconomic Disparities in
Paciﬁc Institute. http://www.pacinst.org/reports/
Environmental Justice Research.” Demography 3(2):
freight_transport/PayingWithOurHealth_Web.pdf
383-399.
Pastor, Manuel, Rachel Morello-Frosch, and James
Morello-Frosch, Rachel A., Manuel Pastor, and
L. Sadd. 2006. “Breathless: Schools, Air Toxics, and
James L Sadd. 2001. “Environmental Justice and
Environmental Justice in California.” Policy Studies
Southern California’s ‘Riskscape’: The Distribution
Journal 34(3): 337-362.
of Air Toxics Exposures and Health Risks Among
Pastor, Manuel, James L. Sadd, and Rachel
Diverse Communities.” Urban Aﬀairs Review 36(4):
Morello-Frosch. 2005. “The Air is Always Cleaner
551-578.
on the Other Side: Race, Space, and Air Toxics
Morello-Frosch RA, Pastor M, Porras C, and Sadd
Exposures in California.” Journal of Urban Aﬀairs
27(2): 127-148.
–17–

Pellow, David N. and Lisa Sun-Hee Park. 2002. The
Szasz, Andrew and Michael Meuser. 2000.
Silicon Valley of Dreams: Environmental Injustice,
“Unintended, Inexorable: The Production of
Immigrant Workers, and the High-Tech Global
Environmental Inequalities in Santa Clara County,
Economy. New York: New York University Press.
California.” American Behavioral Scientist, 43(4):
Sadd, James L., Manuel Pastor, Jr., Joel T. Boer,
602-632
and Lori D. Snyder. 1999. ‘‘Every Breath You
United Church of Christ. 1987. A national report
Take . . .’: The Demographics of Toxic Air Releases
on the racial and socio-economic characteristics of
in Southern California.’ Economic Development
communities with hazardous waste sites. New York,
Quarterly 13(2): 107-123.
NY: United Church of Christ, Commission for
Racial Justice.
Selected Web Resources
• American Lung Association (ALA). The ALA
• California Environmental Protection Agency
is the oldest voluntary health organization in
(Cal/EPA). Cal/EPA has as its mission restoring,
the United States, with a National Oﬃce and
protecting and enhancing the environment,
constituent and aﬃliate associations around the
to ensure public health, environmental quality
country; it ﬁghts lung disease in all its forms, with
and economic vitality. To see the strategies for
special emphasis on asthma, tobacco control and
environmental justice in the state, including
environmental health. See http://www.lungusa.org/
recommendations for state agencies, go to: http://
site/pp.asp?c=dvLUK9O0E&b=22542
www.calepa.ca.gov/EnvJustice
• Bay Area Air Quality Management District
• Center for Community Action and
(BAAQMD). BAAQMD is charged with regulating
Environmental Justice (CCAEJ). The goal of
to achieve clean air to protect the public's health
this center is to build a strong movement for
and the environment in the San Francisco Bay
change that recognizes the connections between
region. See http://www.baaqmd.gov
environmental and worker exploitation, and
• Berkeley Center for Environmental Public
oppression on the basis of race, gender, sexual
Health Tracking (BCEPHT). BCEPHT works
orientation and class. See http://www.ccaej.org
to advance a nationwide Environmental Public
• Center for Health, Environment and Justice
Health Tracking (EPHT) network that provides
(CHEJ). This organization works to level the
and communicates information about relationships
playing ﬁeld by allowing communities members
between environmental factors and health to all
to have a say in the environmental policies and
relevant audiences, including policy-makers and
decisions that aﬀect their health and well-being.
community stakeholders. See http://ehtracking.
See http://www.chej.org/index.htm
berkeley.edu
• Communities for a Better Environment (CBE).
