ArticlePDF Available

The spatial distribution of toxic chemical emissions: Implications for nonmetropolitan areas

Authors:

Abstract and Figures

The United States has experienced considerable industrial decentralization to non‐metropolitan areas during the past thirty years. Whereas researchers have extensively documented the social and economic effects of this process, the environmental consequences of manufacturing activities in rural communities have received less attention. This paper utilizes the U.S. Environmental Protection Agency's Toxic Release Inventory (TRI) to investigate the extent to which nonmetropolitan counties in the state of Indiana serve as locations for toxic chemical emissions from industrial sources. The results of this analysis suggest that particular rural counties experience elevated emissions of industrially produced toxic chemicals. In addition, when compared with urban‐based manufacturing firms, industrial establishments in the state's nonmetropolitan areas tend to emit more toxic chemicals, both per manufacturing employee and per dollar of manufacturing earnings. If manufacturing employment is considered the compensation that communities receive in exchange for exposure to toxic chemical emissions, the residents of nonmetropolitan areas appear to be compensated at a lower rate than urban residents. However, uncertainty remains regarding the actual human health and environmental risks created by industrial activity in nonmetropolitan counties because of data inadequacies and complexities measuring individual exposure incidences.
Content may be subject to copyright.
The Spatial Distribution
of
Toxic Chemical Emissions:
Implications
for
Nonmetropolitan Areas
MAURE
J.
COHEN
Oxford Centre for the Environment, Ethics, and Society
Mansfield College
Oxford University
Oxford, UK
The United States
has
experienced considerable industrial decentralization
to non-
metropolitan areas during
the
past thirty years. Whereas researchers have extensively
documented
the
social
and
economic effects
of
this process,
the
environmental conse-
quences
of
manufacturing activities
in
rural communities have received less attention.
This paper utilizes
the
U.S. Environmental Protection Agency's Toxic Release Inventory
(TRI)
to
investigate
the
extent
to
which nonmetropolitan counties
in the
state
of
Indiana
serve
as
locations
for
toxic chemical emissions from industrial sources.
The
results
of
this analysis suggest that particular rural counties experience elevated emissions
of
in-
dustrially produced toxic chemicals.
In
addition, when compared with urban-based
manufacturing firms, industrial establishments
in the
state's nonmetropolitan areas tend
to emit more toxic chemicals, both
per
manufacturing employee and
per
dollar
of
manu-
facturing earnings.
If
manufacturing employment
is
considered
the
compensation that
communities receive
in
exchange
for
exposure
to
toxic chemical emissions,
the
residents
of nonmetropolitan areas appear
to be
compensated
at a
lower rate than urban resi-
dents. However, uncertainty remains regarding
the
actual human health
and
environ-
mental risks created
by
industrial activity
in
nonmetropolitan counties because
of
data
inadequacies
and
complexities measuring individual exposure incidences.
Keywords environmental equity, industrial hazards, risk, rural industrialization,
rural pollution, rurality, Toxic Release Inventory
The process of nonmetropolitan industrialization in the United States since the mid-1950s
has been extensively documented (Summers et al., 1976; Summers and Selvik, 1979;
Lonsdale and Seyler, 1979).
1
This movement of economic activity is typically considered
to have reached its apex during the 1970s, when nearly 50% of all new manufacturing
employment in the country was generated in rural communities (Lonsdale, 1979;
Schnaiberg, 1986). Although there have been contractions in nonmetropolitan industry in
more recent years (Barkley, 1993), manufacturing still remains a dominant component of
the American rural landscape. The importance of this economic sector is reflected by the
Received
27
February 1995; accepted
23
August
1995.
This research
was
completed while
the
author
was
affiliated with Indiana University's School
of
Public
and
Environmental Affairs. Previous versions
of
this article were presented
at the
annual meet-
ings
of
the Association
of
Collegiate Schools
of
Planning, Tempe, Arizona, USA, November 1994,
and
at
the
conference
for
Regional Development:
The
Challenge
of the
Frontier, sponsored
by the
Negev
Center
for
Regional Development, Ben-Gurion University
of
the Negev
and the
Lewis Center
for Re-
gional Policy Studies, University
of
California
at
Los Angeles,
Ein
Bokek, Israel, December 1993.
The author thanks Melanie Byers
and
Timothy Lehman
for
their valuable research assistance.
Address correspondence
to
Maurie
J.
Cohen, Oxford Centre
for the
Environment, Ethics,
and
Society, Mansfield College, Oxford University, Oxford
0X1 3TF, UK.
E-mail: maurie.cohen@
mansfield.oxford.ac.uk
17
Society
&
Natural Resources,
10:17^11,
1997
Copyright
©
1997 Taylor
&
Francis
0894-1920/97 $12.00
+ .00
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
18 M. J. Cohen
fact that manufacturing currently provides the major source of export earnings for rural
counties in the United States (Deavers, 1992).
2
Research on nonmetropolitan industrialization in the United States has largely fo-
cused on the dynamic social and economic implications of this transformation. The well
established evidence (Summers et al., 1976; Summers and Servile, 1979; Lonsdale and
Seyler, 1979) suggests that for many rural communities industrialization has not occurred
without considerable stress. Beyond these widely considered effects, the introduction of
manufacturing activity into rural communities traditionally dependent on agriculture and
other forms of natural resource production has had several environmental consequences,
most notably deteriorated air and water quality, as well as reductions in open space. This
analysis focuses specifically on the extent to which nonmetropolitan areas may experi-
ence disproportionate emissions of toxic chemicals from industrial sources.
This paper is divided into five sections. The following section briefly reviews several
arguments regarding why rural sites have become attractive locations for environmentally
hazardous manufacturing activities in recent years. This section also considers the unique
vulnerabilities that rural communities have to toxic chemicals. The methodology and data
used to assess the levels of chemical emissions in nonmetropolitan areas are then de-
scribed, followed by a tripartite analysis and a conclusion discussing certain parameters
in which the results of this investigation should be evaluated.
Nonmetropolitan Vulnerability to Toxic Chemicals
The emission of toxic chemicals by industrial sources to air and water media has typically
been viewed as an issue with which the residents of nonmetropolitan areas do not have to
be particularly concerned. Extensive chemical releases from manufacturing facilities have
been associated with more urbanized locations, where industrial activity has traditionally
concentrated. It is because of the pervasiveness of this perception that the enforcement
mechanisms designed during the previous twenty years target urban areas more aggres-
sively than rural. During the 1970s, when the United States was experiencing a period of
rapid nonmetropolitan industrialization, this orientation was perpetuated by researchers
who focused primarily on the effects of this process on population migration, employ-
ment, income, and fiscal policy (Summers et al., 1976; Summers and Selvik, 1979; Sum-
mers and Clemente, 1976; Ford, 1978; Dillman and Hobbs, 1982). Virtually no attention
during this time was devoted to the consequences of nonmetropolitan industrialization on
rural environmental quality.
Although competitive land and labor costs have surely been important catalysts of non-
metropolitan industrialization in the United States, the incrsased latitude for the externaliza-
tion of environmental costs offered by rural locations should not be underestimated. During
the previous two decades, local officials in many rural communities, to attract manufactur-
ing facilities, have given industry assurances that environmental regulations would not be
vigorously enforced (Schnaiberg, 1986).
3
Accordingly, several supplementary reasons can
be identified for why nonmetropolitan areas have proved attractive locational alternatives
for industrial facilities constrained by environmental regulations and why efforts to reduce
toxic chemical emissions in rural communities have often been frustrated.
First, lower population densities in nonmetropolitan areas offer the promise of fewer
conflicts between waste disposal and community demands for responsible corporate man-
agement of environmental hazards (Schnaiberg, 1986; Flora et al., 1992). Most rural
communities have an ample supply of affordable industrial sites that can be used to facili-
tate environmentally damaging industrial processes without public scrutiny.
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Spatial Distribution of Toxic Chemical Emissions 19
Second, a lack of adequate economic diversification in nonmetropolitan areas gives
local manufacturers extensive coercive power over employees and other residents. In par-
ticular, industrial employees in rural communities are vulnerable to negative sanctions in
response to their organizational activities related to environmental hazards (Kazis and
Grossman, 1982; Nelkin and Brown, 1984; Gould, 1991; Robinson, 1991).
Third, the operators of nonmetropolitan industrial facilities often own or control
most of the land in the local community and account for the largest share of the property
tax burden (Gould, 1991). This fiscal dependence often prevents local officials from pur-
suing more aggressive strategies to confront environmental hazards resulting from rural
manufacturing operations.
Fourth, environmental consciousness appears to be correlated with levels of educa-
tional attainment, particularly with respect to ambiguously defined hazards (Schnaiberg,
1986;
Van Liere and Dunlap, 1980). Because of the poorer quality education typically
available in nonmetropolitan areas (Lonsdale, 1979), their residents may have weaker
propensities for interpreting less tangible environmental threats. Most industrial hazards,
because of their long latency periods and the complex etiologies of their associated health
effects, fall into this category.
Fifth, the cultural orientation of nonmetropolitan residents results in lower participa-
tion rates in social and political organizations to protest environmental violations
(Schnaiberg 1980, 1986). Additionally, because of confrontations many rural residents
have previously had with environmental-advocacy groups regarding natural resource ex-
traction practices, alliances concerning the infiltration of nonmetropolitan industry are
likely to be fragile.
Finally, rural communities do not generally have access to the scientific expertise, fi-
nancial resources, and political influence necessary to establish credible claims of human
health impairment resulting from toxic contamination (Schnaiberg, 1986; Cohen, 1996;
Kroll-Smith and Couch, 1990; see also Kasperson and Pijawka, 1985; Kroll-Smith et al.,
in press). These deficiencies make it difficult for nonmetropolitan communities to chal-
lenge corporate claims that the risks of hazardous production processes are minimal.
In aggregate, the presence of these factors can provide industrial employers in non-
metropolitan areas with a high degree of control (Gould, 1991; Schnaiberg, 1980). This
power gives facility operators influence over the local community to dictate the agenda of
acceptable environmental debate and to decide the outcomes of environmental risk as-
sessment conflicts.
Worthy of additional consideration is the fact that the emission of toxic chemicals by
industrial sources may expose rural communities to environmental hazards not generally
considered critical in more densely populated areas. Specifically, the greater reliance of
nonmetropolitan residents on local groundwater supplies and the agricultural orientation
of the economic base create unique vulnerabilities for rural residents.
Nonmetropolitan communities have a heightened dependency on local groundwater
supplies for direct human consumption and irrigation needs (Gulliland et al., 1991; Tobin
and Rajogopal, 1993). These utilization characteristics, in combination with the tendency
for toxic substances to migrate through underground channels, can increase rural expo-
sure to environmental contaminants. Simultaneously, small towns often lack the technol-
ogy necessary to detect the presence of toxins in groundwater, and this lack often results
in failure to identify potential human health hazards promptly. In the event the presence
of toxic chemicals in local groundwater sources is confirmed, implementation of elabo-
rate treatment programs can be financially burdensome for rural communities (Nichols,
1989).
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
20 M. J. Cohen
Two types of toxic chemical emissions from industrial sources, namely ozone pre-
cursors and acidic compounds, can have special consequences for nonmetropolitan areas
because they affect agricultural growth rates. First, ozone, a secondary air pollutant
formed from chemical reactions between nitrogen oxides (NO
X
) and several volatile or-
ganic compounds (VOCs), causes direct leaf injury and tree mortality (Comrie, 1990).
4
Second, acid deposition begins with emissions of sulfur and nitric oxides, primarily from
electric utility plants and smelters. These acids, which can be carried hundreds of miles
by wind and weather, release metals from soils and can accumulate to toxic levels in fish
and other organisms. Although research has focused on the urban production of these
chemicals and their meteorological transport to nonmetropolitan areas (Logan, 1989; Mc-
Keen et al., 1991), extensive industrial decentralization in recent decades suggests manu-
facturing facilities may be releasing large volumes of ozone precursors and acidic com-
pounds directly into rural communities.
5
In addition to these two considerations, nonmetropolitan areas must contend with
several other environmental hazards that derive from their lower population densities and
other special characteristics. First, because of high land values and siting difficulties in
more urbanized regions, rural communities have developed an economic role as disposal
sites for extralocal household and hazardous wastes. Second, agricultural chemicals and
leachates from natural resource exploitation can constitute severe public health threats in
nonmetropolitan areas. Finally, the extensive network of military installations and pro-
duction facilities located across the nonmetropolitan landscape poses complex environ-
mental dilemmas.
Methodology
There are indications that the socioeconomic and sociopolitical characteristics of non-
metropolitan areas, in combination with the well founded efforts of industrial firms to
reduce their regulatory burdens through locational adjustments, have led to reductions
in the environmental quality of many rural communities. The Midwestern state of Indi-
ana provides the context to investigate whether there is any evidence to substantiate
this contention. The state can be divided into three tiers: southern, central, and north-
ern. Indiana's southern and central tiers are extensively agricultural, with production
concentrated in corn, oats, soybeans, and wheat. The northern tier of the state, although
technically part of the American "Rust Belt," retains considerable manufacturing activ-
ity, based largely on primary and fabricated metals, industrial machinery, electric and
electronic components, and transportation equipment. Indiana, despite its status as the
fourteenth most populous state in the United States (5.5 million residents in 1990), is
among the top five states in total toxic chemical releases and transfers from industrial
sources.
6
Indiana has six metropolitan centers surrounded by a rural hinterland containing
many small population concentrations economically dependent on manufacturing and
agricultural and residentiary services. Five of Indiana's metropolitan centers are located
in the state itself (Indianapolis, Fort Wayne, South Bend, Gary and its surrounding com-
munities, and Evansville). A sixth metropolitan center, Louisville, although actually in
the neighboring state of Kentucky, extends into the bordering county in Indiana. These
metropolitan centers can be used to divide the state inlo six functional economic areas
(Figure 1). Indianapolis, by virtue of its primacy, has the largest functional economic
area, whereas the influence of the other five smaller cities is largely confined to the coun-
ties adjacent to each metropolitan center.
7
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Spatial Distribution of Toxic Chemical Emissions
2. SOUTH BEND (ST. JOSEPH COUNTY)
21
1.
GARY ANO SURROUNDING CITIES
(LAKE COUNTY)
4. MOMNAPOUS (MARION COUNTY)
3. FORT WAYNE (ALLEN COUNTY)
6. LOUISVILLE (FLOYD COUNTY)
- EVANSVUE (VANDEHBUROH COUNTY)
Figure 1. Indiana functional economic areas.
To assess the spatial distribution of toxic chemical emissions from industrial sources
in Indiana, this analysis requires a measure of relative ruralness. Several variables have
been previously proposed to capture the notion of rurality, including population density,
distance from a large urban center, economic specialization, and agricultural (or natural
resource) employment (Dewey, 1960; Bealer et al., 1965; Smith and Parvin, 1973, 1975;
Miller and
Lulloff,
1981; Ilvento et al., 1988; Deavers, 1992).
The current analysis defines rurality as a composite function of population density
and distance from a metropolitan center. A rurality measure, r¡, was calculated for each
county in Indiana as the product of population density per square mile, = p¡/a¡ (where
p is population and a is geographic area), and the linear distance between each county's
administrative seat and its corresponding metropolitan center, d¡? The subscript / (/ = 1,
2,
. . . , 92) connotes the county in Indiana. More formally,
In empirical application this formulation is defined as
(2)
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Table
1
Rurality data
by
geographic category
and
county,
state
of
Indiana,
1989
County/Category
Rurality
0
Region*
Populatioi
Area'
Distance''
Density*
Urban counties
Marion
0.02
Lake
0.03
Vanderburgh
0.04
St.
