Printer Friendly

An investigation into the presence and causes of environmental inequity in Denver, Colorado.

Environmental equity, or justice, can be thought of as the achievement of an equitable or proportionate distribution of environmental effects from a given practice or policy. After decades of being overlooked, environmental equity, particularly with respect to race, is now under critical examination. Several studies have found the existence of a correlation between ethnic minorities and pollution or locally unwanted land uses such as hazardous waste facilities (Bunyan and Bryant, 1992; Bullard, 1983; United Church of Christ, 1987; U.S. General Accounting Office, 1983). The study conducted by the United Church of Christ prompted national attention to environmental racism. The presence of environmental racism results in an inequitable exposure to pollution, hazardous waste and toxic facilities by race, resulting in adverse health effects on groups least able to afford medical attention (United Church of Christ, 1987).

Following the results of these studies, there have been recent government policy initiatives to alter the current relationship of minority and low income residents with polluting industry. The Environmental Protection Agency (EPA) created the Office of Environmental Equity in 1993 to improve relations with state and local agencies involved in environmental decisonmaking and to inform and educate low-income and minority communities about environmental hazards. In February of 1994, President Clinton signed Executive Order 12898, Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations (Clinton, 1994). The order guarantees all Americans the right to protection from pollution regardless of income level, race, education or other such factors. Since the policies are relatively recent, their effectiveness is yet to be determined.

It is important to note that there are several scenarios which could result in a community of color experiencing a disproportionate share of environmental hazards. Pure racial discrimination in the siting process is one explanation (Hamilton, 1995). Or, site owners may focus on efficiency as a decision rule without regard for equity. Developers may choose areas with lower land values, higher unemployment or low rates of homeownership, which would lead them to expect less political opposition to site locations. However, there is a correlation between minorities and many of these socioeconomic characteristics. According to the 1990 Census, 25%, 29% and 31% of Hispanics, Blacks and Native Americans, respectively, live below the poverty level. The U.S. average is 13%, with the percentage for whites being 10%. If firms are minimizing land and political costs by siting in areas with low incomes and high unemployment rates, this would create a spurious correlation between site locations and minorities. Furthermore, the sequence of events which led to high levels of pollution in minority neighborhoods is particularly important for policy implications.

Nearly all previous studies have correlated the current characteristics of communities with cumulative number of locally unwanted land uses. This may provide little insight into the demographics of an area at the time the decision to locate a polluting site was made. The correlation between ethnic minorities and pollution could be caused by market forces which are driving lower income or minority people towards polluted communities (Been, 1995). Once a polluting site is built, it may become what is regarded as a locally unwanted land use. Due to its undesirability, surrounding land and home value may decrease and people with the ability to do so move out. However, due to the desirability of cheaper housing, lower taxes or better job opportunities, other people will move in. Owing to the high negative correlation to income, employment and homeownership, minorities may be attracted to these areas. The correlations may be due to deeper rooted discrimination in housing or job markets rather than the influence of racial prejudices in the siting process. Without chronological information, on the sequence of events, we do not know whether the disproportionate distribution of pollution and waste is a consequence of the siting process or of the market system combined with discrimination from other processes. This is extremely important since a policy measure based on the incorrect assumption regarding the cause of environmental equity could prove to be futile and costly.


This study aims to illustrate an approach which allows for a multivariate evaluation of the relative importance of sociodemographic and racial factors in prospective plant siting decisions. To attain these objectives the study will:

1. Estimate correlations between minorities and newly permitted stationary sources of criteria air pollutants in the Denver, Colorado area using zip code level data in order to replicate past studies.

2. Use multiple regression based methods to estimate a causal model relating stationary air pollution sites permitted as a function of racial composition and socioeconomic factors. This analysis will investigate the relative importance of racial and demographic variables present at the time of the siting decision by comparing the statistical significance and magnitude of each variable (standardized using an elasticity measure) in predicting the number of new stationary sources of air pollution permitted during the 1988-1992 time period.

3. Trace the census variable for the minority percentage of the population between 1980 and 1990 in communities where one or more sites were permitted for construction and in those where none were permitted. The purpose is to test the hypothesis that minorities are attracted to communities once polluting sites have been located.


