Citizen Complaints, Regulatory Violations, and Their Implications for Swine Operations in Illinois

Since 1979, the Illinois Environmental Protection Agency (EPA) has operated a livestock waste management program that provides for inspection of livestock production facilities throughout the state. Generally, Illinois EPA inspections are initiated either by citizen complaints or by regularly scheduled random selection of facilities. Illinois EPA takes citizen complaints seriously and responds by sending its agricultural engineers for a prompt site inspection. Citizen complaints can be filed in writing, by telephone, or on the Agency's website. Illinois EPA data document that most facilities were inspected because of citizen complaints.

After corn and soybeans, hogs are the third largest agricultural commodity produced in Illinois. Swine facilities accounted for 62% of the total livestock facilities inspected during 1997–2001 and citizen complaints prompted 59% of the swine facility inspections. From 1997 through 2001, 157 Illinois swine facilities received odor complaints; 180, water pollution complaints; and 81, both odor and water pollution complaints (figure 1).

Citizen complaints may indicate possible noncompliance with environmental and livestock waste regulations, complainers' concerns over the potential impact of the facilities on their health and/or property values, a perceived influence on quality of life, and/or just an odor nuisance. Avoiding citizen complaints is vital to the sustainable development of swine production. This has become particularly important because of the rapid increase in the size of swine operations and the geographic concentration of production over the past two decades. To date, a substantial amount of data has been accumulated from Illinois EPA site inspections regarding the characteristics and regulatory compliance status of operations, along with other details. These data, obtained from all inspections (those as a result of complaints and those conducted on a regular schedule), represent a valuable source of information about factors that may cause citizen complaints and facility regulatory violations.

The purpose of this paper is to explore relationships between facility characteristics, citizen complaints, and regulatory violations using the citizen complaint and Illinois EPA inspection data and suggest implications for the swine industry. To do so, appropriate statistical and economic analyses are used in identifying, assessing, and modeling these relationships. The possible linkages between facility characteristics, citizen complaints, and regulatory compliance are examined using the statistical analysis of categorical data. A latent dependent variable model is used in delineating a facility's regulatory compliance behavior.

This study contributes to the existing literature by providing the first extensive empirical investigation of the regulatory enforcement and compliance issues related to livestock production operations. Our empirical results provide a broad context of various factors that may influence citizen complaints, environmental regulatory enforcement, and compliance and hence, produce valuable policy implications relevant for livestock producers and regulatory enforcement agencies.

Background

The economics of law enforcement and compliance has been an active area of research since Becker's seminal work was published in 1968 (Becker, Polinsky, and Shavell). The enactment of major environmental regulations and policies in the United States and elsewhere in the early 1970s, further stimulated the need to investigate the enforcement and compliance issues involved and the challenges imposed upon environmental regulatory enforcement agencies. A significant body of the economic literature exists on environmental regulatory enforcement and compliance, focusing primarily on theoretic analyses of the norms and behaviors of concerned parties (Heyes; Cohen, 1999). There are relatively few empirical studies; however, they are limited to industries such as oil and pulp and paper where comprehensive data on compliance and enforcement are available (Cohen, 2000).

Feinstein (1989); Helland; Smith; and Eckert are among those empirical studies of methodological relevance. Feinstein (1989) constructed statistical models to study the factors associated with regulatory noncompliance of U.S. nuclear power plants, the variation in detection rates among the Nuclear Regulatory Commission inspectors, and the relationship between undetected violations and abnormal occurrences. Helland used models similar to those proposed by Feinstein (1989) to examine the role of inspections in producing regulatory compliance and self-reporting in the pulp and paper industry under the Clean Water Act. Smith compared productivity of worker-complaint initiated versus generally scheduled inspections conducted by the Occupational Safety and Health Administration (OSHA). The study found that these two types of inspections were similarly productive in detecting safety violations in 1977–79. More recently, Eckert examined the effect of inspections and warnings on environmental regulation enforcement at petroleum storage sites in Manitoba, Canada, using a two-stage probit model. The author showed that though inspections deterred future violations, this effect is small.

