NCD980602163_19980107_Warren County PCB Landfill_SERB C_A Study of Procedural Environmental Equity in Enforcement, DHEC Consent Orders 1993 - 94-OCR

I I
A STUDY OF PROCEDURAL
ENVIRONMENTAL EQUITY IN
ENFORCEMENT: DHEC CONSENT ORDERS
1993-94
BY
GERRI M. MIRKIN, MSPH AND LILLIAN H.
MOOD, RN, MPH
Introduction
• The civil rights movement traditionally defined civil rights and liberties that should be available
to all citizens of the United States, regardless of race. A new dimension to these guaranteed rights is
what defines the concept of environmental equity or, as its more popularly recognized, environmental
justice. In February 1994, President Clinton signed Executive Order 12898 requiring every federal
agency to achieve the principle of environmental justice by addressing and ameliorating the human
health or environmental effects of the agency's programs, policies and activities on minority and low-
income populations in the US [ 1]. The ideas on ways to approach and study environmental equity differ
based on their origin or dissimilarity in scope. The type of equity this study seeks to explore, in the
setting of the state of South Carolina, is procedural equity, which is defined as the extent to which
governmental rules and regulations, enforcement and international treaties and sanctions are applied in a
nondiscriminatory way [4]. This study focuses on a two year "snapshot" of environmental permit
regulation enforcement in South Carolina, as administered by the five main enforcement units of the
Department of Health and Environmental Control (DHEC). Enforcement takes the form of consent
orders issued by DHEC which contain conditions that must be met and adhered to in order to correct
environmental permit infractions, as well as monetary penalties that must be paid for committing said
infractions. Following a review of the environmental justice movement in recent years, the specific
details of the enforcement process dealt with in this study, as well as the structure and parameters of the
study itself, will be described .
1
Environmental Justice: A Brief Background
The tenn environmental racism was coined in 1982 by Benjamin Chavis [10], then head of the
United Church of Christ's (UCC) Commission on Racial Justice, an organization first credited with
taking a comprehensive look at what it proposed was the deliberate targeting of communities of color
for disproportionate burdens of environmental degradation. He stated:
"Environmental racism is racial discrimination in environmental policy-making and enforcement of
regulations and laws, the deliberate targeting of communities of color for toxic waste facilities, the
official sanctioning of the presence of life threatening poisons and pollutants in communities of color,
and the history of excluding people of color from leadership of the environmental movement [2]."
In that same year, the environmental justice movement began, by most accounts, in Warren
County, North Carolina, when the state selected a site (Afton) to host a hazardous waste landfill
containing 30000 cubic yards of PCB-contaminated soil [6]. Residents, mostly African-American,
rural and poor, were joined in their protests by national civil-rights groups, environmental groups,
clergy and members of the Black Congressional Caucus. Although unsuccessful in halting the landfill
construction, the Warren County demonstrations marked the first time that African Americans
mobilized a national broad-based coalition in response to an impending environmental threat [4]. This
demonstration marked the first of many community of color struggles over toxic substances and became
the initiating event for growing awareness, concern and action dealing with environmentally based racial
issues.
A study published in 1983 by the government (USGAO) [14], examining racial inequities in
communities surrounding four of the largest hazardous waste landfills in the south, as well as a 1987
study by the previously mentioned UCC Commission [ 13] that focused on commercial hazardous waste
2
facilities and uncontrolled waste sites, galvanized the movement and provided some much-needed
empirical support for the claims of environmental racism [4]. An extensive analysis of the issues of
environmental justice was published in Bullard' s Dumping in Dixie ( 1990), which fueled the momentum
of the movement by adding further empirical support for the disproportionate burden of toxic waste on
minority communities.
Among other milestones of the movement, in 1992, the USEP A formally established its Office of
Environmental Equity. That same year the Workgroup on Environmental Equity, also established by
the USEPA, finished its report begun in 1990 [15]. This signature report evolved from the
workgroup's focus on two primary tasks: (1) to evaluate the evidence that racial minority and low-
income groups bore a disproportionate burden of environmental risks; and 2) to identify factors that
contributed to different risk burdens and to suggest strategies for improvement [ 4]. Critics of the report
contend that the EPA did not go far enough in examining its current activities, including its own role in
re-enforcing environmental inequalities through its own decision-making procedures [3,9, 11].
Although a number of bills were introduced into Congress at this time, the most noted was the
Environmental Justice Act 1992, first sponsored by Senator Albert Gore (Tennessee) and Congressman
John Lewis (Georgia). Arkansas and Louisiana enacted the first environmental justice laws in states [ 4].
Finally, as described previously, Executive Order 12898 was signed into law by President Clinton on
February 11, 1994.
3
Focusing on Process Equity: Enforcement Issues
• This study focuses on one specific qualifying factor for the achievement of environmental
justice: that of process equity. Process equity, as discerned from outcome equity, refers to some of the
underlying causes of environmental inequities such as basic social inequities, siting decisions, cleanup or
differential enforcement of laws and regulations. This last cause of inequity is the primary focus of this
study. South Carolina is appropriate for an environmental equity study since it is part of the deep south
- a region in which every state can be found towards the bottom of the list on most of the 256
environmental indicators [4].
The concept behind environmental justice is to ensure that no person is discriminated against
such that he or she has to endure unfair environmental conditions [13]. The unfair environmental
conditions can be something obvious in a person's immediate physical environment, such as lack of
clean drinking water in one's community, or something less obvious, such as the differential
enforcement of environmental regulations dealing with potential polluting industries in that community.
This study examines differences in enforcement within a fixed set of parameters. The hypothesis is that
enforcement of regulations that protect the environment from damage due to pollution from various
industries and businesses, as administered by South Carolina DHEC, is carried out in an inequitable
manner for communities that are primarily made up of economically disadvantaged people and/or people
of color, as compared to communities that are primarily affluent and/or white.