Past studies focused on large hazardous waste sites (U.S. General Accounting Office, 1983; United Church of Christ, 1987; Anderton et al, 1994; Been, 1995). While these are important locally undesirable land uses (LULU's), one contribution of this study is to investigate whether the same pattern of results as past studies is evident for another important environmental problem, stationary sources of criteria air pollutants in a large, urban area. Criteria air pollutants are defined by EPA to include carbon monoxide, nitrogen oxides, sulphur dioxide, pm-10, ground level ozone, volatile organic compounds, lead and all hazardous air pollutants. This list is both quite comprehensive as well as clearly containing pollutants which present immediate health threats (e.g., carbon monoxide) as well as long term health threats (e.g., pm-10). An important reason for choosing an urban area is the fact that roughly 75 percent of the U.S. population live in urban areas (U.S. Department of Commerce, 1993). While a large portion of urban air pollution is caused by mobile sources such as cars or trucks, this study will examine stationary sources of criteria pollutants since stationary sources are nonmobile, most of these the emissions are relatively confined to one area. This makes air pollution from stationary sources more logical in evaluating siting decisions relative to residential populations. Stationary sources of air pollution include powerplants, manufacturing facilities and any process equipment which are located at one or more adjacent properties that emit or have the potential to emit one or more air pollutants. Thus stationary sources of air pollution are not as rare as hazardous waste sites and may provide a more comprehensive measure of exposure to environmental pollution.

Denver, Colorado was selected as the study area to illustrate the analytical approach proposed in this article. Denver has serious air pollution problems and is classified as a non-attainment area on several of the criteria pollutants. Denver also has 13% of its population as Hispanic, 6.5% of the population as Black, 2% as Asians and about 1% as Native American. These figures are well within the range for the U.S. as a whole, where Blacks range from 2% to 25% of the population depending on region of the country and Hispanics range from about 1% to in excess of 20% of the population in some regions. For the U.S. as a whole, about 16% of the population is non-white, while for Denver it is 22.5%. Thus, while we make no claims that Denver is a representative city or that our results can be generalized nationwide, it is a typical enough city in terms of its representation of minorities that it serves well to demonstrate the advantages of our empirical approach.

In the analysis to be discussed, sixty eight zip codes in the Denver metropolitan area were analyzed using Bureau of the Census Data for 1990.(1) By using a small area such as a zip code, there is likely to be less variance in the characteristics of a community than a larger area such as a county (Allen et al., 1995) or state (Lester et al., 1994). Zip codes were used in the United Church of Christ study, although both Anderton et al. and Been (1995) use Census tracts. There appears to be some debate over the advantages and disadvantages of the different units of analysis (see Been, 1995). We simply note that our multivariate regression approach can be applied using data from Census tracts as well as zip codes.

This study examines the characteristics of a community at the time when a stationary source was permitted by the State of Colorado for construction. This is a distinct advance over previous studies in which the relationship of all cumulative sites in an area to ethnic minorities was examined using the current census data regardless of when the sites were permitted or built. To establish the sequence of events, it is crucial to examine the socioeconomic characteristics of a particular area when the site was permitted for construction.


It is important to establish the variables which may be influencing the location of polluting sites. The race variables to be used in this study are Asian, Black, Hispanic [TABULAR DATA FOR TABLE 1 OMITTED] Origin and Native American. The variable "minority", which is comprised of the nonwhite percentage of the population, is also used as an aggregate race variable in some regressions in order to avoid the high correlations between the individual race variables on the right hand side (this is a common practice in this literature, see Been, 1995, p. 5 for an example). The number of polluting sites in a particular zip code area may be explained by several or a combination of many different socioeconomic variables.

Table 1 provides a list and definitions of possibly influential variables. The table also contains the expected sign of the variable in relation to the number of newly permitted sites in the zip code region.