There is a general dearth of research on environmental regulatory enforcement and compliance in agricultural sectors in general and in livestock industries, in particular. Using an analytical model of regional livestock production, Innes examined the efficiency of different environmental regulations such as scale regulations, chemical fertilizer taxes, and waste storage standards in controlling spills from animal waste stores, nutrient leaching and runoff from manure land application, and ambient pollution from livestock facilities. Results of the study indicate that varying types of facilities and sites need to be regulated differently to maximize social welfare. Recently, Mullen and Centner theoretically investigated the environmental effect of the number of regulated confined animal feeding operations (CAFOs), monitoring effort, and compliance standards. Their results show that while there is no doubt about the positive environmental effect of greater monitoring effort, increasing the number of regulated CAFOs alone may not achieve a desired improvement in environmental quality. Moreover, the impact of raising compliance standards to improve environmental quality may depend on how noncompliance is prosecuted and penalties for violation are set. Though facility inspection and environmental regulatory enforcement have become a major activity of the EPA, we have not seen any empirical studies addressing issues related to environmental regulatory enforcement and compliance in livestock industries.

Dasgupta and Wheeler assessed factors affecting citizen environmental complaints using an econometric model estimated with a Chinese provincial panel data set. To our knowledge, there is little formal research on the causes and implications of citizen complaints against swine or other livestock operations. In an earlier study, Hardwick counted the number of livestock facilities in the United Kingdom that were causing justifiable odor complaints and found that among 1,820 pig, cattle, and poultry farms, 46% of the complaints were associated with manure land applications, 25% with building odors, and 19% with manure storage (Bradshaw). Recently, Messenger reported a survey of Iowa pork producers conducted by Kliebenstein and Lorimor, who found that 21.7% of the 354 producers responding had received a complaint in the past five years. Their preliminary results show that complaints were not necessarily related to farm size and that neighbors within 1/8 to 1/2 miles filed more complaints than those living farther away. However, these findings were limited because they might be subject to potential response bias (e.g., producers who had received complaints might be more or less likely to respond to the survey than those with no complaints) and lacked adequate statistical evaluation.

The remaining sections are designed to answer the following questions: Are citizen complaints and regulatory violations related to production characteristics of swine facilities such as operating capacity and the type of manure storage? Are citizen complaints related to characteristics of the surrounding communities and their citizens such as education, income level, and property values? Are complaint-initiated inspections as effective as regularly scheduled inspections in detecting air and water regulatory violations of the facilities and what are the implications of this analysis on the EPA's inspection resource allocation? What are some of the factors that may influence a producer's likelihood of having a regulatory violation?

Facility Characteristics, Citizen Complaints, and Regulatory Violations

The characteristics of a livestock facility recorded in an Illinois EPA inspection include types of livestock raised or boarded; operating capacity in terms of the National Pollutant Discharge Elimination System (NPDES) defined animal units (AU)¹; types of livestock waste storage structures; number of lagoons or outside holding ponds; types of building structures (total confinement or others); and existence of a concrete settling basin. Specifically, Illinois EPA categorizes the operating capacities of livestock facilities into six groups, ranging from less than 50 to more than 7,000 AU. Since operations with a capacity of 1,000 or more AU are subject to more restrictive environmental and livestock waste management regulations, we regrouped inspected swine facilities into two capacity categories: less than 1,000 AU or more than 1,000 AU (table 1).

Similarly, based on the available inspection data, facilities are also divided into two categories using the following pairs of nominal variables: with at least one or with no lagoon/holding pond; consisting of total confinement buildings only or otherwise; with or without an open feedlot; and with or without a concrete settling basin. In addition to categorization based on these characteristics, facilities are also grouped by whether there are complaints against them and/or they are in regulatory violation. Specifically, facilities are categorized according to whether they receive complaints about odor, water pollution, or either of these two issues. Regulatory violations are distinguished by air emission, water pollution, and any regulatory violations.²

We assume that inspected facilities that are not receiving complaints are chosen without regard to specific facility characteristics. This assumption is reasonable since in most cases, the Illinois EPA inspectors do not have information about a facility's production characteristics prior to their visit regardless of whether the inspection is complaint-prompted or not. Statistical analysis of categorical data is used to assess the relationship between pairs of variables, that is, the relationship between the column variables, such as facilities receiving an odor complaint or receiving no odor complaint and row variables, such as operating capacity less than 1,000 AU or greater than 1,000 AU. The null hypothesis (H₀) of no association between the row variable and the column variable was tested using the Pearson chi-square test.³