4
Study Design: Building on Previous Analyses
This study borrows heavily, in terms of definition of parameters and implementation, from a
1994 Masters of Science thesis by Danika M. Holm in the Department of Geography at the University
of South Carolina. Holm's thesis tested three hypotheses: 1) hazardous waste treatment, storage and
disposal sites throughout the state of South Carolina are not randomly distributed; 2) a disproportionate
number of sites are located in minority and economically-disadvantaged communities; and 3) inequities
result from original location of hazardous waste facilities in minority communities (7]. The thesis
focused on analyzing outcome equity by specifically addressing whether minority and economically
disadvantaged communities in South Carolina are disproportionately burdened with the State's
hazardous waste facilities [7]. The study involved 79 facilities listed in the State's Hazardous Waste
Management Tracking System (12]. Findings of the study supported hypothesis #1, but did not support
the 2nd and 3rd hypotheses.
As mentioned, this study builds on the method used in Holm's thesis and the environmental
equity parameters defined in it for South Carolina. Specifically, this study utilizes Holm's definitions of
what identifies a community as economically disadvantaged or a minority one in South Carolina. Once
communities are defined as either economically disadvantaged or minority, as well as both or neither of
these, then their status is compared to the manner in which enforcement of environmental regulations is
meted out, by SC DHEC, to potentially polluting facilities located within these communities.
5
The Data Set: DHEC Consent Orders
The enforcement actions administered by DHEC, the focus of our study, are called consent
orders. Consent orders are defined by two parameters for the purpose of this study, the monetary
penalties assessed by DHEC for infraction(s) committed and the total number of days from the initiation
of the enforcement action to its identifiable conclusion or resolution -termed the "enforcement time".
These parameters will be discussed in detail in upcoming sections. This definition of consent orders is
quite simplistic in view of the entire enforcement process that a given facility (henceforth referred to as
the "violator") must go through once they violate environmental requirements of their operational
permit, as established and issued by DHEC. Businesses and industries may be subject to permit
requirements in several program areas such as Air Quality Control, Water Pollution Control and
Hazardous Waste.
There are vast differences in the types of facilities that can have similar permit requirements. An
important fact to note, however, is that businesses can receive consent orders for violating state or
federal environmental laws that are entirely separate from their permit regulations, but fall within the
enforcement jurisdiction of DHEC as well. A good example of such an instance was enforcement
actions brought against Suffolk Chemical Company by DHEC from, approximately, August of 1982
until the site's subsequent order to cease operation and it's closure on July 3 1, 198 7. For its Chapin site
in Lexington county, Suffolk had obtained a National Pollutant Discharge Elimination System (NPDES)
permit to release non-contact cooling water and uncontaminated stormwater runoff into a ditch that
drains into a Lake Murray tributary. In the five year period noted above, on-site releases of various
chemicals into the air, such as chlorine gas, occurred several times. Although Suffolk had no air
discharge permit with DHEC, the Commissioner of DHEC used his authority as a public health official
6
to invoke statute 44-1-140 which grants the authority to close a facility that poses an immediate threat
to the public health and safety. This led to the closure of the site and was the basis for all enforcement
action taken against Suffolk, even though no permit with DHEC existed for regulation of the air
discharges involved.
The consent orders in this study were all issued in the period from January of 1993 to December
of 1994. The total number of consent orders for this time period was 664. These orders were issued by
the enforcement divisions of five program areas within DHEC: Air Quality Control (A), Drinking
Water Protection (DWP), Water Pollution Control (W), Solid Waste (SW) and Hazardous Waste
(HW). All information concerning these orders was obtained from the proper authorities and official
records within DHEC.
A number of the original 664 orders were omitted from the study for one of the following
reasons: (I) If there were no actual regulatory violations committed by the violator; for example, the
order was for establishing some form of compliance schedule and/or permit limitations between the
party responsible for a site and DHEC. (2) The violator was from out of state or was a mobile
transporter of waste or other materials that had no fixed location within the state of South Carolina. (3)
The exact location ( within limits) of the occurrence of the original cited violation could not be
pinpointed by any means available. ( 4) The violation occurred in a program area that did not fit into
the methodology of the study. For example, consent orders for infectious and radioactive waste were
not included because these types of violations contain within them, by definition, elements of the 3
previous reasons cited here for exclusion of a consent order. These omissions brought the total number
of orders used for this study to 511.
7
Research Parameters: How to measure Inequity?
In order to measure the results of comparing the selected independent variables for this study,
i.e. the minority and economic status of the community in which a consent order was issued and, to a
lesser extent, the specific bureau that issued the order, dependent variables were identified. For this
study, two parameters were chosen as adequate for measuring the impact on differing community
compositions~ these parameters will be referred to as the dependent variables.
The first parameter is the amount of dollars initially assessed as a penalty for the infraction(s) of
environmental regulations by the violator. This monetary amount is listed in the final draft of the
consent order, along with any provisions for the payment of that amount. Several options are available
to a violator after the penalty amount has been assessed. A company may not be financially capable of
paying its initially assessed penalty and therefore may have it reduced or suspended completely in order
to expedite the remediation and clean up of the polluted site. There is also an appeal process which any
violator can choose to undertake if they feel any part of the enforcement process is not being carried out
fairly, including the penalty assessed. To keep the amount assessed each facility a standardized value,
the monetary amount that was initially assessed the violator is designated as the "penalty" amount. This
is the civil penalty deemed an adequate amount by DHEC to compensate for the environmental
infractions committed by the violator. If taken at the time of issuance, this is a standardized value for all
enforcement actions involved since no permutation of the amount has been considered or performed
based on specific circumstances of the particular violator or violation. Our hypothesis is validated if
there are significantly higher penalties assessed to violators located in white and/or aflluent communities
compared to minority and/or poor communities.