Organizing the variables into equation form for estimation yields Equation 1:

Siteden = [B.sub.0] + [B.sub.1]Asian + [B.sub.2]Black + [B.sub.3]His + [B.sub.4]Namer + [B.sub.5]College + [B.sub.6]Homev + [B.sub.7]Inc + [B.sub.8]Pop + [B.sub.9]Rent(%) + [B.sub.10]Unemp (1)

The dependent variable (Siteden) is the number of newly permitted sites in a given zip code divided by the square miles of the zip code area. The number of sites, is measured by the number of new sites permitted for construction in years within two years of when the independent variables were collected by the Census Bureau (e.g., 1988 - 1992). This is critical because by using the characteristics of a community at the time the decision to build a polluting site was made, information regarding the siting process is uncovered. Another distinguishing feature of this study as compared to the few other multivariate analyses is the use of number of sites rather than simply presence or absence of sites. Been (1995) uses a logit analysis of presence or absence of hazardous waste sites. While this is a thorough multivariate analysis, the magnitude of environmental exposure is likely influenced by the number of these sites. Using the number of sites also utilizes more of the information on strength of the relationship than simply relying on a more qualitative indicator such as presence or absence.

The number of sites was divided by the area to create the new site concentration or density of newly permitted sites in order to control for the difference in the geographic size of the zip codes. An analogy to population density illuminates the importance of this transformation to site density. If, for example, one million new people moved to California, an impact would be felt. However, if one million people moved to Rhode Island, the relative impact would be much greater regardless of the existing population. The land area of California is approximately one hundred times that of Rhode Island which suggests there would be quite a difference in the effect of a migration of one million people. In our case, the range in land area for zip codes is from .8 of a square mile to 294.5 square miles, it is necessary to use the area to standardize the harm from pollution concentration.

There are three types of independent variables. The race variables are the key variables for testing whether ethnicity of the population at the time of siting has a significant influence on the density of newly permitted sites in an area. Variables such as College and Income are included for two reasons. First, including College and Income allows us to control for demographic influences when assessing the significance of the minority variables. If environmental racism is operating, then site density should be high in middle income minority communities as well poor minority communities. Second, inclusion of demographic variables allow us to test whether construction of new point sources of air pollution are being targeted at low income areas in general, irrespective of the racial composition of these areas. Significance of these variables would suggest "classism" rather than "racism" may be occurring.

Rent, or the percentage of renter occupied units is included in the model as a proxy for two influences. First, the higher the percentage of renters, the lower the percentage of homeowners. Since owner-occupied housing is most individuals single largest investment, there is often strong homeowner opposition to polluting sites likely to drive their property values down. Alternatively, renters tend to be more mobile, since they often have the ability to move on a months notice or at most have a one year lease. Renter's frequent mobility provides less attachment to a particular community. Thus the higher percentage of renters in an area, the greater we would expect site density to be.

Home value (Homev) would be expected to be negative as this variable serves as a proxy for land values in the area. Higher land values would make an area less profitable for industrial sites. Unemployment rate is expected to be positive because areas with high unemployment may welcome, rather than oppose, new industrial projects. If an area is receptive to new industrial projects (even if it is polluting industry) it may offer incentives for firms to locate there and would certainly not require expensive mitigation actions of new firms.

Market Dynamics -Post Site Activity

As noted previously it is also beneficial to observe the change in the race variables after a site has been permitted or constructed. If environmental inequity persists, either without discrimination or even with policy measures aimed at the siting process, there may be other factors which lead to the correlation between ethnicity and pollution. As discussed earlier, market dynamics in the form of cheaper housing or deeper rooted social problems such as housing and job discrimination may lead minorities into polluted areas after a site is built. In order to uncover information regarding market dynamics, the rate of growth of the race variables will be traced over time. Since the variables are defined as the race's percentage of the population, we do not need to make additional controls for the change in overall population. By using the minority percentage of the population, the change in the composition of the community is observed. The simple percentage change was calculated for each [race.sub.i] (where i = White, Asian, Black, Hispanic, and Native American) between the 1980 census data and that of 1990 as shown in Equation 2:

[Delta]% = % Population(1990)[Race.sub.i] - % Population(1980)[Race.sub.i] / % Population(1980) [Race.sub.i] (2)

The percentage change in the minority (or non-minority) percentages of the population can be compared in communities with no sites permitted in the initial time period, 1978 to 1982, to those communities with one or more sites at the initial time period. If the percentage change in minorities over the next decade is significantly greater in communities with sites, we would have evidence that shows that after sites are built or permitted, the minority percentage of the population tends to grow faster relative to areas without sites. However, if there is no significant difference or if the percentage change is greater in communities with no sites, no evidence of post site market dynamics is established. By examining a simple percentage change such as this we are not controlling for other trends such as changes in job opportunities, home value or average rent changes. This analysis simply observes the percentage change in the minority percentage of the population over time after a site is permitted or built and compares it to that of communities where no sites existed at the same point in time.