Table 2 shows that operating capacity greater than 1,000 AU was statistically associated with more odor complaints (χ² = 21.84, p < 0.001) and more air emission violations (χ² = 11.52, p < 0.001). However, capacity less than 1,000 AU was associated with more water pollution complaints (χ² = 6.51, p = 0.011), water regulatory violations (χ² = 20.66, p < 0.001), and overall regulatory violations (χ² = 5.09, p = 0.024). It is not unreasonable that small operations may have more water pollution complaints and violations because only animal feeding operations of 1,000 AU or greater are required to prepare and maintain waste management plans (Illinois Department of Agriculture). These results suggest that waste management plans may affect water pollution control and extension of this policy to small operations may reduce their water pollution complaints and violations.

Contrary to what we anticipated, outside lagoons/holding ponds were associated with fewer odor complaints (χ² = 5.19, p = 0.023), that is, facilities with no outside lagoons/holding ponds were more likely to receive odor complaints though there was no difference in air emission regulation compliance between these two types of facilities. However, we found no association (p > 0.05) between waste storage type and other complaints/regulatory violations. A plausible explanation is that facilities with lagoons or outdoor holding ponds may be less odor-offensive because they are located relatively farther away from residential homes or communities. In Illinois, setbacks for livestock facilities are regulated based only on facility capacities and community population densities (Illinois Department of Agriculture). Our results suggest that in order to alleviate the potential conflict between livestock producers and local communities, facility characteristics such as waste storage type may also be appropriate to consider when establishing setback regulations.

Facilities with only total confinement buildings were associated with more odor complaints (χ² = 25.96, p < 0.001) and air emission violations (χ² = 4.35, p = 0.037), but with fewer water complaints (χ² = 9.40, p = 0.002) and water regulatory violations (±² = 66.15, p < 0.001). When both odor and water pollution complaints and air and water violations were considered, total confinement facilities were related to more citizen complaints (χ² = 4.16, p = 0.041) while facilities other than total confinement were related to more regulatory violations (χ² = 39.31, p < 0.001). This suggests that compared with nontotal confinement facilities, total confinement facilities were more of a nuisance than a real source of pollution to local communities. Facilities with open feedlots were associated with more water pollution complaints (χ² = 11.66, p < 0.001), increased water pollution violations (χ² = 71.75, p < 0.001), and more overall regulatory violations (χ² = 45.19, p < 0.001). However, facilities with no open feedlots were associated with a higher number of odor complaints (χ² = 25.71, p < 0.001), more air emission violations (χ² = 4.02, p = 0.045), and probably more overall citizen complaints (χ² = 2.91, p = 0.088). Finally, facilities with a concrete settling basin were associated with fewer odor complaints (χ² = 7.55, p = 0.006) but more water and overall regulatory violations (χ² = 16.08, p < 0.001; and χ² = 8.24, p = 0.004, respectively), suggesting that careful management is more important for regulatory compliance than facility characteristics since concrete settling basins are technically superior to nonconcrete settling basins.

In summary, small operations, nontotal confinement operations, and facilities with an open feedlot were generally associated with more water pollution and other nonambient emission regulation violations and complaints. Large operations, total confinement facilities, and those with no open feedlots were related to more air emission violations and complaints. Such information is valuable for regulatory enforcement agencies, particularly when inspection resources are limited. Educational and regulatory efforts to achieve compliance can proceed with this information.

Citizen Complaints and Community Characteristics

Economic theory suggests that citizens are more likely to complain to authorities about pollution when the expected benefits from agency action are likely to exceed the expected costs for their own investment of time and effort. According to Dasgupta and Wheeler, factors affecting citizen complaints in a region include pollution damage suffered by the individual, the individual's understanding of the problem (which is assumed to be a function of education), and the cost of a complaint (assumed to be a function of income).