Joseph
0.04
Allen
0.05
Floyd
0.15
Urban totals
Suburban counties
Clark
0.23
Porter
0.23
Elkhart
0.26
Johnson
0.27
Hendricks
0.33
Madison
0.34
Hamilton
0.36
LaPorte
0.37
Warrick
0.38
Suburban totals
First-ring rural counties
Monroe
0.40
Delaware
0.41
Hancock
0.42
Howard
0.43
Morgan
0.44
DeKalb
0.47
Clay
0.47
Adams
0.47
Tippecanoe
0.47
Whitley
0.48
Huntington
0.48
Marshall
0.49
Bartholomew
0.49
Scott
0.51
Vigo
0.51
Harrison
0.52
Shelby
0.52
Posey
0.53
Boone
0.53
Noble
0.54
Kosciusko
0.54
Wells
0.55
Henry
0.57
Wayne
0.60
Grant
0.62
Gibson
0.62
Starke
0.63
Stueben
0.67
Fayette
0.67
Washington
0.68
Blackford
0.72
Clinton
0.73
Dubois
0.73
4
4
4
4
4
3
4
3
4
3
3
2
4
6
4
6
4
5
4
3
2
3
4
4
4
5
2
3
4
6
3
4
5
797,159
475,594
165,053
247,052
3OO,83i5
64,404
2,050,103
87,777
128,932
156,19(5
88,109
75,717
130,669
108,936
107,06(5
44,920
928,324
108,973
119,659
45,527
80,827
55,920
35,324
24,70:5
31,09:5
130,593
27,651
35,427
42,182
63,657
20,991
106,107
29,890
40,307
25,9615
38,147
37,877
65,294
25,94»
48,139
71,951
74,169
31,913
22,747
27,446
26,015
23,717
14,067
30,974
36,616
396
501
236
459
659
150
2,401
376
418
466
321
409
453
398
600
391
3,832
385
392
385
293
409
364
360
340
502
336
366
444
409
191
405
369
413
409
423
413
540
370
394
404
415
490
309
308
215
516
166
405
429
1
1
1
1
1
10
12
17
23
20
20
34
36
25
17
46
51
21
50
27
21
15
20
60
19
23
23
40
29
70
22
27
18
26
27
37
23
44
68
69
26
33
40
56
24
45
41
46
2,013
949
699
538
457
429
854
233
308
335
274
185
288
274
178
115
242
283
305
118
276
137
97
69
91
260
82
97
95
156
110
262
81
98
63
90
92
121
70
122
178
179
65
74
89
121
46
85
76
85
(continued)
22
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Table
1
Rurality data
by
geographic category
and
county, state
of
Indiana,
1989
(continued)
County/Category
First-ring rural counties
LaGrange
Spencer
Tipton
Putnam
Knox
Montgomery
Dearborn
Miami
Lawrence
Decatur
Cass
Jay
Owen
Daviess
Fulton
Wabash
Brown
Rush
Jackson
First-ring rural totals
Second-ring rural counties
Jennings
Orange
Jefferson
Jasper
Vermillion
Pike
Randolph
Green
Perry
Carroll
Crawford
Ripley
Franklin
Ohio
Union
Fountain
Pulaski
White
Parke
Newton
Sullivan
Martin
Switzerland
Warren
Benton
Second-ring rural totals
Grand totals
Rurality"
0.74
0.77
0.77
0.79
0.80
0.80
0.81
0.82
0.83
0.86
0.86
0.90
0.90
0.90
0.91
0.92
0.93
0.93
0.98
1.00
1.00
1.00
1.02
1.02
1.03
1.07
1.09
1.10
1.11
1.11
1.12
1.14
1.21
1.25
1.26
1.27
1.28
1.28
1.30
1.39
1.45
1.63
1.78
1.93
Region*
3
5
4
4
5
4
4
4
4
4
4
3
4
5
2
4
4
4
4
4
6
4
1
4
5
4
4
6
4
6
4
4
2
4
4
2
4
4
1
4
5
4
4
4
Population
29,477
19,490
16,119
30,315
39,884
34,436
38,835
36,897
42,836
23,645
38,413
21,512
17,281
27,533
18,840
35,069
14,080
18,129
37,730
2,140,354
23,661
18,409
29,797
24,960
16,773
12,509
27,148
30,410
19,107
18,809
9,914
24,616
19,580
5,315
6,976
17,808
12,643
23,265
15,410
13,551
18,993
10,369
7,738
8,176
9,441
425,378
5,544,159
Area
17
380
400
260
482
520
505
307
369
452
373
414
384
306
432
369
398
312
406
513
20,151
378
408
363
561
260
341
454
546
382
372
307
447
385
87
162
398
435
506
444
401
452
339
223
366
407
9,424
35,808
Distance
1
'
42
29
37
39
49
44
84
68
65
47
69
45
46
52
42
74
39
39
70
62
45
82
46
67
39
69
66
61
62
40
69
66
90
67
71
47
75
57
57
81
64
92
71
86
Density*
78
49
62
63
77
68
126
100
95
63
93
56
56
64
51
88
45
45
74
106
63
45
82
44
65
37
60
56
50
51
32
55
51
61
43
45
29
46
35
34
42
31
35
22
23
45
155
Source: Author's calculations based
on
U.S. Bureau
of
the Census, Census
of
Population and Housing (1990).
"Rurality index,
r
r
^Functional economic region:
1 =
Gary
and
surrounding communities;
2 =
South Bend;
3 =
Fort Wayne;
4 =
Indianapolis;
5 =
Evansville;
6 =
Louisville.
c
Area
in
square miles.
''Distance
in
miles
to
metropolitan center
as
calculated
by
author.
"Number
of
persons per square mile.
23
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
24 M. J. Cohen
and is referred to as the rurality index (Table 1).
Whereas the use of population density as a variable in the rurality index is straight-
forward, the utilization of linear distance as a proxy for actual travel distance between
each county center and its respective functional economic area is contingent on four sim-
plifications inherent in Indiana's economic and physicd geography. First, Indianapolis,
the state's capital and primary city, is located in the geographic center of Indiana. Sec-
ond, nearly all of the state's county-level administrative seats are situated near the geo-
graphic centers of their respective counties.
9
Third, Indiana has no appreciable topo-
graphic variation (except for in the extreme southern part of the state) or geographic
barriers (e.g., major rivers, large lakes). Finally, the state has a relatively dense and un-
congested transportation network.
Based on this definition of rurality, Indiana's ninety-two counties have been divided
into a geographic hierarchy comprising four discrete categories: urban, suburban, first-ring
rural, and second-ring rural. First, the six counties with low rurality indexes, < 0.15, are
characterized as urban jurisdictions. Second, the nine counties with rurality indexes in the
range of 0.15 < r¿ < 0.40 are considered suburban. Third, the first-ring rural classification
comprises the fifty-two counties with rurality indexes of 0.40 < r,- < 1.00. The twenty-five
counties with rurality indexes of > 1.00 are in the second-ring rural category.
To examine the spatial distribution of toxic chemical emissions from industrial facili-
ties across the different classifications of rurality, this analysis considered the three chem-
icals with the largest number of emission sources in Indiana: 1,1,1-trichloroethane,
toluene, and sulfuric acid.
10
First, 1,1,1-trichloroethane, or methyl chloroform (CH
3
CC1
3
), is a colorless, nonflam-
mable solvent often used as a cold-cleaning agent to dissolve greases and oils (U.S. Envi-
ronmental Protection Agency [EPA], 1990a). The chemical is also used as a dry cleaning
agent, a vapor degreasing agent, and a propellant (Sittig, 1985). Although not considered
carcinogenic, trichloroethane is classified as both a hazardous waste and a priority toxic
pollutant by the EPA. Both acute and chronic low-level exposures to the chemical have
been found to cause liver and cardiac malfunctions (Sittig, 1985).
Second, toluene, or methyl benzene (C
6
H
5
CH
3
), is a clear, colorless, noncorrosive
solvent widely used as a chemical building block (U.S. EPA, 1990a). Classified by the
EPA as a hazardous substance, hazardous waste, and priority toxic pollutant, toluene is a
VOC used as a feedstock for the manufacture of toluene diisocyanate, phenol, and other
chemicals. Commercially, it is used as a solvent for paints and coatings and as a compo-
nent of automobile and aviation fuels (Sittig, 1985). Because of its mainly hydrocarbon
composition, toluene is a precursor chemical in the foimation of ozone (Comrie, 1990;
U.S.
EPA, 1990a). Acute exposure to toluene results in central nervous system depression
and has also been found to cause liver and kidney damage (Sittig, 1985).
Finally, sulfuric acid (H
2
SO
4
), a colorless and odorless liquid, is one of the largest
production-volume chemicals manufactured. It is considered a hazardous material by the
EPA and is used as a feedstock by a wide range of industries to manufacture acetic acid,
hydrochloric acid, citric acid, and other compounds that call for a relatively inexpensive
acid. Sulfuric acid is widely used in the production of synthetic fertilizers, nitrate explo-
sives,
artificial fibers, dyes, Pharmaceuticals, detergents, adhesives, paints, and paper, as
well as for electrolyte in storage batteries (Sittig, 1985). Sulfuric acid is also used as a
"pickling" chemical to clean rust from iron and steel prior to the application of paint or
coating (U.S. EPA, 1990a). This chemical, in addition to causing pronounced skin irrita-
tion because of its water-reactive properties, is associated with respiratory and bronchial
damage and digestive disturbances (Sittig, 1985).
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Spatial Distribution of Toxic Chemical Emissions 25
Data
The chemical disaster in Bhopal, India, in 1984 motivated the federal government in the
United States to establish the Toxic Release Inventory (TRI) as part of the Emergency
Planning and Community Right-to-Know Act (EPCRA) of 1986. This act, also referred
to as Title III of the Superfund Amendments and Reauthorization Act of 1986, requires
industrial facilities to report on their release and transfer of hundreds of chemicals (Had-
den, 1989; U.S. General Accounting Office [GAO], 1991a, 1991b). TRI is presently
available as a database that provides unprecedented public access to information on toxic
chemical emissions from industrial sources.
11
Under EPCRA, all facilities in the manufacturing sectors corresponding to Standard
Industrial Code (SIC) classifications 20 through 39 and employing ten or more full-time
employees must report the quantity of each of their toxic chemical emissions to the EPA
if they (a) produce, import, or process 25,000 pounds or more of one or more of 300
chemicals or (b) use 10,000 pounds or more of a listed chemical (U.S. EPA, 1990a).
12
Re-
ported data are required to be disaggregated by emissions released directly into the air,
land, and water, or sent to locations that treat, store, or dispose of hazardous wastes (U.S.
GAO,
1991a).
Approximately 3,500 separate toxic chemical emission sources were reported in In-
diana for 1990.
13
From this total, the aggregate sum for each of the three subject chemi-
cals (trichloroethane, toluene, and sulfuric acid) was assembled by county (Table 2). Al-
though these data provide an opportunity to assess the extent and distribution of
emissions of potentially harmful substances from industrial sources, the TRI has several
deficiencies that presently impinge on its utility as a comprehensive source.
First, as currently constituted, the TRI does not collect data from nonmanufacturers.
14
Several major emission source categories exist outside the manufacturing sector, includ-
ing mineral mining and processing, electricity production, oil.and gas extraction, and
agricultural operations.
13
However, the most critical omissions from the TRI are perhaps
the numerous federal facilities not represented by the database, including the vast net-
work of industrial installations operated by the Department of Defense and the Depart-
ment of Energy.