The data for the multiple regression model was retrieved from the zip code summary of the 1990 Census of Population and Housing Summary Tape File 3B on CD-ROM. The variables retrieved are listed in Table 1. Zip code area size was obtained from the Environmental System Research Inc. County Boundary Data File for 1995 (Environmental Systems Research Inc., 1995).

The data for number and location of stationary sources of criteria air pollutants within the Denver metropolitan area are available on the AIRS database produced by the Colorado Department of Public Health and Environment, Air Quality Division (CDPHE) in association with the EPA. The data extracted from AIRS is a listing of sites that have been issued permits for construction by the State of Colorado during the time period bracketing the 1990 Census (1988-1992). Because we are interested in the permitting of "new sites" that are sources of criteria air pollutants, any site that already had a pollution permit prior to 1988 was not included in our study as a new site, even if an additional pollution permit was issued in the 1988-1992 period.(2) Information regarding the AIRS database was downloaded from the Colorado Department of Public Health and Environment Bulletin Board on the Internet. The permitted sites were then located by the zip codes in the Denver Metropolitan Area. The 1990 census data on demographics were combined with the number of sites permitted for construction between 1988 and 1992.


In order to remain consistent with the previous work, it is necessary to establish that a correlation exists between ethnic minorities and polluting industry. Our results show the significant correlations of the different ethnic group's percentage of the population with newly permitted site density. The correlation with site density of Hispanic and Native Americans is .36 and .45, respectively. These are significant at the .01 level. The correlation on white is -.27, which is significant at the .05 level. Neither Black or Asian are significant (r = .06 and. 1, respectively). These results suggest that Hispanic and Native Americans are more likely than whites to be exposed to newly permitted sources of air pollution. Unlike the previous studies, a strong correlation between Blacks and polluting sites was not observed.

However, it is important to realize that there are also strong correlations of several other socioeconomic variables and polluting industry as well. College, Home Value and Income are negative and significantly (at the .01 level) correlated with permitting of new stationary sources (r = -.2, -.34, and -.58, respectively). Percent of housing units that are rentals (Rent) and the unemployment rate (Unemp) are positive and significant (at the .01 level) with new stationary sources (r = .57 and .46, respectively).

Like the previous literature, a correlation between ethnic minorities and pollutants was obtained. However, this correlation alone does not provide much insight into establishing an appropriate policy instrument since demographic variables are also highly correlated with site density. Since we used only the sites permitted in a time period surrounding the time of the census data, we can infer the communities had these, or relatively similar socioeconomic characteristics and demographics when the sites were permitted. At this point, we do not know which race or socioeconomic variables were the stronger determining factors. It is in this context that multiple regression is particularly useful. However, the regression analysis that follows will be limited by the strong correlation that exists between the independent variables such as minority and the demographic variables. This will introduce some multicolinearity in the regression. Despite our effort to carefully select demographic variables that were not correlated with each other, the multicolinearity is likely between some of the demographic variables such as Income, Home value and Rent. While multicolinearity does not result in biased coefficient estimates it does increase the variance often increasing the likelihood of insignificant coefficient estimates (Kmenta, 1986). Several of the leading econometrics texts have asserted that correlations of less than .8 are not severe and need not be dealt with in any particular manner (Greene, 1990; Kennedy, 1984). Nonetheless, some of the coefficients and t-statistics reported below are influenced by multicollinearity, but other than the minority variable, the signs of the variables are intuitively correct.


The Box-Cox regression model was used to transform the right hand side variables in equation 1 and estimate the most appropriate functional form (Greene, 1990). The Box-Cox transformation indicated a linear model as the best functional form, given the linearity of the dependent variable. Regressions are reported in Table 2 for 12 different [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] models beginning with minority as the only independent variable. Other socioeconomic variables were then individually added to observe the additional explanatory power associated with the inclusion of each of these variables.