The relationship between citizen complaints and community characteristics is assessed using county-level data and an econometric model with the proportion of swine production facilities receiving complaints as the dependent variable (table 3). Among the independent variables, average swine operation scale (ASOS), swine inventory intensity (SII), and soil productivity rating (SPR) are used to capture the pollution damage potential caused by swine production; proportion of residents with a high school diploma or higher as a proxy for education attainment of the residents in a county, to represent the individual's understanding of the problem; and median household income in 2000, to capture the cost of a complaint. Other county characteristic variables such as distance to the nearest city with a population over 50,000, rural-urban continuum code (Beale code), average farmland price, average home price, population density, and proportion of residents aged sixty-five and over also were tried but eliminated from the analysis because of collinearity problems or statistical insignificance. For further description of these variables, see Huang et al.

Table 4 shows that the three models of citizen complaints produce similar estimation results and all are statistically significant. In addition, the signs of the coefficients are, in general, as expected. Higher SII leads to a higher percentage of swine facilities being complained against. This finding is consistent with our expectation that increased swine production intensity (number of hogs per square mile) may generate greater environmental pollution and hence a higher proportion of complaints. The estimated coefficient of the ASOS bears a negative sign; that is, given the number of hogs in a county, more concentrated production was associated with a lower proportion of facilities receiving citizen complaints. One possible explanation for this result is that as average operation scale increases, there is less overall environmental pollution at the county level and therefore a lower proportion of facilities receiving citizen complaints.⁴ In addition, most large operations are relatively new and rely on more advanced production technologies that may be less offensive to local communities.

Our results also show that higher SPRs were related to a larger proportion of facilities receiving complaints. SPRs were used to capture pollution damage potential from swine production. Therefore, higher soil productivity means greater marginal pollution damage. On the other hand, owners with higher SPRs have a greater incentive to protect their land. Moreover, consistent with economic theory, a higher household income tended to cause a lower proportion of complaints about facilities because the opportunity cost of complaints was greater in high-income counties. However, citizens' education level did not have a statistically significant influence on their environmental complaint behavior, though the estimated coefficients suggest that more educated citizens might have a greater tendency to complain in most cases. Finally, it is worth noting that the adjusted R² in all three estimated equations is low (less than 0.2), suggesting that factors affecting citizen complaints are far more complicated than what we have modeled and more theoretical and empirical research is needed.

Efficiency Comparison of Complaint-Initiated and Regularly Scheduled Inspections

The usefulness of citizen complaints for regulatory enforcement agencies to allocate inspection resources is controversial (Smith; Dasgupta and Wheeler). One view is that complaints are undoubtedly a source of low-cost information, since pollution and regulatory violations of a facility are often apparent to their neighbors even if they are invisible to governmental agencies. Conversely, complainers may lack sufficient information to distinguish between a nuisance and a true regulatory violation. In addition, some individuals or communities may have a higher propensity to complain than others, regardless of the objective situation. Therefore, if agencies respond to complaints, aggressive complainers may capture most of the available resources.

As noted, the Illinois EPA was responsive to each complaint with a site inspection and complaint-initiated inspections composed a majority (59%) of the agency's swine facility inspections. In order to assess the relative efficiency of complaint-initiated versus regularly scheduled inspections, swine facility inspections are divided into odor complaint initiated, water pollution complaint initiated, both odor and water pollution initiated inspections, and regularly scheduled inspections. The efficiencies of these four types of inspections in detecting different regulatory violations were compared using the statistical analysis of categorical variables as described earlier. Table 5 shows the specific regulatory violations that were examined and the related inspection data summary on which the analysis was based. Table 6 presents the related hypotheses and statistical test results.

Compared with regularly scheduled inspections, odor complaint initiated inspections were more efficient in detecting air emissions violations but less efficient in detecting various water pollution-related violations. On the other hand, water pollution complaint initiated inspections were more efficient than regularly scheduled ones in detecting violations in water quality standard, runoff control requirements, manure handling/storage requirements, and field application criteria. However, there is no statistical difference between a water complaint and a regular inspection in detecting violations in effluent standards and air emissions. Between regularly scheduled inspections and those prompted by odor and water pollution complaints, the latter show a higher efficiency in detecting air emission and field application criteria violations. There is no statistical difference between the two in detecting water quality standard, effluent standard, runoff control requirement, and manure handling/storage requirement violations. Our results tend to support the view that citizen complaints are a valuable source of information. Inspections prompted by citizen complaints were more likely to identify facilities with violations than regularly scheduled visits, suggesting that the Illinois EPA has responded properly if the goal is to detect violations.

Factors Affecting Regulatory Violations

According to the theory of rational crime (Becker), a profit-maximizing facility will violate an environmental regulation as long as the compliance cost exceeds the expected penalty of noncompliance. The basic premises of this theory help formulate appropriate variables to include in the analysis and to interpret the results, even though most swine producers likely abide by the existing laws and regulations. Following this theory, three factors influence a facility's regulatory violation: the cost of compliance, the cost of the penalty, and the likelihood of the penalty. However, such data are usually unavailable.

In this analysis, a facility's production characteristics such as operating capacity and type of waste storage, are used as a proxy for compliance cost because they are important determinants of compliance cost. The expected economic penalty of noncompliance for swine facilities in Illinois can be a fine and the cost of compliance.⁵ Again, due to the lack of data, we use community characteristics to proxy the expected economic penalty of a violation. This treatment is reasonable as inspections and the stringency of environmental enforcement are usually determined by the economic situation of the surrounding communities (Helland). Other factors that may contribute to noncompliance include the history of past violations and the difference in inspectors' ability to detect violations. Noncompliance history is important because often violations can only be corrected with a capital investment such as the installation of a new abatement technology (Helland). In our analysis, we use the number of on-site visits by EPA staff as a proxy for the violation history of a facility and we expect that past violations are positively related to current violations (Magat and Viscusi). It has been documented that inspectors may differ substantially in their detection of violations for various reasons (Feinstein, 1989). Inspectors not only determine whether a violation exists, but also influence a facility's compliance behavior.⁶ Hence, the probability of a violation can be modeled as a latent variable that is a function with the following form as suggested in the regulatory violation literature (Helland; Feinstein, 1989):

(1)

where Y_j is the observed binary variable as defined above; X denotes an array of variables likely to affect the probability of a facility's regulatory compliance as discussed above; β is a vector of parameters to be estimated; u_j is a disturbance term representing unobservable facility and community characteristics and factors that affect the costs and benefits of compliance; and j is a subscript index for facility. Assuming that u_j is normally distributed, model (1) therefore becomes a probit model that can be estimated using conventional maximum likelihood techniques.⁷

In specification, facility characteristic variables include current operating capacity in terms of the NPDES-defined AU, number of outside lagoons/holding ponds, building type (total confinement or otherwise), and type of settling basin (concrete or otherwise) (table 7). Feedlot type (open feedlot or otherwise) is omitted due to its high correlation coefficient (0.96) with building type. Community characteristic variables include distance to the nearest city over 50,000, rural-urban continuum code (Beale code), population density, annual household income, education of the residents, SII, and average scale of swine operation.⁸ Other variables include number of on-site visits by EPA staff, investigator, and year when violations occurred, hoping to capture the trend over time in facilities' compliance behavior.

To alleviate concerns about statistical sampling issues that may arise from the inspection data, we divide them into one category consisting of all the facilities that were receiving complaints and the other of all the facilities that were inspected based on a regular schedule. The former represents a complete population of facilities that were complained against while the latter is assumed to be a random sampling from a large population of facilities not subject to complaints. The model was separately estimated for the two data sets using the SAS probit procedure (table 8).

Our results show that the probability of regulatory violations of a facility significantly depended on the EPA inspector. This is consistent with the existing literature that an inspector's ability to detect a violation or her strictness in regulatory enforcement can substantially influence a facility's compliance behavior. This result suggests that a random assignment of inspectors to cases may achieve more effective regulation enforcement. The probability of violation might decrease over time but the trend was not statistically significant. As expected, the coefficient for number of visits by EPA staff was positive and significant, indicating that violations take time to correct and that follow-up inspections, as pursued by Illinois EPA once a violation was detected, were necessary for compliance assurance monitoring.