16
Second, the TRI's exclusion of manufacturing firms with fewer than ten employees
eliminates from consideration a potentially large number of emission sources. EPA esti-
mated in 1988 that there were approximately 164,500 manufacturing facilities nationwide
with fewer than ten employees (U.S. GAO, 1991a). Although not all of such facilities use
toxic chemicals in excess of the reporting thresholds, the magnitude of this excluded cate-
gory provides an indication of the scale of this exemption.
Third, by 1990 an estimated 10,000 required facilities nationwide (of a total of
30,000) were still failing to fulfill their TRI obligations, many apparently because they
were unaware of the reporting requirement.
17
Some states have estimated nonreporting
rates as high as 50% (U.S. GAO, 1991b), but financial constraints have limited the ability
of state environmental agencies to ensure more thorough compliance.
Finally, the TRI also suffers from the EPA's lack of financial resources to verify the
accuracy of reported data. Given the complexity of the reporting requirements and possi-
ble efforts by individual facilities to underrepresent their emissions, the reliability of
these data is difficult to ascertain.
18
One estimate of the shortcomings of the TRI is provided by the Congressional Office
of Technology Assessment (OTA), which has indicated the database may undercount
total toxic chemical emissions in the United States by at least 95% (U.S. GAO, 1991a).
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Table
2
Selected toxic chemical emissions
by
county
and
geographic
category,
state
of
Indiana,
1990
(emissions volumes
in
thousands
of
pounds)
County/Category
Urban counties
Marion
Lake
Vanderburgh
St.
Joseph
Allen
Floyd
Urban totals
Suburban counties
Clark
Porter
Elkhart
Johnson
Hendricks
Madison
Hamilton
LaPorte
Warrick
Suburban totals
First-ring rural counties
Monroe
Delaware
Hancock
Howard
Morgan
DeKalb
Clay
Adams
Tippecanoe
Whitley
Huntington
Marshall
Bartholomew
Scott
Vigo
Harrison
Shelby
Posey
Boone
Noble
Kosciusko
Wells
Henry
Wayne
Grant
Gibson
Starke
Stueben
Fayette
Washington
Blackford
Clinton
Rurality
index
0.02
0.03
0.04
0.04
0.05
0.15
0.23
0.23
0.26
0.27
0.33
0.34
0.36
0.37
0.38
0.40
0.41
0.42
0.43
0.44
0.47
0.47
0.47
0.48
0.48
0.49
0.49
0.51
0.51
0.52
0.52
0.53
0.53
0.54
0.54
0.55
0.57
0.60
0.62
0.62
0.63
0.67
0.67
0.68
0.72
0.73
0.73
Trichloroethane"
Sites
19
6
5
6
4
1
41
0
3
24
2
0
3
2
3
0
37
3
3
1
2
1
2
0
4
4
1
3
2
5
1
2
0
2
1
0
2
3
0
2
0
2
2
0
2
2
1
0
2
Emissions
513.0
533.0
82.9
99.5
287.2
60.9
1,576.5
0.0
551.4
1,461.7
37.7
0.0
176.4
77.5
1,080.3
0.0
3,385.0
69.5
96.1
60.8
242.4
14.5
428.1
0.0
422.0
169.7
16.9
40.0
32.2
309.9
12.8
98.2
0.0
49.3
0.0
0.0
176.2
1,225.8
0.0
463.1
0.0
52.6
75.2
0.0
101.9
54.5
5.5
0.0
122.0
Sites
16
11
7
9
7
1
51
4
2
21
1
0
2
1
3
2
36
5
1
1
2
1
2
1
4
4
1
3
3
7
0
3
2
1
2
0
2
6
1
0
3
3
0
0
1
2
1
2
2
Toluene
Emissions
723.7
1,215.8
386.7
1,054.5
435.9
9.7
3,826.2
240.9
17.9
595.9
198.8
0.0
281.3
42.1
41.1
81.3
1,499.3
93.8
143.6
1.5
26.8
15.2
39.5
182.2
73.7
1,443.8
0.8
98.9
245.5
135.5
0.0
382.2
38.0
17.4
382.4
0.0
45.0
3,303.0
13.0
0.0
162.9
36.6
0.0
0.0
2.3
5.3
32.0
1,534.7
21.4
Sulfuric acid
Sites
21
9
2
4
5
3
44
4
4
6
1
0
3
0
4
0
22
1
7
2
6
0
1
0
2
5
1
0
1
1
0
3
0
1
1
0
1
4
1
2
1
1
0
0
1
2
0
0
2
Emissions
126.1
220.8
0.5
28.3
25.6
135.5
536.8
2.5
21.7
1.5
0.3
0.0
11.4
0.0
1.0
0.0
38.3
38.0
26.2
0.5
7.6
0.0
0.0
0.0
0.0
130.5
0.3
0.0
0.9
0.3
0.0
289.8
0.0
0.5
0.1
0.0
14.0
1,051.3
0.2
0.7
118.4
0.3
0.0
0.0
0.5
0.4
0.0
0.0
0.3
(continued)
26
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Table
2
Selected toxic chemical emissions
by
county
and
geographic category, state
of
Indiana,
1990 (emissions volumes
in
thousands
of
pounds) (continued)
County/Category
First-ring rural counties
Dubois
LaGrange
Spencer
Tipton
Putnam
Knox
Montgomery
Dearborn
Miami
Lawrence
Decatur
Cass
Jay
Owen
Daviess
Fulton
Wabash
Brown
Rush
Jackson
First-ring totals
Second-ring rural counties
Jennings
Orange
Jefferson
Jasper
Vermillion
Pike
Randolph
Green
Perry
Carroll
Crawford
Ripley
Franklin
Ohio
Union
Fountain
Pulaski
White
Parke
Newton
Sullivan
Martin
Switzerland
Warren
Benton
Second-ring totals
Grand totals
Rurality
index
0.73
0.74
0.77
0.77
0.79
0.80
0.80
0.81
0.82