The significance, explanatory power and elasticities associated with the results in Table 2 are given in Table 3. First, we discuss the influence of multicollinearity. From Table 2, note the sign on the minority variable changes from positive to negative when variables such as Income and/or Unemp are included in the regression. One possible reason for this is the high correlation of these variables to the minority variable. Minority has a relatively high correlation with all variables except for Population and Rent. However, the sign of the estimated coefficient on Minority only changes when the correlated variable(s) is significant in the regression. For example, the second regression shows College as an insignificant variable. Even though College is as highly correlated with Minority as Income is, since College is not significant in the regression it does not affect the sign of Minority which is a significant variable. Nevertheless, even though multicollinearity leads to less reliable coefficient estimates and t-statistics, the explanatory power of the variables is not affected (Kmenta, 1986).

Significance, Explanatory Power and Elasticities

Table 3 provides the variables, the additional explanatory power by including only that variable with Minority, the consistency of the significance of the variable (given by the t-statistic range), and the calculated elasticity for the significant variables. Elasticity was calculated for variables with significant coefficient estimates using the regressions from model 11 and 12 in Table 2. These fully specified models should avoid or lessen the omitted variable bias that may be present in models 1-10. The formula to calculate elasticity is given in Equation 3:

[Epsilon] = ([Delta]Y/[Delta]X) * [X.sub.m]/[Y.sub.m] (3)

where ([Delta]Y/[Delta]X) is the coefficient estimated from the regression and [X.sub.m] and [Y.sub.m] are the means of X and Y, respectively.

Income, Rent and Unemp add the most explanatory power to the regressions. Rent and Unemployment are both positive and consistently significant at the 95% level for all models in which they were included. The elasticity estimate for income, obtained from Model 11 is -6.97. A 1% rise in income indicates a decline in new site density of approximately 7%. Unemployment had the highest elasticity indicating a one percent change in the unemployment rate would increase the site density of newly permitted [TABULAR DATA FOR TABLE 4 OMITTED] sites by over 8%. The unemployment elasticity is much higher than that of Rent which is .5 and .6 in Model 11 and 12 respectively. A negative elasticity for Minority was calculated using Model 12 which would indicate an increase in the minority percentage of the population would decrease the new site density. However, this estimate is counter-intuitive, and the result appears to be influenced by the strong multicollinearity.

The results of the estimation of these models indicates that low-income, higher unemployment and higher renter occupied areas attract more polluting industry. Low income areas are often associated with cheaper land value which would attract more developers of polluting sites. A higher percentage of renter occupied units could be inviting to developers of locally unwanted land uses since less homeowner opposition would be anticipated. Also, the higher rates of unemployment may cause residents to welcome new job opportunities. The communities receiving the bulk of new sites may have a greater than average minority percentage of the population, however, it does not appear to be the driving force in explaining the density of new sites. The high correlation of minorities to other socioeconomic factors may be causing the high but apparently spurious correlation with new sites and pollution, if the socioeconomic factor in question is a significant variable in explaining the location of new sites.

All of the models were also estimated using the four individual race variables. Table 4 provides a summary of the results. Income and Rent are the most significant. The Native American percentage of the population is the only significant variable and as with the minority variable, it is significant only in cases when it is not highly correlated with another significant variable. It is important to note that there is a correlation between Native American and Hispanic of .67. This may be affecting the t-statistics of the Hispanic and Native American variables.

Elasticities, consistency of significance, and explanatory power are summarized in Table 5. The additional explanatory power from including the socioeconomic and demographic variables is lower than in the regressions using minority as the race variable. Using the four individual race variables involves using correlated variables but provides more explanatory power than the general measure of minority. Nonetheless, the elasticities on unemployment and income are still the largest.



Throughout estimation of the models with the different measures of ethnicity, unemployment remained the most consistently significant variable and maintained the largest elasticity. Income and Rent also appear to play a vital role in the density of new sites and College and Home value, possibly to a lesser extent. We now turn to an analysis of events occurring after a site was constructed.