• California Air Resources Board (CARB). CARB
CBE is an environmental health and justice non-
is responsible for promoting and protecting public
proﬁt organization, whose unique three-part
health, welfare and ecological resources through the
strategy provides grassroots activism, environmental
eﬀective and eﬃcient reduction of air pollutants
research and legal assistance within underserved
while recognizing and considering the eﬀects on the
urban communities. See http://www.cbecal.org
economy of the state. See http://www.arb.ca.gov/
•
homepage.htm
Community Action to Fight Asthma (CAFA).
CAFA is a network of asthma coalitions in
• California Environmental Health Tracking
California working to shape local, regional and
Program (CEHTP). CEHTP is a multi-agency
state policies to reduce the environmental triggers
collaborative seeking to develop a comprehensive
of asthma for school-aged children where they live,
standards-based, coordinated, and integrated
learn, and play. See http://www.calasthma.org
system, at the state level, that enables public health
•
actions through linkage, monitoring, reporting, and
Environmental Health Coalition (EHC).
communication of health eﬀects and environmental
Dedicated to achieving environmental and social
hazards and exposure data. See http://www.
justice, this coalition believes that justice is
catracking.com
accomplished by empowered communities acting
together to make social change, and supports broad
–18–

eﬀorts that create a just society which fosters a
special interests in the area of environmental justice.
healthy and sustainable quality of life. See http://
See http://www.sﬀ.org/grantmaking/enviro_ehji.
www.environmentalhealth.org
html
• Environmental Inequality. This site contains a
• Transportation and Land Use Coalition (TALC).
report detailing environmental disparities in Silicon
TALC is a partnership of over 90 groups working
Val ey and Santa Clara County. Refer to the maps
for a sustainable and socially just Bay Area with a
comparing toxic sites with income level and racial
focus on analyzing county and regional policies,
compositions. See http://www.mapcruzin.com/EI/
and working with community groups to develop
• Golden Gate University Environmental Justice
alternatives. See http://www.transcoalition.org
and Law Clinic. In 1994, Golden Gate was one
• Urban Habitat (UH). UH builds power in
of the ﬁrst law schools in the country to establish
low-income communities and communities of
an environmental justice clinic. Working under
color by combining education, advocacy, research
the close supervision of two ful -time professors,
and coalition building to advance environmental,
students directly represent environmental
economic and social justice in the Bay Area. See
organizations and community groups in low-income
http://urbanhabitat.org
and minority communities in real-life public
•
health, toxics, and environmental justice matters.
US EPA’s 1999 National Air Toxics Assessment
See http://www.ggu.edu/school_of_law/academic_
(NATA) website can be found at: http://www.epa.
law_programs/jd_program/environmental_law/
gov/ttn/atw/nata1999/
environmental_law_justice_clinic
• US EPA’s 2003 Toxic Release Inventory Program
• Greenaction for Health and Environmental
can be accessed at: http://www.epa.gov/tri/tridata/
Justice (Greenaction). Greenaction mobilizes
tri03/index.htm
community power to win victories that change
government and corporate policies and practices
to protect health and to promote environmental
Photo Credits:
justice. See http://www.greenaction.org
Front cover
• National Resource Defense Council (NRDC).
NRDC works to foster the fundamental right of al
West Berkeley community members and allies
people to have a voice in decisions that aﬀect their
march to protest toxic emissions from a local steel
environment. See http://www.nrdc.org
foundry . . . by Peter Guerrero, November 11,
2006.
• Paciﬁc Institute. The Paciﬁc Institute is an
independent, nonpartisan think-tank studying issues
Page 1
at the intersection of development, environment,
“On April 11, 2006, Bayview Hunters Point
and security. Some of its recent work includes
community resdients...” by Greenaction.
pioneering studies of environmental disparities,
particularly using community-based participatory
Page 2
research. See http://www.pacinst.org
“Organizing for a healthy environment, great
• Reﬁnery Reform. The Reﬁnery Reform
jobs, and healthy foods...” by PODER.
Campaign is a national eﬀort seeking to clean up
Page 3
America’s oil reﬁneries and reduce our dependence
on fossil fuels. See http://www.reﬁneryreform.org/
“In July of 2005, PODER and the Mission Anti
Displacement Coalition...” by PODER.