8
A similar parameter, also used to assess environmental equity, was that of penalties imposed in
court for RCRA violations as shown in a 1992 investigation by the National Law Journal [8]. These
penalties were assessed as a result of civil cases concluded in federal courts from 1985 to 1991 against
various facilities which the EPA took to court for violating the federal hazardous waste law. This study
found that the average penalties imposed in court for RCRA violations vary dramatically with the racial
composition ( 500 percent higher if the facility polluted white communities than if they violated the law
in minority communities), but not the wealth, of the communities surrounding the waste sites.
The second dependent variable defined was enforcement time. One must define an enforcement
action as having an identifiable beginning and end in order to consider the time between these two
points a standardized, usable value. The five bureaus within DHEC came to a consensus on the
definition of these two points in time.
For the purposes of this study, the initiation point of an enforcement action was taken as the
date of the last inspection, performed by DHEC personnel, of the site where the violation occurred.
This inspection was the last to be conducted before enforcement action was initiated by the enforcement
division of each bureau responsible for the permit issued to said site. The notification of an enforcement
action started against a person or persons responsible for a site is in the form of a Notice of Violation
(NOV) sent to the violator -usually by registered mail. Upon response to this notice, a meeting is set
up between DHEC and the violator to discuss whatever issues of environmental and enforcement
concern are deemed necessary by DHEC . If the two parties reach a consensus on what needs to be
done to remedy the situation at the site, then a consent order is drawn up by legal counsel stating all
-pertinent findings and facts of the situation, as well as remedial measures to be taken by the violator in
order to bring the site into a state of compliance with relevant environmental regulations. This
document must be signed by the violator, their legal counsel, DHEC's legal counsel and other
9
responsible authorities within DHEC. One signature required on every consent order, for it to be
considered official and complete, is that of the Commissioner of DHEC (currently Doug Bryant). The
date that the Commissioner signs the document was chosen as the end point, or the date of completion,
of an enforcement action by DHEC. Our hypothesis is upheld if the average length of time it takes for
an enforcement action to be completed in a primarily nonwhite and/or poor community is greater, by a
statistically significant amount, than in a comparable white and/or affluent community.
It should be noted again that this enforcement process, as described, is a fairly ideal chain of
events. Although occurring as described more often than not, it can be interrupted, postponed or
delayed for any length of time, and for a variety of reasons, at numerous points during the process. A
thorough review of all of the possible reasons for the lengthening of the enforcement process
(environmental inequity being one of them) is beyond the scope of this study. We have assumed, for
our purposes, that due to the size of the study sample and the variety of the bureaus and type of actions
involved, confounding variables (for example, legal action taken by the violator in the form of an appeal
or other suits) are likely to cancel each other out within our data set, which is, as nearly as possible, a
census of a two year period of DHEC enforcement action.
A third independent variable was also included in the statistical analysis applied to our data: the
bureau issuing the consent order. The fact that each bureau within DHEC deals with largely different
criteria and approaches for enforcement, based on the types of facilities and pollution involved in their
program areas, became evident as the study progressed. Therefore, the bureau issuing a consent order is
regarded as a secondary or side focus of this study, since differences in penalty amounts and/or
enforcement times between bureaus are expected.
Comparing Communities: Separating South Carolina into Census Tracts
Once parameters were defined as the basis for comparison of different communities that have
violating facilities located within them, these communities were defined as to economic and racial
composition. As mentioned before, for this purpose we made use of a set of definitions in Danika
Holm's thesis. As in her investigation, this study utilized demographic data obtained from the 1990 US
Census. This data was combined with the census tract boundaries from the Census Bureau TIGER
(Topically Integrated Geographic Encoding and Referencing) data.
In order to obtain the geographical locations of each site where consent orders were issued, we
made use, primarily, of two services. Geographic Data Technology (GOT) is an address geocoding
service which used the latest version of Matchmaker/2000® for Windows™ to provide latitudinal and
longitudinal coordinates for sites of violations, once a reasonably complete address was found for the
facility. The software program MapExpert 2.0 for Windows™ by Delorme Mapping® (1993) was then
used to locate the coordinates for the rest of the locations that GOT had been unable to identify. Any
missing locations still left at this point were found through sources within the various offices of DHEC
across South Carolina. Even with all of these available resources, there were, regrettably, 30 locations
that we could not identify, and this missing data is recognized as a potential source of error and/or bias
in our study. However, these missing consent orders did not have any uncommon penalties or
enforcement times associated with them, so the characteristics of these omissions were very similar to
the consent orders for which locations were identified.
With this coordinate data, the Geographic Information System (GIS) was employed to analyze
the attributes of the community immediately surrounding the location of a site that had received a
consent order from DHEC in the specified time window (1/93 -12/94). This study also utilized the
11
ARC/INFO geographic information system software, developed by Environmental Systems Research
Institute (ESRI}, to perform all of the GIS functions required. ARC/INFO has the capability to store,
manipulate and analyze both spatial, geographic information and non-spatial, attribute information. In
addition, the study relies on the plotting and macro-creating subsystems provided by ARC/INFO for
cartographic and tabular outputs [7].
Geographical data describe objects from the real world in terms of (I) their positions with
respect to a known coordinate system, (2) their attributes which are unrelated to their position, and (3)
their spatial interrelations with each other (topological relationships), which describe how they are
linked together [7].
In this study, analysis of the data was performed at the Census tract level of data aggregation.