0.83
0.86
0.86
0.90
0.90
0.90
0.91
0.92
0.93
0.93
0.98
1.00
1.00
1.00
1.02
.02
.03
.07
.09
.10
l.ll
.11
1.12
L.14
L21
1.25
.26
.27
.28
.28
1.30
1.39
1.45
1.63
1.78
1.93
Trichloroethane"
Sites
2
2
0
0
2
0
0
1
0
2
0
4
1
0
0
1
2
0
0
2
74
0
1
1
0
0
0
1
0
2
0
0
0
0
0
0
0
0
1
0
1
0
0
1
0
0
8
160
Emissions
40.0
99.0
0.0
0.0
88.8
0.0
0.0
29.3
0.0
17.0
0.0
145.0
16.3
0.0
0.0
25.3
56.9
0.0
0.0
1,642.0
6,498.8
0.0
3.1
0.3
0.0
0.0
0.0
195.9
0.0
36.2
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
9.4
0.0
7.3
0.0
0.0
16.2
0.0
0.0
268.4
11,728.7
Sites
21
0
1
0
2
1
1
2
1
1
0
1
1
0
0
0
1
0
0
3
102
0
3
2
0
1
0
5
0
2
1
0
2
0
0
0
1
0
0
0
0
0
0
2
0
0
19
208
Toluene
Emissions
801.7
0.0
58.5
0.0
19.2
1.0
25.6
135.1
9.2
19.4
0.0
14.9
11.8
502.6
0.0
0.0
16.6
0.0
0.0
93.1
10,185.7
0.0
104.1
15.7
0.0
706.0
0.0
691.1
0.0
101.0
207.0
0.0
191.3
0.0
0.0
0.0
262.4
0.0
0.0
0.0
0.0
0.0
0.0
41.6
0.0
0.0
2,320.3
17,831.5
Sulfuric acid
Sites
3
0
0
0
0
1
2
1
0
0
2
1
0
0
0
0
3
0
0
2
62
1
0
2
0
1
0
0
0
0
0
0
0
0
0
0
2
0
2
0
1
0
0
0
0
0
9
137
Emissions
0.5
0.0
0.0
0.0
0.0
46.6
0.0
0.3
0.0
0.0
1.0
0.3
0.0
0.0
0.0
0.0
54.8
0.0
0.0
21.5
1,807.6
0.0
0.0
2.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.4
0.0
4.0
0.0
0.1
0.0
0.0
0.0
0.0
0.0
7.1
2,389.7
Source: Author's calculations based
on
U.S. Environmental Protection Agency, Toxic Release Inventory, (1990b).
"1,1,1-Trichloroethane (Methyl chloroform).
27
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
28 M. J. Cohen
For example, the EPA reported that national TRI emissions in 1987 were 6.2 billion
pounds, compared with an OTA estimate of approximately 400 billion pounds.
19
Despite
these allegedly large discrepancies, more recent appraisals of the TRI's accuracy suggest
that the EPA has improved its ability to account for the volume of toxic chemicals re-
leased from industrial sources.
These observations indicate that the TRI is an evolving database. Nonetheless, if it is
assumed that omissions and errors are spatially evenly distributed, the TRI can be used to
conduct a preliminary investigation of the spatial distribution of toxic chemicals emitted
by industrial facilities to identify potential environmental and human health hazards asso-
ciated with nonmetropolitan industrialization.
Analysis
The first phase of this analysis considered the bivariate relationship between rurality and
each of the subject toxic chemical emissions (trichloroethane, toluene, and sulfuric acid)
using a spline-smoothing technique. Spline smoothing is a statistical tool for describing
empirical data in low-information situations in which a mathematical model is inappro-
priate. This approach provides an approximation of data using a piecewise cubic polyno-
mial.
20
The data are divided into subintervals, or knots (specified by X.), and a cubic poly-
nomial is fit for each subinterval (Goodall, 1990). This ¡¡et of third-degree polynomials is
then spliced together such that the resulting curve is continuous and smooth at the knot
points (Lehman et al., 1989).
21
Spline smoothing, which is expected to provide only a rough approximation of the
trend between rurality and toxic chemical emissions, was applied on an experimental
basis.
The final fit for each substance required the application of a flexible spline with
several knots (K = 0.0001). The results suggest a relationship between rurality and
toxic chemical emissions that is bimodal or, in certain cases, multimodal (Figures 2, 3,
and 4).
As expected, Indiana's urban counties, r
f
- < 0.15, experience elevated levels of toxic
chemical emissions. As rurality increases into the range of the suburban counties, 0.15 <
< 0.40, a sharp decrease in releases is observed. Another steep rise (in some cases sur-
passing the levels recorded for the urban counties) in emissions is observed for the first-
ring rural counties, 0.40 s < 1.00. As rurality increases in the second-ring rural coun-
ties,
Tj- > 1.00, releases decline asymptotically to zero. The coefficients of determination
can be interpreted roughly as measures of the variation accounted for by the dependent
variable. These R
2
values indicate that the relative degree of rurality (or alternatively, ur-
banity) explains approximately 20% to 30% of the emissions of the three subject chemi-
cals.