Similar to our finding regarding the association between citizen complaints and operating capacities (see table 2), the probability of violations was independent of a facility's operating capacity. Larger operating capacity might be expected to be associated with higher compliance cost and hence, greater violation probability. However, the literature on regulatory compliance also suggests that larger facilities may be more likely to comply because of the lower cost per unit of emission removal when there are economies of scale (Gray and Deily). Therefore, the influence of operating capacity on compliance behavior is an empirical issue. We did not find evidence that the capacity of a facility was associated with the probability of violation. Also, we did not find evidence that the number of lagoons/holding ponds was related to the probability of violation. However, we did find evidence that total confinement facilities tended to have a lower probability of violation in both study populations. This finding is not surprising since total confinement facilities are usually new and better equipped, suggesting that total confinement facilities may have a lower compliance cost. Incidentally, this finding is in accordance with an earlier result that nontotal confinement facilities were associated with more regulatory violations (see table 2). Our results regarding the impact of concrete settling basin on the probability of violation were mixed: among the EPA-selected facilities, a concrete settling basin was associated with a lower probability of violation. Among the citizen-complained facilities, a concrete settling basin was possibly associated with a higher probability of violation, although this association was not statistically significant.

The community characteristics included in our analysis exhibit no statistically significant influence on the violation probability of the facilities inspected on a regular schedule. Among the facilities with citizen complaints, our results show that facilities located in counties with higher SII also have a higher probability of violation. If we assume that community characteristics partially capture the expected penalty of violation, it is reasonable to argue that the expected violation penalty could be less severe in major hog producing counties than elsewhere. Another interesting finding is that there was significant evidence that the income level of the communities was not associated with the probability of violation.

In general, our results show that a facility's compliance behavior was not obviously affected by community characteristics. The existing literature generally supports that community characteristics such as income and education level may influence industrial firms' regulatory compliance behavior (Pargal and Wheeler; Cohen, 1999). The absence of this influence in our empirical results may partly be due to the use of community data aggregated at the county level, which may not be specific enough to reflect the difference in pressure a local community places upon an animal feeding operation. Future research based on more location-specific data is needed.

Conclusions

Our statistical and economic analyses constitute an extensive empirical description of the relationships among citizen complaints, swine production and community characteristics, EPA inspections, and regulatory violations. We found that hog operations above 1,000 AU were related to more odor complaints and air emission violations. Operations of less than 1,000 AU were associated with more water pollution complaints and violations, suggesting that extending the waste management plan policy to smaller animal feeding operations might be beneficial. Also, facilities with outside lagoons/holding ponds were associated with fewer odor complaints. However, there was no difference in regulatory compliance between facilities with or without lagoons/holding ponds. Moreover, total confinement facilities were associated with more citizen complaints despite the fact that they were more likely in regulation compliance than nontotal confinement facilities. These findings suggest that setbacks may need to be regulated not only on production capacity and community population but also on other facility characteristics.

Citizens in counties with high hog production intensity and low production concentration tended to complain more than people in counties with more concentrated production. Together with recent empirical results of studies of animal feeding operations on property values (Huang et al.; Ready and Abdalla; Herriges, Secchi, and Babcock), policies encouraging concentrated production may be more desirable from an overall social welfare perspective. Other interesting results regarding citizen complaints included a positive effect of SPRs and a negative impact of income levels on complaint rates, suggesting that it may be advisable for producers to locate production in counties with lower SPRs and higher income levels. Our results also identify the need for more theoretical and empirical research to better our understanding why citizens complain.

Our results support the view that citizen complaints were valuable information for enforcement agencies to allocate inspection resources since complaint-initiated inspections generally were more efficient in terms of regulatory violation detection rates than regularly scheduled inspections. Our empirical results concerning facility compliance behavior agree with the existing literature that EPA inspectors influence a facility's probability of violation. Such influence may result from an inspector's ability to detect a violation or her strictness in enforcement. To mitigate this influence, it may be helpful to provide regular training for enforcement staff and randomly assign staff to field inspections. Our results also suggest that follow-up inspections were useful in violation correction and compliance assurance monitoring. In addition, among the facility characteristics examined, total confinement facilities were found to be related to a significantly lower probability of regulatory violation while operating capacity and waste storage did not show any statistically significant impact on compliance behavior. Finally, except for production intensity, we did not find any evidence that county-level community characteristics were associated with a facility's compliance behavior.