Census tracts are small geographic areas, with each tract averaging approximately 4000 people, the
standard representative community size [7]. They are considered the most specific unit for the
definition of a community, compared to other possible units such as block groups and counties, and
contain the smallest average number of people. Thus, census tracts were selected as the best unit of
comparison for the purposes of this study, and throughout the remainder of this work, the words
"tracts" and "communities" are used interchangeably. A social profile based on race and percent below
poverty level was compiled for each of the communities serving as hosts for facilities that were issued
consent orders in our study. These profiles are split into 8 categories based on 1990 averages for the
state of South Carolina for the composition of race and economic status within census tracts. The state
averages were 31.84% and 16.36% of a census tract being nonwhite residents and residents below the
federally defined poverty line, respectively. A tract was considered high minority (HM) if it had a
percentage of nonwhite residents that was greater than the state average, and, conversely, low minority
(LM) if its percentage of nonwhites was less than the state average. This same type of dividing line was
12
used for the economic status of a census tract, splitting each tract into high poverty (HP -greater than
16.36% below poverty line) and low poverty (LP -less than 16.36% below poverty line) categories.
These four categories were then combined to get an idea of what effect, if any, the combination
of poverty and minority status had on a community in relation to the enforcement actions conducted
within it. All 8 of the tract categories are listed in Table I with their defining characteristics, total
number of tracts in the category, and the number of total consent orders, in effect enforcement actions,
that were issued in each category during the study time period.
Figures I .A and l .B show the distribution of minority (HM) and economically disadvantaged
(HP) census tracts (853 in total), respectively, across South Carolina. Figure 2 displays the distribution
of the 4 combinations of census tracts outlining minority and economic status across the state. The
data set used in this study is superimposed on both displays. This map allows visual identification of
clusters or pockets of violations that were committed in certain census tracts of South Carolina. It also
gives a visual representation of the size of the data set dealt with in this study as compared to the total
size of the area containing that data set.
Data Analysis and Results
In order to conduct the most accurate and clear comparison of our equity variables across
bureaus and census tracts, SAS was used through the University of South Carolina. Specifically, the
procedure used for this data was the Multivariate Analysis of Variance (MANOVA) hypothesis test
method. This procedure is essentially an extension of the univariate ANOV A, allowing inclusion of
more than one dependent variable, thus reducing the likelihood of a Type I error and increasing the
13
overall power of the test. The MANOV A test allowed the comparison of the means of high/low
minority, high/low poverty and the 5 bureaus across both of our equity variables (penalty amounts and
enforcement times), simultaneously.
The MANOVA was the first and most important test criteria applied to the data and evaluated
the null hypothesis of the difference of the means between any given independent (bureau, minority
status and economic status), or group of independent variables and either one, or both, of our
dependent variables simultaneously. The multivariate statistics used, as a group, gave F approximations
for the hypothesis of no overall effect of the independent variables being tested. That is, at a 95%
confidence level, if certain statistics are within 0.05 of the F value (association is very unlikely due to
chance) for an independent variable, then it relates to either, or both simultaneously, of the dependent
variables (means differ by a statistically significant amount). The statistics used included Wilks'
Lambda, Pillai's Trace, Hotelling-Lawley Trace and Roy's Greatest Root. A detailed description or
derivation of these values is beyond the scope of this discussion, and so only the results of one particular
statistic, Wilks' Lambda, will be considered as it applies to our data set. This statistic is used because it
is more robust to violations of the standard assumptions of normality et al than the other statistics. A
list of the Wilks' Lambda values for all the independent variables is displayed in Table 2 with the
appropriate degrees of freedom for each F statistic listed as well.
The next test applied to the data was the overall F test (univariate) that yielded P values which
evaluate the null hypothesis for any two or more sample means being compared. Table 3 and 4
summarize these values for all three of the independent variables tested and their combinations as they
affect penalty amounts issued and enforcement time, respectively. The F test for any combination of the
independent variables tested as compared to either dependent variable ( one at a time) was used to
generate P values for each test. The means of the independent variables being compared on the
14
dependent variables differ by a statistically significant amount if the F value is greater than the critical F
value for the appropriate degrees of freedom. Consequently, from the F test, the P value generated
must by< 0.05 in order for the means of the variables to be considered significantly different (i.e. there
is less than a 5% probability that the means differ by chance). At this point, it is important to note that
the values for tables 2, 3 and 4 are based on a type III sum of squares calculation whereby the sum of
squares for each variable is calculated taking into account all of the other variables.
An example of this analysis and the conclusions that can be drawn from it, for this data set, is
that if the independent variable of race (high or low minority) was related to the dependent variable,
penalty amount, then the P value for the mean of the race category population would be less than 0.05
for the F test. Therefore, these results serve as a support mechanism to the MANOV A test shown in
table 2 in order to identify relationships and significant differences between means of samples.
The next step was the application of Tukey's studentized range (HSD) test for each of the
dependent variables. Whenever the analysis of variance leads to a rejection of the null hypothesis of no
difference among population means, the question naturally arises regarding just which pairs of means
are different [5]. This test is a multiple comparison procedure developed by Tukey and is frequently
used for testing the null hypotheses that all possible pairs of treatment means are equal when the
samples are all the same size. Spjotvoll and Stoline have extended the Tukey procedure to the case
where the sample sizes are different and that is the procedure utilized here. When this test is employed,
we select an overall significance level of a (which is equal to 0.05 for this analysis) [5]. This
comparison was made between the mean values of each dependent variable for each bureau, in all
possible combinations, at a 95% confidence level. The differences between means computed that yield
an absolute value exceeding the HSD (honestly significant difference -the calculation of which is
15
beyond the scope of this discussion), and are therefore considered significant at the 0.05 level, are listed
in Table 5.
Finally, Table 6 lists the means of the actual data collected for both of the dependent variables,
grouped by each independent variable acting on them in tum followed by some significant interactions
of these variables.
Discussion: Evaluating Our Hypothesis
Penalty Amounts:
When inspecting table 6 there are some obvious characteristics that come into play when
analyzing and trying to draw conclusions from this data. The samples used to calculate means and
standard deviations for each category vary greatly in size. The smallest sample size is that for the Solid
Waste bureau issuing orders in low minority, high poverty census tracts: 1 consent order in total. This
small sample size reduces the standard deviation to zero and makes this sample, as well as others in the
set, less useful when comparing means. We will first discuss the results for the larger categories of
census tracts as they apply to the dependent variable of penalty amount and keep in mind when doing
this that as categories of census tracts become more specific, the sample size must be taken into account
to see if the statistics being compared to each other are meaningful.
Table 2 lists the independent variables that could possibly exhibit a statistically significant
relationship with penalty amount based on the MANOV A test. The variables that stand out as having a
significant effect on penalty amounts issued (P<0.05) are bureau (P = 0.0001), race or minority status of
16
a census tract (P = 0.0329) and the interaction of these two variables (P = 0.0152). All the other
variables and possible interactions of these variables do not exhibit a significant relation to penalty
amount according to this test at the appropriate degrees of freedom. We now look to our supporting
tables, 3 and 4, to see if the univariate test of each independent variable supports these relations. Table
3 lists the P values for penalty amount and shows the same relations to be significant as table 2. All
other independent variables and combinations of such are, again, not significantly related to penalty
amount at the appropriate degrees of freedom.
The first independent variable that exhibits a significant relationship to the penalty amount issued
is the type of bureau issuing the penalty in a particular census tract. To find out if there is a significant
difference between Bureaus and if so, which Bureaus are significantly different from each other, we use
Tukey's Studentized Range (HSD) listed in table 5. The significant differences in mean penalty
amounts issued are between the Bureau of Hazardous Waste and each of the Bureaus of Air Quality,
Drinking Water Protection and Water Pollution Control. Although the absolute magnitude of the
difference between mean penalty amounts issued by the bureaus of hazardous waste and solid waste is
comparable to the other differences cited as significant ($96914.00), the upper and lower confidence
limits of this difference contain the value of 0 within them and that denotes it as being not significant in
the boundaries of the HSD test.
What we can conclude from these analyses is that the Bureau of Hazardous Waste, on average,
issues significantly higher penalties to violators in all census tracts as compared to the Bureaus of Air
Quality, Drinking Water Protection and Water Pollution Control. This result is expected considering
that waste handling Bureaus (HW and SW) deal more, on average, with larger, more substantial
companies. For these types of companies, environmental infractions are usually more far-reaching,
damaging and take longer to remediate, on average, as compared to the types of facilities that are dealt
17
with by the other 3 bureaus. Thus, larger penalties being issued by the bureau of hazardous waste falls
in line with the types of facilities these penalties are, most often, levied against. It does not seem a
coincidence, therefore, that the two highest penalty amounts for the entire sample were both issued by
the Bureau of Hazardous Waste.
The next significant independent variable relating to penalty amount is race. The mean penalty
amount for the high minority tracts, across our entire sample, are substantially higher than the mean
penalty amount for the low minority tracts (Table 6). It is noticeable that the standard deviation for the
penalty amounts of the high minority tracts is substantially greater (by 21 times) as that for the low
minority tracts -showing the high minority tracts to be a much more widespread sample. From this
data, we can ascertain that, on average, higher penalties are issued by DHEC in high minority
communities than in low minority ones.
Finally, the significant interaction between bureau and race and this interaction's relationship to
the penalty amounts issued is examined. This is done with the help of the interaction plot in Figure 3.
This is a plot of the mean penalty amounts issued by each bureau (labeled on X-axis) in high minority
and low minority communities (two different lines displayed). This plot shows that the mean penalty
amounts issued are higher in low minority communities for the Bureaus of Air Quality, Solid Waste and
Water Pollution Control. Subsequently, the mean penalty amounts are higher in high minority
communities for the Bureaus of Hazardous Waste and Drinking Water Protection. It should be noted
here that size varies greatly among these samples categorized by race and bureau, as can be seen from
table 6.
No comparisons can be made between bureaus or across the total sample for the relationship of
the economic status of the census tracts with penalty amounts issued in those tracts. Economic status
was not found to correlate significantly with penalty amounts for any of our statistical tests and we must
18
I I ' .I
I )
i
I 1
i
I
I I
assume, therefore, that any differences between penalty amounts for different census tracts, based on
economic status, are not statistically significant and can be deemed due to chance. Thus, the last four
lines of table 6 are included for the sake of curiosity and, again, any seeming relation between economic
status and penalty amount (or enforcement time for that matter) are not due to this independent
variable, but are a result of the real significance in the relationship between minority status of the census
tracts and penalty amount.
Enforcement Time:
The same significant relationships identified by table 2 for penalty amounts apply to enforcement
times as well. However, one difference that can be noticed immediately is that the interaction of bureau
and race is not supported by Table 4 where the univariate F test is applied to enforcement time. The
independent variable, bureau, is supported as significant (P = 0.0008) and the variable of race is also
very close to significance (P = 0.0663). Therefore, we examine the effects of each of these variables
separately and then scrutinize more closely the effect of their interaction on enforcement time.
To identify any significant differences between bureaus, we look to the HSD test again and find
that the Bureau of Water Pollution Control differs significantly in mean enforcement time from the
bureaus of Drinking Water Protection and Air Quality (table 5). The mean enforcement time is higher
for enforcement actions taken by the Bureau of Water Pollution Control than for these other two
bureaus. All other mean enforcement times do not differ significantly among the bureaus. The
underlying reason for this elevated mean enforcement time for the Bureau of Water Pollution Control
(also illustrated in the mean values of table 6) is not readily obvious. One possible explanation is that
the initial enforcement conference between this bureau and the violator can come before or after the
19
initial NOV pertaining to the order being issued. This means that there can be long lags in time (more
pronounced than some other bureaus) between the first interaction between this bureau and the violator,
and later official meeting times. This characteristic, in tum, is due to the fact that enforcement actions
by this bureau involve the issuance of previous orders prior to the ones being considered (in our case,
those from 1993-94). These previous orders are not necessarily fully resolved by the time a new
consent order or enforcement action is being considered against the violator, due to the logistics of the
types of violations and environmental problems this bureau encounters.