Although there is still considerable variability that this approach fails to capture, the
manufacturing sector appears to have shifted a large volume of its toxic chemical emis-
sions from conventional urban locations. In fact, TRI data for 1990 indicate that only
25.6%
of Indiana's trichloroethane emission sources were located in urban counties, and
13.4%
of its emission volumes occurred in urban locales (Tables 3 and 4). In contrast,
46.3%
of Indiana's trichloroethane emission sources were situated in first-ring rural
counties and 55.4% emission volumes of this substance occurred at these sites. Similar
disparities in the spatial distribution of toluene and sulfuric acid are evident. These obser-
vations independently should not be surprising and simply reflect the larger land area in
the state defined as within the first-ring rural classification (56%) and the extensive
process of nonmetropolitan industrialization that has occurred in Indiana.
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Spatial Distribution of Toxic Chemical Emissions
29
1500-
M
•a
3
O
a
X)
co
«
o
1000-
ë 500-
ca
I
2
0"
0.0
0.5
1.0
Rurality Index
2.0
Smoothing Spline Fit Statistics
A = 0.001
R
2
= 0.25
Figure 2. Distribution of trichloroethane emissions in Indiana, 1990.
However, often underlying policy debates on industrial location is the notion that
manufacturing employment and earnings represent compensation for the local environ-
mental deterioration and possible deleterious human health impacts that may accompany
industrialization. To ascertain any differences in this trade-off across the geographic hier-
archy, the second step in this analysis was to evaluate the spatial association between
toxic chemical emissions and manufacturing employment.
First-ring rural counties received in 1990 more than 50% of Indiana's releases of
each of the three subject toxic chemicals (see Table 4). However, as indicated in Table 5,
only 42.4% of the state's manufacturing jobs were located in these communities. In con-
trast, 34.8% of Indiana's manufacturing jobs were found in urban counties, but less than
25%
of the state's trichloroethane, toluene, and sulfuric acid emissions were produced in
this geographic category.
To take another perspective, emissions:employment quotients were calculated for re-
leases of the three subject chemicals in each classification of the geographic hierarchy.
These emissions:employment quotients are ratios of the volume of chemical emitted per
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
30
M. J. Cohen
3500
-500
0.0 0.5 1.0
Rurality Index
1.5
2.0
Smoothing Spline Fit Statistics
A = 0.001
R
2
= 0.24
Figure 3. Distribution of toluene emissions in Indiana, 1990.
manufacturing sector job. For example, industrial firms located in Indiana's urban coun-
ties supported 217,337 manufacturing jobs in 1990 (see Table 5) and released 1.6 million
pounds of trichloroethane (see Table 4) for an emissions:employment quotient of 7.3
(Table 6). In other words, urban industrial firms emitted 7.3 pounds of trichloroethane
into the environment for each manufacturing job. In comparison, the state's first-ring
rural manufacturers employed 264,903 workers and emitted 6.5 million pounds of
trichloroethane during the same period, for an emissions:employment quotient of 24.5.
On the basis of these emissions:employment quotients, it can be inferred that the volumes
of toluene and sulfuric acid released are roughly two to three times greater for each man-
ufacturing job in the first-ring rural counties than in the urban counties. Indiana's first-
ring rural counties, in comparison with the state's urban counties, do not appear to re-
ceive comparable compensation in the form of manufacturing employment for local
emissions of the three subject chemicals.
As previously noted, the prevalence of a lower wage scale in rural communities is
conventionally considered an important inducement for industrial migration to nonmetro-
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Spatial Distribution of Toxic Chemical Emissions 31
1100
0.5 1.0
Rurality Index
1.5
2.0
Smoothing Spline Fit Statistics
A = 0.001
R
2
= 0.20
Figure 4. Distribution of sulfuric acid emissions in Indiana, 1990.
politan locations. In Indiana, this urban-rural wage differential is reflected by the fact
that the mean annual earnings for workers employed in urban-based manufacturing was
$32,790 in 1990, 114% of the statewide average (see Table 5). By comparison, industrial
employees in the state's suburban counties earned a mean annual salary of $27,697. An-
nual earnings for first- and second-ring rural manufacturing workers averaged $26,731
and $21,804, respectively.
In the third stage of this analysis, these observations suggesting that first-ring rural
counties receive less compensation in terms of both manufacturing jobs and earnings than
urban communities in exchange for tacit approval to generate toxic chemical emissions
were combined. By assessing the direct economic benefits—as measured by aggregate
annual payroll—that industrial production conferred on different levels of the geographic
hierarchy, indications of possible inequities in this tradeoff were derived. As depicted in
Table 5, both urban and first-ring rural manufacturing establishments expended $7.1 bil-
lion on wages in 1990, approximately 39% of the statewide total. However, as indicated
above, the urban counties experienced only 13.4% of the state's trichloroethane ends-
Downloaded by [University of Wisconsin - Madison] at 08:45 16 May 2016
Spatial Distribution
of
Toxic Chemical Emissions
33
Table
4
Toxic chemical emission volumes
by
geographic category,
state
of
Indiana,
1990
(emission volumes
in
thousands
of
pounds)
Geographic
category
Urban
Suburban
First-ring rural
Second-ring rural
Statewide
Trichloroethane
Volume
1,576.5
3,385.0
6,498.3
268.3
11,728.1
Percent
13.4
28.9
55.4
2.3
100.0
Toluene
Volume
3,826.2
1,499.3
10,185.7
2,320.3
17,831.5
Percent
21.5
8.4
57.1