• Silicon Val ey Toxics Coalition (SVTC). This
diverse organization is engaged in research,
Page 11
advocacy, and grassroots organizing to promote
“On October 18, 2005, the bay area Ditching
human health and environmental justice in response
Dirty Diesel Collaborative...” by Ditching Dirty
to the rapid growth of the high-tech industry. See
Diesel
http://svtc.etoxics.org/
Page 13
• The San Francisco Foundation. The Foundation
seeks to mobilize resources and act as a catalyst for
“Youth from southeast San Francisco work
change to build strong communities, foster civic
with PODER and the Chinese Progressive
leadership, and promote philanthropy. It has had
Association...” by PODER
–19–

Bay Area Environmental Health Collaborative
The Bay Area Environmental Health Collaborative (BAEHC) is a multi-year partnership among six broad
coalitions and numerous organizations working for the adoption of speciﬁc measures to protect public health
in communities that are heavily impacted by air pollution. BAEHC’s diverse membership includes community
representatives, environmental health and justice advocates, scientiﬁc and technical experts and public health
professionals.
The goal of BAEHC is to assure better health outcomes for local residents by improving Bay Area air quality
through community capacity and the establishment of protective public policy measures that reduce the
cumulative eﬀects of air pollution and ensure public access to decision-making processes, particularly in highly
impacted areas. The BAEHC is supported in part by The San Francisco Foundation.
Environmental Justice Air Quality Coalition
La Clinica de La Raza
Greenaction for Environmental Health & Justice
Martinez Uniﬁed School District
Communities for a Better Environment
Mt. Diablo Uniﬁed School District
West County Toxics Coalition
Neighborhood House of North Richmond
Youth United for Community Action
Parents for a Safer Environment
Hunters View Mothers Committee
Shields-Reid Community Center
Healthy San Leandro Col aborative
West County Toxics Coalition
Immigrant Power for Environment Health and
Ditching Dirty Diesel Collaborative
Justice
Bayview Hunters’ Point Community Advocates
Chinese Progressive Association
Center for Community Action and
People Organizing to Demand Environmental and
Environmental Justice
Economic Rights
Coalition for Clean Air
Environmental Law and Justice Clinic
Communities for a Better Environment
Healthy San Leandro Col aborative
Bay Area Clean Air Task Force
International Longshore and Warehouse Union,
American Lung Association of California
Local 10
Bayview Hunters Point Community Advocates
Natural Resources Defense Council
Bluewater Network
Neighborhood House of North Richmond
Breathe California, Golden Gate Partnership
Paciﬁc Institute
Natural Resources Defensive Council
Regional Asthma Management and Prevention
Our Children’s Earth Foundation
Initiative
Regional Asthma Management and Prevention
West Oakland Environmental Indicators Project
Initiative
Sierra Club
Regional Asthma Management Prevention
Union of Concerned Scientists
Initiative
TRANSDEF
Oakland-Berkeley Community Action to Fight
West Oakland Environmental Indicators Project
Asthma
Ethnic Health Institute
Contra Costa Asthma Coalition
San Francisco Department of Public Health
American Lung Association of California
San Francisco Asthma Task Force
Asthma Community Advocates
Communities for a Better Environment
Environmental Law and Justice Clinic
Community Action to Fight Asthma
Community Clinic Consortium of Contra Costa
For more information, please contact:
Community Focus
Amy S. Cohen, Campaign Director
Community Health Initiative
Bay Area Environmental Health Col aborative
Concerned residents of Contra Costa County
c/o Environmental Law and Justice Clinic,
Contra Costa Child Care Council
Golden Gate University School for Law
Contra Costa Community Services
Mail: 536 Mission Street; Oﬃce: 62-First Street,
Contra Costa Council
Suite 522
Contra Costa Health Services
San Francisco, CA 94105-2968
John Muir Health
Tel: 415.442.6656 Fax: 415.896.2450
Kaiser Permanente
E-Mail: acohen@ggu.edu