Table 6 shows a that the mean enforcement time is greater for high minority communities than
for low minority ones. Again, the standard deviation for the high minority communities shows more
variance in enforcement times for these census tracts as compared to that of low minority communities.
20
I .
I: 1
Potential Confounds
One fact worthy of discussion for this sample set is that two of the penalty amounts issued
during our study time period were larger by an order of magnitude of 10 times than any other penalties
in the sample. This fact, by itself, does not seem so extraordinary, but the fact that both of these
extremely large penalties fall into the same category of race and bureau for our census tracts lends some
elaboration on the results seen here.
The two penalties, $1825000 and $3909000 issued to Stohler Chemical and Laidlaw
Environmental, respectively, both were issued in high minority communities by the Bureau of
Hazardous Waste. This explains the considerably greater variance of penalty amounts in high minority
communities as compared to low minority ones.
This occurrence in our particular sample seemed worth investigating; the same statistical analysis
conducted on the full sample was also calculated for that sample minus these two consent orders. The
result was that bureau was the only independent variable proving significant, although the interaction of
all three independent variables showed a more prominent effect on the dependent variables
simultaneously. The effects of these two outliers on our study sample are negated by the statistical
analysis that was applied, but they do lend an interesting possibility that the data acquired perhaps could
not have been duplicated if the study had been conducted for a different period of time.
21
Conclusions
For the dependent variable of penalty amount, our hypothesis is not supported by the data
acquired and analyzed here. The mean penalty amounts issued in high minority areas were, on average,
higher in high minority areas than in lower minority ones, contrary to our hypothesis. No statistically
significant relationship existed between economic status of any community and the penalty amounts
issued or the enforcement time taken for a consent order issued.
Our hypothesis was supported for the interaction of the independent variable, minority status or
race, with enforcement time, since on average, the enforcement time was greater in high minority areas
than in low minority communities. That is, it took longer, on average, for DHEC to complete an
enforcement action taking place in a high minority community than in a low minority one.
Recommendations
1) Enforcement tracking systems should record a start date and completion date for each enforcement
action.
2) Effort should be put into geocoding specific locations of enforcement actions.
3) Enforcement staff should identify and examine the factors that contribute to the length of time
involved in enforcement actions, e.g., district referral, enforcement unit NOV issuance, legal office
consultation, and public involvement.
4) Attention should be given to modifying factors in the process that contribute to a lengthier
enforcement time in minority communities.
22
References
(1) Bullard, R.D.: Dumping in Dixie: Race, Class and Environmental quality. Westview Press,
Boulder CO, 1990.
"-"· Overcoming Racism in Environmental Decisionmaking. Environment 36, 10-20, 39-44,
1994.
(2) Chavis, B.F. jr.: Preface in Bullard, R.D., editor, Unequal Protection: Environmental Justice
and Communities of Color. Sierra Club Books xi -xii, San Francisco, CA, 1994.
(3) Collin, R. W.: Environmental Equity: A Law and Planning Approach to Environmental Racism.
Virginia Environmental Law Journal 11, pp. 495-546, 1994.
"----": Environmental Equity and the Need for Government Intervention: Two Proposals.
Environment 35, pp. 41-43, 1993 .
(4) Cutter, S.L.: Race, class and Environmental justice. Progress in Human Geography 19, 1 pp .
107-118, 1995.
(5) Daniel, W.W.: Biostatistics: A Foundation for Analysis in the Health Sciences. 4th Edition,
John Wiley and Sons, NY, 1987.
(6) Geiser, K. and Waneck, G.: PCBs and Warren County. Science for the People, July/August,
13-17, 1983.
(7) Holm, D.M.: Environmental Inequities in South Carolina: The Distribution of Hazardous
Waste Facilities. Master of Science Thesis in the Dept. of Geography at the University of South
Carolina, 1994.
(8) Lavelle, M., Environmental justice. In World Resources Institute, editor: The 1994
Information Please Environmental Almanac. Houghton-Mifflin, Boston, MA, 183 -92, 1994.
(9) Mohai, P. and Bryant, M.: Environmental Racism: Reviewing the Evidence. In Bryant, B.
and Mohai, P., editors, Race and the Incidence o(Environmental Hhazards: A Time of
Discourse. Boulder, CO: Westview Press, pp. 163-176, 1992.
"----": Environmental Injustice: Weighing Race and Class as Factors in the Distribution of
Environmental Hazards. University of Colorado Law Review 63, pp. 921-932, 1992.
(I 0) Mushak, B.: Environmental Equity: A New Coalition for Justice. Environmental Health
Prospectives 101,478 -83, 1993 .
(11) Roque, J.A.: Environmental Equity: Reducing Risk for All Communities. Environment 35,
pp . 25-28. Commentary and reply -Environment 35, p.4, 1993 .
(12) South Carolina Department of Health and Environmental Control: Hazardous Waste
Activities Reported in South Carolina for 1991. Columbia, SC : SC DHEC, 1992
(13) United Church of Christ, Commission for Racial Justice: Toxic Wastes and Race in the
United States. United Church of Christ Commission for Racial Justice, New York, 1987.
(14) United States General Accounting Office: Siting ofHazardous Waste Landfills and Their
Correlation With Racial and Economic Status of Surrounding Communities. Government
Printing Office, Washington, DC, 1983.
(15) US Environmental Protection Agency: Environmental Equity: Reducing Risk for All
Communities. Washington, DC: Government Printing Office, 1992.
Tables and Figures
Table 1: Categories of Census Tracts based on Race and Economic Status
Name of Catl'gOfY Defining Characteristic I Number of Census ; Number of Consent
~ > ••
(Symbol) . . . ( Trads in C.ntegonf ; • Orders Issued
I >
High Minority (HM) > 31.84% Nonwhite Residents 357 223
Low Minority (LM) < 31 .84% Nonwhite Residents 497 288
High Poverty (HP) > 16.36% Below Poverty Line 356 241
Low Poverty (LP) < 16.36% Below Poverty Line 498 270
High Minority and High > 31.84% Nonwhite Residents and
Poverty (HM/HP) > 16.36% Below Poverty Line 281 182
. . .
High Minority and Low > 31 .84% Nonwhite Residents and I l l
Poverty (HM/LP) < 16.36% Below Poverty Line I 75 I 41 I
-----········ .. ······································---··-----····· ........................................................................................................................................ ~
~::;;:;d ffigh : :~::::::::h~:v::::: and ! 76 I 59 I
Low Minority and Low < 31 .84% Nonwhite Residents and
Poverty (LM/LP) < 16.36% Below Poverty Line 422 229
Table 2: MANOV A Test Criteria and F Approximations for the Hypothesis of no Overall Effect of
Given Independent Variable on Either or Both Dependent Variables•
Bureau 0.91909544 j 5.2779 8 0.0001
.......................................................... · ......................................................................................................................... ··························· ............................................... .
ESb l 0.99649094 I 0.8627 2 0.4226
......................................................... i ....................................................................... ~ ................................................. ·························· ·················································
Bureau and ES i 0.98582960 I 0.8773 8 0.5351
···································· .. ······ .. ············~································································· ... · .. l ................................................ ··························· ················································
•• ::::•;t•ce.. ·'=••············· :.::;::::: ~~ =!·~~=:.:::: ..................... : ......................... :.:::: =
Bureau, Race and ES \ 0.99325494 j 0.4152 8 0.9122
··························································l·· .. ···································································:···································· .. ··········1. ......................................................................... .
• Characteristic roots and vectors calculated for each variable based on: E Inverse • H, where H = Type III Sum of Squares and
Cumulative Polynomial Matrix for given variable and E = Error of said matrix
b Economic Status
Table 3: F and P values for Type m Sum of Squares Test as Applied to Penalty Amountc
lnde11endent Variable DF ! FValue I P Value !
Bureau 4 5.77 0.0002
Economic Status 1 0.20 0.6525
Bureau and Economic Status 4 0.15 0.9624
Race 1 4.10 0.0435
Bureau and Race 4 4.60 0.0012
Race and Economic Status 1 0.18 0.6719
Bureau and Race and Economic Status 4 0.25 0.9105
Table 4: F and P values for Type m Sum of Squares Test Applied to Enforcement Time
Independent Variable DF l F Vaine l P Value ! l
Bureau 4 4.81 0.0008
Economic Status 1 1.61 0.2048
Bureau and Economic Status 4 1.61 0.1716
Race 1 3.39 0.0663
Bureau and Race 4 0.14 0.9658
Race and Economic Status I 0.00 0.9729
Bureau and Race and Economic Status 4 0.58 0.6741
0 Sum of Squares for Error= 31567568.76, DF = 491
Table 5: Tukey's Studentized Range (HSD) Test for Penalty Amount and Enforcement Time Among
Different Bureaus of DHECd
• Dependent l Bureaus Simultaneous Lower Difference : Simultaneous Upmr I • Variable j Com ared Confidence. Limit Befiveen Means '. Confidem:e Limit
Penalty Amount HWandW 13273 96101 178928
HW and A I 26083 100055 174027 I
••••••••••••••• ........ ••••--•OOOOOOOOO••ooonoHOOOOOOOOO •••••••••••• .. ••••••••••••• .............. ••·•••u••••••••••••·•••••--•••••••••• .. ••••••••••••••••• .. ••·•••••••••••• .. •••••••••••• .. •••••••••• .. •••••• .. •••••••••••••••< HW and DWP : 27283 103564 179846 :
Enforcement Time Wand DWP 53.86 148.11 242.35 ...................................................................................... _ _.., _______ _,.__ ....................................................................................................... .
Wand A 67.17 157.91 248.65
Table 6: General Linear Models Procedure: Means and Standard Deviations for All Possible
t . • .
H 223 31258.99 288030.66 315.86 306.94
L 270 . 14746.20 : 111378.92 300.04 232.52 : ............................................ ···········i·········t··············i'1i I 5151.35 I 9224.13 252.04 190.50 1
................................................................... P.~ ...... ! ................ }.~?.····--·····}§1.~.:9..?. ........... L ..... 53 56.43 .............. }.~.~.:.~.?. ................ 309 •. 56 ....... !
HW j 65 105206.39 ! 529322.52 313.89 169.56 j
............. -............................ -...................... sw ....... t .................. 41 ................. ) ............ 8292.23 ........... L .... 11284.6.1 .............. 3_14.38 ............... ..186.20 .... ..J
W I 89 I 9105.84 \ 12720.42 409.96 335.70 I
H -A . 92 __ 4294.57 ! 5460.79 240.83 162.76 :
···········'i····················....... • • • A j 79 6149.11 ···••j••··12191.396 ···········265.1·0 ........ ······218 .• 80·····••j
H -DWP : 48 : 2836.46 I 8738.32 340.46 481.97 i ····························· ......................................................... .,. ........................................... : ............................................................................................................................................... ~
L DWP l 91 l 1012.09 : 1719.13 220.38 143.49 :
H -HW l 21 287400 i 918832.49 317.71 173.75 i ···········i············· •••••••••••••••••••••••••••••••• ·······irw······r················4·4 18250.34 j 19118.88 312.01 169.52 j
H -SW \ 18 l 3394.72 \ 3563 .60 335 236.56 : .............................................................. ························t········· .................................................................................................................................................................................. ,.
L -SW I 29 : 11332.07 : 13285.52 301.59 150.14 \
............. ~ ....................................................... ~ .......... , ................... 1.i......... 119 5 .46 L ..... ?.?.§.~.:1?. ................ 1}.?.:.~.9. .............. }}.?.J?. .... J
L W ! 45 , 10387.11 ! 14940.12 383 .31 337.83 l
H H -i 182 l 26768.16 l 289482.71 300.07 309.24 : ····························· ......................................................... +······································· ................................................ +···· .............................. ····························--.... ·····························••-: • H L : 41 : 51193.90 i 284151.45 385.95 289.84 i
............ ~ ........................... !! ...................................... L ................. ?..?. ........................... 4990.11 I 12938.16 263 .32 201.96 i
L L i 229 1 822oi3 • i 13751.61 284.66 211.90·······1
d All numbers are absolute values
• Enforcement Time
lilm'e LA: MiDolity Tndl
[I
&d'oa..-.... Adioa Locatiom
llpre 1.B:
Tncta nh Mme Tban 16.36"
Below the Powaty Line
Bufo.ooaa,em Action Locatiom
----======•---iic:===HI IIIIIUI
---••====•=-----"c:::::==::51W IIIIIUI
acmmc-
H Percent Persom Below the Poverty Line L Pen:ent Nonwhite L H L = Lower Than the State Average H = Higher Than the State Average • Number of TractB Liat.ed in Box • • Enforcement Action Locatiooa Flgure 2: Minority and/or Economically Disadvantaged Tracts 0 25 50 75 100 MILl!S Saarce: 1990 C-.
~ -C :, 0 E c( ~ iii C Q) 0.. C "' Q) == Figure 3: Affect of the Interaction Between Bureau and Race on Mean Penalty Amount Issued 1000000 100000 10000 1000 100 10 1 1•it=i=i=i=~=Ji=i~=it=i=i=~=JiWitt~tii=i:it=ilJli=iiJitt=Iitt~~tf~tit~t}i~tti:ltJi~fi~~=ti~fi=i=i=iWfiWi=i~~=tJit=ti=i=i=i=i}i~t=i=i=tt=i~tt}it~WJiWi~~~=~fi~~W~~Wit~Wifi=ifitWififi}i}iffi{i ---A DWP HW Bureau SW w --+-Mean Penalty Amount ($) for HM -Mean Penalty Amount ($) for LM
ACKNOWLEDGMENTS
We would like to thank Environmental Quality Control ofDHEC for funding of this project. We also
greatly appreciate the time, advice and other resources put forth by all of the DHEC staff, in Columbia
and in the district offices across South Carolina, throughout this study.
Special thanks and recognition goes to the GIS staff ofDHEC: Pat Horton, Paul Laymon, and Jeannie
Eidson for their ingenuity, expertise, and commitment to this project.
We are grateful to Danika Holm for the foundation that her work provided and for her consultation on
this project.
Finally, we are indebted to Dr. Susan Cutter for her insight and expertise on the subject of
environmental justice.
PRINTED DECEMBER 1995
Total Printing Cost -$3 5 .20
Total Number of Documents Printed -100
Cost Per Unit -$0.352
" ,, ,
A STUDY OF PROCEDURAL
ENVIRONMENTAL EQUITY IN
ENFORCEMENT: DHEC CONSENT ORDERS
1993-94
A D D E N D U M
The enforcement of environmental law within the State of South
Carolina is carried out by the Office of Environmental Quality
Control (EQC) of the S.C. Department of Health and Environmental
Control (DHEC) in accordance with the Uniform Enforcement Policy
(U.E.P.), approved by DHEC's Board on December 12, 1991. This
policy was created with the intention of providing uniform
procedures for the conduct of enforcement actions and uniform
criteria for the assessment of civil penalties. Each Bureau may
establish procedures and guidelines to better enable them to carry
out the U.E.P., but they must be consistent with the policy.
The civil penalty amounts may vary greatly between orders and
between Bureaus depending upon several factors including, but not
limited to, the type of violation, the potential for harm to the
public or environment, frequency or duration of the violation, past
record of compliance, and economic benefit to the Respondent as a
result of the violation. The statutory maximum civil penalty
amounts also vary between programs, i.e. $500 for a first offense
violation under the State Recreational Waters Act to $25,000 per
day of violation for violations of the S.C. Hazardous Waste
Management Act. The settlement amounts often reflect this
relationship to the statutory maximum penalty.
1
The locating of facilities or sites with the potential to
pollute is a result of multiple factors apart from which EQC
enforcement staff exercise no influence or control.
It is the practice of EQC staff to provide compliance
assistance to those experiencing problems, should circumstances
warrant. Reasonable opportunity is thereby given to a Respondent
to regain compliance status before an enforcement option is
pursued. Formal enforcement actions are reserved as a last resort.
It should be noted that the time frames for many of the enforcement
cases included in this study encompass that period of time allowed
for the Respondent to attempt compliance restoration. Also, the
length of time needed to complete an enforcement action is often
dictated by the complexity of the factors surrounding the
noncompliance event as well as the intensity of negotiation when
multiple factors are considered or when significant civil penalties
are involved.
The decision to initiate an enforcement action is usually
based upon the existence of a documented, unresolved violation
brought to the attention of enforcement staff either through a
referral for enforcement action by DHEC compliance staff, e.g., as
the result of a complaint or spill investigation, or through the
routine review of monitoring reports. Monitoring reports may be
generated either by DHEC staff as the result of on-site compliance
inspections, or by the permittee and submitted to DHEC in
fulfillment of a permit or order requirement . Information
pertaining to the demographics or economics of a community in which
2
\
an alleged violation has occurred is not provided to enforcement
staff and is therefore not a factor in prioritizing or managing an
enforcement case. Cases are prioritized based upon the degree of
endangerment to the public's health or the environment and upon the
nature and extent of the violation alleged.
3