The second hypothesis regarding the cause of environmental inequity involves market dynamics. Table 6 shows the racial composition of the communities between 1980 and 1990 for areas with no sites permitted and areas with one or more sites permitted. The percentages sum to more than 100% due to the Census Bureau's definition of Hispanics as "Hispanic" origin of any race which includes Asians, Blacks, Whites, or Native Americans of Hispanic origin. Table 6 also displays the percentage change of the racial composition between 1980 and 1990.

There is a higher percentage of minorities living in zip codes with one or more sites as compared to zip codes with no sites. To test whether siting was attracting minorities into these areas, equation 2 was used to calculate %[Delta], the rate of growth of the ethnic group's percentage of the population between 1980 and 1990 for zip code areas with zero sites permitted for construction between 1978 and 1982 and areas where one or more sites were permitted in the same time period. These figures were calculated to observe if the minority percentages of the areas grew faster after sites were built as compared to areas where no sites were constructed (e.g., our test of market dynamics leading minorities to polluting areas).

Table 6 shows that the percentage of the population for Blacks and Hispanics grew faster in communities with no sites built in the initial time period than in those zip code areas where polluting sites were located. The percent increase in Native American percentage of the population was virtually identical in zip code areas with one or more sites and zip codes with no sites in the Denver metropolitan area. These results show no indication of market forces leading minorities into areas after the sites were permitted for construction. In fact, just the opposite is happening. Communities without sites in the initial time period actually show a faster and larger increase in the Black and Hispanic percentage of the population. While areas with one or more sites have lower than average increases in the minority percentage of the population, the non-minority (white) variable showed a bit of a possible effect of market dynamics. The percentage of the population of whites declined slightly faster in communities with one or more sites. This is particularly interesting since the minority percentage of the population was growing faster in the areas with no sites. If the overall minority percentage of the population increased, it must be inferred that the white percentage of the population decreased. The fact that the percentage of the population that is white is going down at a greater rate in areas where there were sites built may be an effect of the polluting industry.


From this analysis, evidence of market dynamics leading minorities to polluting sites was not found in Denver. It appears to be the case that sites are being permitted in Denver where minorities are located but not necessarily because of racism in the siting process. Other socioeconomic and demographic factors, which tend to be correlated with the minority percentage of the population, appear to be stronger determinants of the location and density of new sites in our study area. Factors such as less homeowner opposition, less political opposition due to the creation of new jobs, and possibly lower land costs may be drawing firms to locate polluting sites into these areas. It does not appear that racism, as a basis for the location of new stationary sources of criteria air pollutants in the Denver metropolitan area, is prevalent.


In summary, we illustrated an empirical approach that overcomes some of the shortcomings of past correlational or univariate studies that simply relate ethnicity and historic number of hazardous waste sites. Our multivariate approach controls for demographic and economic variables when attempting to relate the number of newly permitted stationary sources of air pollution to minority status of the population at the time the permitting decision was made. Applying this multivariate approach to Denver, Colorado, we found that socioeconomic and demographic factors other than race have the most explanatory power in determining the location of newly permitted stationary sources of air pollution. If our findings are replicated in other communities and for other sources of pollution, then it may be the case that site owners are choosing communities based on efficiency reasons such as low cost land and less homeowner and political opposition. Nonetheless, the decision to locate sites based on efficiency fails to account for the undesirable distributive effects of such policy. The correlation of minorities to the other socioeconomic conditions such as low income, low education levels, unemployment, less homeownership may indicate a pattern of deeper rooted racism. For example, the correlation of minorities to lower economic conditions could come from discrimination in labor markets, educational institutions, or bank credit processes. Thus, even if racism is not present in the siting process, it is the inequitable effect on minorities which establishes the need for policy measures that involve equity concerns such as Executive Order 12898. This policy measure may enhance less fortunate people's right to choose their level of exposure to pollution. However, in the long run, raising income and education levels are

the primary vehicles to provide more meaningful choices to all people, regardless of ethnicity.

Acknowledgment: The authors wish to thank Robert Kling of Colorado State University and Gloria Helfand of University of Michigan for their suggestions. Steve Davies, Colorado State University provided valuable assistance with the econometrics. All these helpful individuals are absolved from what follows which is the sole responsibility of the authors.


1. An analysis was also performed using the 1980 Census Data. However, this analysis is somewhat more limited since not all of the variables needed were available in the 1980 data. The results are qualitatively similar to what was found with the 1990 data. See Sabina Shaikh. "An Examination of the Presence and Causes of Environmental Inequity in Denver, Colorado." (Master Thesis, Colorado State University, 1995) for more details.

2. Because the AIRS database only records permitted sources of criteria air pollutants since 1972 it is possible that we interpreted air pollution permits issued in the 1988-1992 time period to pre-1972 sites as "new" sites. This situation would arise if a pre-1972 source did not make any modifications or expansions requiring a permit during the 1972 to 1988 time period and then made changes that required a permit in the 1988-1992 time period. Misinterpreting additions to pre-1972 plants as new plants reduces our ability to link recent air pollution permitting decisions to ethnicity at the time of the decision since the original plant location decision may have been made decades before. The reader should keep this limitation in mind when judging the results that follow. We appreciate an anonymous referee's suggestion to describe how a "new" site was defined in this study.


Allen, D., J. Lester, and K. Hill. (1995). Prejudice, profits and power: Assessing the eco-racism thesis at the county level. Paper prepared for the 1995 Annual Meeting of the Western Political Science Association. Department of Political Science, Colorado State University, Fort Collins, CO.

Anderton, D., A. Anderson, J. Oakes, and M. Fraser. (1994). Environmental Equity: The Demographics of Dumping. Demography, 31:229-247.

Been, V. (1994). Locally Undesirable Land Uses in Minority Neighborhoods: Disproportionate Siting or Market Dynamics? The Yale Law Journal, 103:1383-1421.

-----. (1995). Analyzing Evidence of Environmental Justice. Journal of Land Use and Environmental Law, 11:1-36.

Bullard, R.D., (1983). Solid Waste Sites and the Black Houston Community. Sociological Inquiry, 53: 273-288.

Bunyan, B. and P. Mohai. (1992). Race and the Incidence of Environmental Hazards: A Time for Discourse. Boulder, CO: Westview.

Clinton, W. (1994). Presidential Documents: Executive Order 12898 of February 11, 1994: Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations. Federal Register, 59:7629-7633 Washington DC.

Colorado Department of Public Health and Environment (CDPHE), Air Quality Control Commission (1995). Regulation Number 3, Concerning general provisions applicable to construction permits and operating permits. Downloaded from the CDPHE bulletin board on the Internet.

Environmental System Research Inc. (1995). ESRI ARC VIEWII county boundary data file. Boulder, CO.

Greene, W. (1990). Econometric Analysis. New York: Macmillan.

Hamilton, J.T. (1995). Testing for Environmental Racism: Prejudice, Profits, Political Power Journal of Policy Analysis and Management, 14:107-132.

Kennedy, K. (1984). A Guide to Econometrics. Cambridge, MA: MIT Press.

Kmenta., J. (1986). Elements of Econometrics, 2nd edition. New York: MacMillian Publishing Co.

Lester, J., D. Allen, and D. Lauer. (1994). Race, class and environmental quality: An examination of environmental racisim in the American states. Prepared for the 1994 Annual Meeting of the Western Political Science Association. Dept of Political Science, Colorado State University, Fort Collins, CO.

Shaikh, S. (1995). An examination of the presence and causes of environmental inequity in Denver, Colorado. Master Thesis, Colorado State University.

United Church of Christ, Commission for Racial Justice. (1987). Toxic Wastes and Race in the United States: A National Report on the Racial and Socioeconomic Characteristics of Communities with Hazardous Waste Sites. New York: UCC.

U.S. General Accounting Office. (1983). Siting of Hazardous Waste Landfills and Their Correlation with Racial and Economics Status of Surrounding Communities, RCED-83-168. Washington DC.

U.S. Department of Commerce. (1993). Census of Population and Housing, 1990: Summary Tape File 3B on CD-ROM. Washington DC.
COPYRIGHT 1999 Elsevier Science Publishers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Shaikh, Sabina L.; Loomis, John B.
Publication:The Social Science Journal
Date:Jan 1, 1999
Previous Article:Gains and losses of migrants and income through intercounty migration in the U.S., 1992-1993.
Next Article:Rational choice and out-of-wedlock births among African American females: the influence of AFDC and unemployment.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters