1. Introduction

Securing regulatory compliance poses significant challenges across the globe. Among the nine planetary boundaries widely posited as critical to stable global ecological conditions, six high-risk thresholds have already been crossed, including nitrogen and phosphorus flows from anthropogenic processes, underscoring severe implementation gaps in international environmental governance [1,2]. A paradigm shift in artificial intelligence innovation has sparked widespread demands for regulation and widespread debate on the appropriate compliance burdens to impose on foundation model developers [3,4]. In the United States, the Internal Revenue Service battles a half-trillion dollar annual gap between taxes owed and collected [5]. Hazardous work conditions cause more than 100,000 deaths in the U.S. each year, revealing the limits of current enforcement of occupational health and safety regulations [6]. And 60–75% of major facilities permitted under the 1972 Clean Water Act (CWA) self-report being in noncompliance annually [7], with at least one in four regulated facilities in significant noncompliance [8]. In many areas, enforcement resources have dwindled. The U.S. Environmental Protection Agency’s (EPA) budget has decreased by 41% since 1980, and 40 states have cut staff since 2010 [9]. The challenges posed by these shrinking resources are amplified by regulatory design choices that, as former EPA Assistant Administrator Cynthia Giles argues, fail to actively promote compliance, relying instead on mistaken assumptions of widespread compliance regardless of costs, incentive structures, or enforcement level [8]. Any hope to improve this situation must start from understanding what has been tried, and how it has fared.

Theoretical frameworks for understanding the dynamics of regulatory compliance and evasion date to the “rational choice” model introduced by Becker [10], which posits that regulated facilities will compare the costs associated with compliance (e.g., maintaining or upgrading water treatment systems) with the expected costs of evasion, weighing the size of sanctions for noncompliance by the (perceived) probability that their violations are detected and subject to sanctions. For regulators, this model suggests three policy levers by which they can influence the behavior of regulated entities: (1) increase the probability that violations are sanctioned, either by investing more resources in inspections and enforcement or through policy changes that ensure detected violations are consistently penalized; (2) increase the size of sanctions themselves; or (3) reduce the cost of compliance by, for instance, providing monetary or technical assistance. To the extent that regulated facilities are indeed rational actors, an equivalent change to any term of this calculation (e.g., reducing compliance costs by $1, increasing sanctions by $1, or increasing the probability of a $10 sanction by 10%) should have an equivalent impact on facility behavior, leaving regulators free to identify the most cost-effective allocation of their resources across the three levers. However, empirical literature has critiqued this equivalence implied by classical theory, pointing to considerable evidence of differential responsiveness to otherwise calibrated levers in criminal deterrence [11,12]. Several important questions remain inconclusively addressed, particularly in the domain of environmental regulation: To what extent are regulators in fact able to affect sanction probabilities, sanction sizes, and compliance costs? Is the behavior of regulated facilities more responsive to changes in some of these factors relative to others? That is, is the calculus as simple as weighing expected costs against each other, or are there other factors that might limit facilities’ ability to adjust to changes in this equation (e.g., political context, role of upstream actors, etc.)?

Here, we explore these core questions of regulatory compliance by examining a change in California’s enforcement of the CWA, specifically violations of regulated pollutant levels and reporting requirements in permitted wastewater facilities, wherein California introduced mandatory minimum penalties (MMPs) and compliance assistance for eligible jurisdictions through penalty conversions. This work makes three contributions: First, we present a mixed-methods analysis of the MMP program, exploring how heterogeneity in its effectiveness at addressing different types of violations – including reporting violations and effluent exceedances of pathogens, biochemical oxygen demand, solids, and ammonia – reflects frameworks that treat polluters as rational actors weighing the relative costs and benefits of compliance, and the underappreciated importance of understanding differential compliance costs by violation and non-monetary costs such as political pressure. Second, we explore the effectiveness of reducing compliance costs through monetary assistance, addressing a gap in environmental policy literature, namely the potential value of such assistive approaches within a rational choice model of compliance. And, third, we look beyond the aggregate impact of these approaches on water quality metrics to investigate the distributive impacts of eligibility design and implementation fidelity, defined as the alignment between intended and actual implementation, for penalty conversions.

Our paper proceeds as follows. In Section 2, we review prior theory, as well as related empirical work examining environmental compliance programs. From our literature review, we establish three key hypotheses corresponding to the three main policy levers of the Beckerian model. In Section 3, we detail the policy and program context for our specific case study, including the origins of California’s MMP program and penalty conversions, as well as the 2008 enforcement initiative which serves as a natural experiment. In Section 4, we present our research design, including explanation of data sources and methods. In Section 5, we organize our findings into three subsections: (A) quantitative evidence, including an event study and two synthetic control analyses; (B) qualitative evidence, structured around five within-case comparisons; and (C) a distributive equity analysis. In Section 6, we discuss the implications of our findings. Section 7 concludes.

2. Theory and related work

In the 1960s, Gary Becker, building on the seminal theories of Beccaria [13] and Bentham [14] from the 18th century, published “Crime and Punishment: An Economic Approach” [10], the first modern formalization of deterrence that has shaped generations of subsequent research in economics, criminology, law, and sociology [11,12,15–47], as well as generations of actual enforcement and sanction regimes, for better or worse. As aforementioned, its core contribution was to formulate a rational choice model of criminal behavior in which individuals weigh, in particular, the certainty and severity of punishment against the benefits of some unauthorized course of action. Critiques of Becker’s economic formalization, whether seeking to extend its reach through increasingly intricate modeling of expected utility, or questioning its fundamental ability to capture the inherent irrationality of certain worldviews and beliefs, all ultimately converge on the importance of evaluating theory against real-world implementation [16–18]. In fact, when Becker first introduced his theory, he himself explicitly acknowledged that empirical evidence of deterrence effects was severely limited in his time, yet essential for understanding discrepancies he agreed would be inevitable between optimal and real policies.

In the decades since, a large body of research has investigated the conditions necessary for effectively promoting compliance in practice, such as the proportionality of sanctions to behavior [20–22], or the swiftness and certainty of enforcement [15,23–25]. In a comprehensive literature review, Chalfin and McCrary [12] point to an emerging consensus from both theoretical extensions of Becker and empirical studies that offenders are likely more responsive to changes in the probability versus the severity of sanctions, due to nuances in risk aversion insufficiently captured by the classical model [11,19]. Although much of the foundational work in rational choice theory focuses, as Becker did, on crimes by persons against other persons and property, both this conceptual framework as well as the core question of the equivalency of policy levers affecting these Beckerian factors are relevant to other policy settings, particularly those involving regulatory compliance of institutional actors like industrial facilities.

In the environmental context, reviews of facility-level deterrence have likewise suggested responsiveness to increases in the probability of sanctions, with both detection-focused instruments (e.g., self-monitoring reports, facility visits) and enforcement-focused actions (e.g., telephone warnings, administrative orders) seen to yield reductions in violations as well as pollution [26,27]. However, when examining the role of sanction type and size in particular, just as in the criminal justice literature, results are more mixed. Studies have found that the existence and magnitude of fines reduced pollution at pulp and paper mills [28,29], chemical manufacturing plants [30], and sewage treatment plants [31,32]. Meanwhile, other studies have found no significant relationship between fines and environmental compliance among polluting facilities [33,34]. The isolated examination of diverse violation conditions and interventions across these works underscores the value of further empirical research that can study interconnected systems of deterrence components.

Recent works also increasingly focus on a third lever missing from the original Beckerian equation: increasing the opportunity cost of a future criminal act, i.e., increasing the incentive or reducing the disincentive associated with alternative, socially desirable activities. Unlike the traditional levers which focus on varying the probability or severity of sanctions, assistive approaches, such as pretrial diversion into treatment programs [35,36], fundamentally expand the policy toolkit with opportunities to address the root causes of criminal behavior and to break the cycle of apprehension and punishment. However, assistive interventions, like deterrence approaches, are diverse and have been found to produce varied outcomes, including in the environmental setting. A number of studies have found that compliance with environmental regulations may be improved with relatively low-cost interventions such as reporting deadline reminders [37], establishment of personal contacts [38], promotion of pollution prevention practices [39], and adoption of management standards [40,41]. Studies in other contexts, however, have had less promising findings, such as work by Evans et al. [42] that concludes voluntary self-audits had no long-run impact on environmental compliance, and other works that suggest compliance assistance may only be effective at small facilities [43] or alongside threatened sanctions [44]. Giles [8] argues that assistive approaches only work for facilities that participate, in contrast to the general deterrence produced by sanctions. Monetary compliance assistance is particularly understudied, even more so in the environmental context. Though monetary incentives have been found to increase compliance with school nutrition standards [45] and radiation safety [46], neither of these contexts included monetary sanctions as an alternative or directed incentives towards continued compliance. To our knowledge, only Anica and Elbakidze [47] have explored monetary assistance in the context of environmental regulation, finding that federal grants marginally improve CWA compliance among wastewater facilities.

Rarely in empirical literature, particularly within the environmental compliance scholarship, do studies have the opportunity to examine the interaction between multiple levers in one setting, making it difficult for regulators and researchers to understand how different interventions stack up against a fixed regulatory backdrop. In the present case study, we leverage a uniquely rich opportunity to study the interplay between all three key levers – deterrence-based changes to probability and severity of sanctions, as well as assistance-based approaches – of the Beckerian rational choice equation for polluters. In California’s MMP program, although penalties are in principle consistently applied based on self-reports, actual enforcement levels vary dramatically over time. An explicit step-up in enforcement in 2008 offers a rare look at how the behavior of regulated entities responds to changes in the likelihood of being sanctioned – the lever considered more effective in literature – and allows for the application of more robust identification strategies. Moreover, while individual sanctions are fixed in size, MMPs can still accumulate towards tipping points at which overall sanction liability influences facility behavior. In addition to observing the impacts of these variations in deterrence, we can also observe the impacts of a penalty conversion option which essentially reduces the cost of compliance for eligible facilities. Finally, our work focuses on publicly owned treatment works (POTWs; that is, facilities that treat municipal wastewater), allowing us to explore how such actors navigate this compliance calculus amidst the unique dynamics of the public sector, where utility rates may be subject to public votes, and budgets and operational parameters may be set by political agents. Although POTWs are the primary focus of CWA regulations, most research on regulatory compliance in this space has exclusively considered private sector actors; only Earnhart’s [31,32] studies of deterrence-based approaches and Anica and Elbakidze’s [47] research on the impact of federal grants explore means of promoting compliance within this distinctive population.

Few have studied the impact of California’s MMP program on the behavior of regulated facilities. A report from a public interest advocacy group [48] employed a pre-post design to measure effluent violation reductions in the two years following introduction of California’s MMP program as well as a similar program in New Jersey. In addition, a doctoral dissertation [49] found negative associations between region-level MMP enforcement and effluent violations, as well as between facility-level MMP enforcement and pollution, but the analysis was limited to regression models lacking an untreated comparison group. Kang and Silveira [50] used data from the California MMP program to develop a theoretical model of regulatory discretion, finding that discretion leads to a better social outcome than a one-size-fits-all enforcement policy (so long as regulations reflect local preferences). While these prior works provide some suggestive evidence for the ability of mandatory penalties to change facility behavior, our work provides a more extensive analysis of this program by combining a robust inference strategy with evidence from within-case analyses of POTWs and interviews of POTW operators. Using these mixed methods, we test three hypotheses: whether changes to probability of sanctions, severity of sanctions, and cost of compliance, respectively, lead to improvements in water quality.

3. Case study description

A. California’s MMP program

Under the National Pollutant Discharge Elimination System (NPDES), established by the CWA, facilities must obtain a permit to discharge pollutants into U.S. waters. NPDES permits specify limits on the quality and volume of the discharge (“effluent”), requiring permittees to perform regular testing and report results via Discharge Monitoring Reports (DMRs) submitted to both state and federal authorities. In most cases, CWA enforcement is delegated by EPA to state authorities, which have the discretion to set more stringent environmental protection standards than the federal baseline. Using this authority, California enacted a MMP program in 1999, requiring penalties of at least $3,000 per violation of certain NPDES permit conditions, regardless of the pollutant, to be paid to the State’s Cleanup and Abatement Account [51]. Violations are defined based on pollutant-specific effluent exceedance thresholds. All 266 POTWs in California are regulated by this program, which is one of only two in the U.S. imposing mandatory penalties. The California Water Boards, which include the State Water Resources Control Board and nine Regional Water Quality Control Boards, began implementing the MMP program when the statute came into effect in 2000. One of us (Mustain) is directly involved in MMP program enforcement through her role in the State Board. While effluent violations (i.e., exceeding pollutant discharge limits) were eligible from 2000, reporting violations (i.e., submitting a DMR late) were not subject to MMPs until 2004 (Cal. Wat. Code § 13385.1); see Fig 1. The Boards have other compliance tools alongside MMP enforcement; for example, for facilities under extended compliance schedules, MMP exemptions may be granted provided discharges remain in compliance with the relevant orders, even if they violate permit conditions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Timeline of key events in the implementation of California’s MMP program.

https://doi.org/10.1371/journal.pwat.0000326.g001

C. The 2008 enforcement initiative

Prior to 2000, California’s enforcement of NPDES violations was minimal [60]. Initial MMP implementation from 2000-2007 was hindered by resource constraints, as well as data quality and completeness issues [61,62]. As a result, with no time limit for the Regional Boards to assess MMPs, by 2007 there were more than 12,000 backlogged MMP violations from more than 450 facilities facing no enforcement action [62]. The backlog extent varied by region, ranging from 18% to over 90% of all MMP violations. While an overall 25% MMP enforcement rate in the early 2000s (see Fig 2) represented an increase over pre-2000 levels, MMP enforcement was inconsistent. Sanctions were thus hardly mandatory, and instead reflected a built-in Beckerian uncertainty of ever coming into fruition.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. 2008 MMP enforcement initiative.

Y-axis is a metric designed to represent the MMP assessment rate, calculated as the annual number of MMPs assessed divided by the one-year lagged count of MMP-eligible violations, presented as a three-year moving average. Pink line denotes POTWs; brown line denotes non-POTWs. The rate can exceed 1.0 when MMPs are assessed for backlogged violations.

https://doi.org/10.1371/journal.pwat.0000326.g002

In 2008, the State Board undertook an initiative to reduce the MMP enforcement backlog, thereby increasing the probability of sanctions. To streamline resolution of uncontested violations, an expedited settlement program was introduced, which allowed dischargers to simply acknowledge and pay their accrued penalties without undergoing a more resource-intensive, formal enforcement process. Fig 2 illustrates the sharp increase in MMP enforcement in 2008. By March 2009, settlement offers had been made to 455 violating facilities (including POTWs and non-POTWs), and 99% of all backlogged violations had been addressed [62]. Our analysis leverages the natural experiment provided by the 2008 MMP enforcement initiative’s sharp increase across the state in enforcement, and many facilities being functionally unaware of the program prior to 2008, i.e., they had not been assessed penalties prior to 2008 due to enforcement backlogs.

Since 2010, the State Board has required that MMP violations be enforced within 18 months of occurrence [55]. As of May 2011, the Boards claimed to have initially addressed all known MMP-eligible violations, though enforcement proceedings were ongoing in many cases [63]. By the mid-2010s, ongoing MMP enforcement was more swift and consistent: 80% of violations subject to MMPs in 2016 statewide were enforced within two years [64].

4. Research design

Though the MMP program applies to all NPDES permittees in California, including “major permittees” defined as POTWs with design flows of greater than one million gallons per day and/or approved industrial pretreatment programs, and certain private facilities based on EPA ratings criteria, we primarily focus our analysis on POTWs given they make up about 78% of major permittees. This is to ensure comparability between facilities in California and other states, and for consistency with analysis of penalty conversions, available only to POTWs.

A. Data sources

We draw data from two main sources: the California Integrated Water Quality System (CIWQS), California’s database for wastewater-related programs, and EPA’s Integrated Compliance Information System (ICIS), which tracks permits, DMRs, compliance, and other data related to NPDES programs across the United States [65,66].

From CIWQS, we obtain data on MMP enforcement actions and penalty conversions. From ICIS, we obtain legal effluent limits and reported effluent values from DMRs, as well as other permit information. ICIS has numerous data issues, such as missingness and data entry errors [67]. As a result of missing historical data, we analyze effluent data post-2004. In a typical year, our sample includes over one hundred POTWs submitting over one thousand DMRs containing several hundred exceedances to ICIS (see Table 1). Even when we examine only POTWs, these DMRs are reported at multiple frequencies, and include test results for hundreds of pollutants aggregated in multiple ways over multiple time periods. To control for this variability, we focus our analysis on four commonly reported groups of pollutants: pathogens, including E. coli and other bacteria; biochemical oxygen demand (BOD); solids, both dissolved and suspended in effluent; and ammonia. To account for reporting frequency variation, we aggregate effluent data annually. For more detail on the significant efforts to clean and process ICIS data, see Appendix A in S1 Text.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. POTW sample summary.

https://doi.org/10.1371/journal.pwat.0000326.t001

B. Methods

i. Synthetic control analyses.

Synthetic control methods estimate the causal effect of an intervention by comparing the treated unit to a weighted combination of untreated units that best replicate its pre-intervention characteristics, and are commonly used in policy evaluations involving a small number of aggregate units, such as countries and states [68–71]. We use the synthetic control method [68] to estimate the effect of MMP enforcement on effluent limit exceedances among POTWs (see Appendix B.1 in S1 Text for details). The treated unit is California, and the donor pool for the control group contains 33 other states (see Table 1). For effluent limit exceedances, we calculate “exceedance percentage” for each reported effluent value as the percentage difference between the reported and limit value, with negative and positive values reflecting compliance and noncompliance with permit limits, respectively. Exceedance percentages are averaged by facility-year, then by state-year, to create the outcome variable, mean exceedance percentage. We use a posttreatment analysis period of 8 years to account for the extended timelines of many POTW projects [47].

To estimate the impact of penalty conversions, we employ a facility-level generalized synthetic control approach [72] (see Appendix B.2 in S1 Text for details). Treated units include facilities that completed a penalty conversion, and the donor pool of controls includes all facilities that only settled their MMPs through monetary payments. Because facilities may complete multiple penalty conversions over time, we compare the timing of only a treated facility’s first penalty conversion with the first instance of MMP enforcement for control donor pool facilities. The synthetic control pretreatment fit is assessed for the eight-quarter (two-year) window before facilities received relevant MMP enforcement. The posttreatment period is also eight quarters, starting from notification of MMP enforcement (with the opportunity for penalty conversion), as facilities likely begin responding upon notification, not settlement. Due to data missingness, we focus our evaluation of penalty conversions on the all-parameter mean exceedance percentage pooled over exceedances of the specific effluents (out of pathogens, BOD, solids, ammonia, and metals) cited in each facility’s MMP enforcement case, based on manual review of administrative documents. The resulting sample size is small, leaving only 7 treated facilities for this analysis (see Fig C in S1 Text).

ii. Qualitative analysis.

To provide a richer understanding of the mechanisms by which MMPs and penalty conversions may affect outcomes, we also evaluate the MMP program qualitatively. First, we conduct within-case comparisons of four POTWs (see Fig 6 and Table 2). We select illustrations using a “diverse case” method [73] in order to reflect variation along two key dimensions: (1) penalty conversion eligibility and (2) the extent to which compliance appeared to improve post-enforcement (see Appendix E in S1 Text for more details). We analyze within-case comparisons collectively in order to identify cross-cutting themes, which are first extracted through process tracing as coded patterns [74], refined and grouped into themes, and finally exemplified through curated excerpts. At this level of granularity, we are able to scrutinize and document the fidelity with which Regional Boards defined and approved a facility’s eligibility for penalty conversions, and the appropriateness of a facility’s proposed project. Second, we review local news reports about the impacts of the MMP program on facilities other than those chosen for within-case analysis (see Appendix F in S1 Text). Third, we supplement within-case comparisons with a small set of semi-structured interviews with California POTW operators with exposure to MMPs (see Appendix G in S1 Text).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Additional within-case comparison information.

https://doi.org/10.1371/journal.pwat.0000326.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Strong negative association between MMP enforcement and reporting violations among non-POTWs.

Time series of annual total number of reporting violations, subject to MMPs in 2004, in California. Pink line denotes POTWs, brown line denotes non-POTWs.

https://doi.org/10.1371/journal.pwat.0000326.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Limited evidence for an effect of MMP enforcement on mean effluent limit exceedances among POTWs.

Analysis using pollutant category-specific mean exceedances as covariates. Pink line represents gap between treatment group (California) outcome and synthetic control outcome. Gray lines display outcomes when the treatment group is permuted across the donor pool.

https://doi.org/10.1371/journal.pwat.0000326.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Penalty conversions yield large practical decreases, but robustness and generalizability of result are limited by data quality and sample selection.

(a) Average treatment effect on the treated (ATT). Effects are estimated using a facility-level synthetic control, and bootstrap 95% confidence intervals are plotted. (b) Average effluent exceedances on the treated (pink) and counterfactual (gray) units. (c) Observed mean exceedance percent event studies for three selected treated facilities. Y-axis shows average mean exceedance percent, and vertical lines indicate calendar quarter of MMP enforcement.

https://doi.org/10.1371/journal.pwat.0000326.g005

iii. Ethics statement.

The research was approved by Stanford University’s Institutional Review Board Panel on Non-Medical Human Subjects Research (Protocol ID IRB-69001). Consent documentation was emailed to participants prior to the interview and, if they did not see this, it was also read to them at the start of the interview. All interview participants gave verbal informed consent (a waiver of documentation of informed consent), witnessed prior to the interview, in order to participate in the study.

iv. Distributive impact analysis.

We also assess the distributive impacts of the eligibility criteria for penalty conversion among communities served by POTWs active in 2022 (see Appendix D in S1 Text for more details). The Boards do not track service areas and eligibility for all POTWs, so we first retrieve service area and population information through manual reviews of multiple data sources. Because our curated dataset better reflects the actual populations served by POTWs than heuristics employed by prior studies, we release the resultant dataset (http://arcg.is/0aT8TG) to support future research on wastewater infrastructure. Once service areas are identified, we use population size and median household income data from the American Community Survey for eligibility determinations under the most recent criteria. We also incorporate environmental justice data from CalEnviroScreen 4.0 (CES) [75], including the overall CES score, overall pollution burden, and water quality-related risks and burden, and apply a population-weighted procedure to aggregate tract-level CES data to our curated POTW service areas. We test for sociodemographic and environmental differences between eligible and ineligible communities.

5. Findings

In “Quantitative evidence on treatment effects”, we begin our quantitative findings with (i) an event study examining late reporting, reflecting a Beckerian intuition that the relatively lower cost of compliance associated with on-time reporting might render this outcome particularly responsive to the step-up in enforcement, for both POTWs and non-POTWs. We follow with our primary synthetic control analyses of (ii) MMP enforcement and (iii) penalty conversions for POTWs. In “Qualitative evidence on treatment heterogeneity”, we supplement our quantitative findings with within-case comparisons of individual POTWs. In “Distributive impacts of penalty conversions”, we evaluate the distributive equity of penalty conversions.

C. Distributive impacts of penalty conversions

Penalty conversions are, in principle, available only to eligible communities, as defined on the basis of population size and financial hardship. We find that these definitions are in tension with typical definitions of environmental justice [111,112]: as detailed in Table 3, eligible POTW-served communities have proportionally larger non-Hispanic White populations (p < 0.05), while ineligible communities have a higher proportion of Hispanic, Black, and Asian (p < 0.1) constituents. This disparity cannot be explained by a difference in the risk of water pollution: eligible and ineligible communities do not have statistically significant differences in risks of surface, groundwater, drinking water pollution, or overall pollution burden. Nor do they substantially differ by CalEnviroScreen score, the primary index used to guide public investments improving public health, quality of life, and economic opportunity in California’s “disadvantaged communities.” Although penalty conversion eligibility qualifications are race-blind, they yield racial disparities seemingly inconsistent with environmental justice principles.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Environmental justice and sociodemographic characteristics of communities by penalty conversion eligibility.

https://doi.org/10.1371/journal.pwat.0000326.t003

Fig 7 underscores the degree to which different race-blind allocation strategies can have dramatically different equity implications. The first row represents the eligibility criteria based on financial hardship as actually applied to penalty conversions, while the four alternative strategies below it target groundwater risk, drinking water risk, surface water risk, and CalEnviroScreen score, respectively, and per our simulations would allocate the same total scope of penalty conversion benefits to progressively more diverse communities across the state.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Within-case comparison information.

(a) Map of within-case comparison locations, indicating the approximate population served by each POTW. (b) Distribution of community-serving POTWs along two selection dimensions, with population density in gray. (c) Compliance trajectories for POTWs. Y-axis shows annual MMP effluent violations, and red dots indicate MMPs with size proportional to total penalty amount.

https://doi.org/10.1371/journal.pwat.0000326.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Racial/ethnic composition of eligible populations under alternative allocation strategies.

In each simulation, we use a different vulnerability measure to select the same number of people targeted by the current financial hardship criteria and calculate the racial composition of the selected population.

https://doi.org/10.1371/journal.pwat.0000326.g007

6. Discussion

Taken together, many of our findings are broadly consistent with general tenets of compliance theory and support our three key hypotheses related to three levers of the Beckerian equation, with several indications of the MMP program’s effect on compliance appearing to be proportionate to the certainty and scale of penalties, as well as suggestive evidence from our penalty conversion analysis that reducing the cost of compliance can also shape facility behavior. Yet our deeper look at this policy also surfaces limits to classical theory, including the complexity of facility decision-making situated in its social and political context, and the less-considered equity implications of enforcement decisions. Below, we review the key implications of our findings for regulators, starting with the three Beckerian levers and continuing with environmental justice considerations, the importance of data quality, the unintended consequences of policy changes, and the scale of MMPs. We finish the section with a discussion of study limitations and avenues for future work.

In our quantitative analysis, we find that a step-up in MMP enforcement (i.e., changing only the probability of sanction) was associated with a reduction in reporting violations for private permittees. These violations are presumably cheap to fix and have been shown to be responsive to lightweight compliance assistance interventions, such as deadline reminders [37]. POTWs faced less of a baseline problem with late reporting, likely because they are in the business of water treatment and more familiar with reporting protocol. However, diverging from the conclusions of prior works, for more substantive effluent violations, the effect of MMP enforcement is unclear, with no significant exceedance reduction for all but the most tractable (pathogens) of four pollutants considered. Our qualitative analysis supports this conclusion, suggesting the severity of sanctions governs in this case: individual MMPs are small relative to compliance cost and only effective once accumulated. This interaction between certainty and scale of sanctions is, of course, consistent with standard notions of deterrence, wherein a rational polluter will not comply if the expected value of penalties is less than the cost of compliance. However, where prior works largely fail to disaggregate analyses by pollutant type, our case study surfaces a diversity of enforcement outcomes across four key exceedance types and underscores that compliance cost variation – ranging from mere process quality improvements to large capital improvements implicating myriad stakeholders – materially shapes facility behavior, and should not be left abstracted away in empirical studies. Furthermore, traditional deterrence models have not accounted for other, non-monetary factors affecting POTWs’ strategies, such as the behavior of upstream actors as well as the political environment (with rate increases often subject to popular approval and budgets set independently by city governments). One of us (Mustain) recalls one occasion in which a POTW operator urged regulators to assess a penalty at their facility in order to nudge their city council to approve a budget for needed upgrades. While concerns around public perception and media coverage may be relevant for both private and public facilities, they are likely particularly acute for public operations such as municipal wastewater facilities. Because these facilities are a primary focus of the CWA, it is important for both theoretical and empirical works seeking to understand environmental compliance to account for the unique incentive structures and constraints they face.

Our results for penalty conversions are similarly nuanced, with suggestive impacts considerably limited by the small number of facilities with high-quality data. Still, an improvement in compliance would be consistent with previous work that suggests compliance assistance paired with the threat of sanctions can be an effective combination [44], including in the context of financial assistance on wastewater compliance [47]. However, because the size of penalty conversion assistance reflects MMP totals, again the small scale of penalties relative to the cost of compliance may limit the impact of penalty conversions, even as they accumulate across multiple violations. The average grant studied by Anica and Elbakidze [47] was $7.6 million, while the median penalty conversion in this case study was roughly $10,000. For cash-strapped communities like Colfax, multiple penalty conversions resolving large accumulated penalties appear to have made a significant difference. That being said, the Regional Board’s approval of retroactive penalty conversions in Colfax’s case, while technically acceptable under the State’s enforcement policy, is normatively debatable. On one hand, retroactive approvals allow Regional Boards, vested the discretionary powers of street-level bureaucrats [113], to flexibly reward good-faith, proactive compliance efforts and not force municipalities to wait for an administrative order before taking action. On the other hand, they complicate efforts to measure impact, and, even with state guidelines, create an opening for inconsistent implementation between regions and facilities.

Penalty conversions also raise important tensions with environmental justice, codified in California as “the fair treatment and meaningful involvement of people of all races, cultures, incomes, and national origins, with respect to the development, adoption, implementation, and enforcement of environmental laws, regulations, and policies” [114]. Despite the Boards’ commitment to racial equity [115], their eligibility criteria are still exclusionary for larger, more urban communities of color that have been historically underserved. Such criteria do not necessarily affect the efficacy of penalty conversions, but should benefits exist, they dictate whether those benefits are equitably distributed across communities. The California legislature can make this compliance assistance more consistent with its own priorities by reforming eligibility criteria. A more inclusive and environmentally conscious approach could be to incorporate water-related data from CalEnviroScreen to better direct program benefits to historically underserved communities (see Fig 7) [116]. That being said, it is worth keeping in mind that penalty conversions are fundamentally conversions of certain municipal budget line items, which would have gone towards paying penalties, into budget line items that can now go towards paying for infrastructure improvements. Regulators should not lose sight of the fact that an environmental externality – the pollution from the original violations – is inextricably baked into this conversion. Lagged, retroactive, and discretionary implementation could effectively render penalty conversions a sanctioned form of infrastructure “fundraising” where – perversely – it pays to pollute.

Our findings underscore the importance of clear enforcement guidelines, which support program integrity and consistency across disparate geographies and regulated entities, as well as render policies like the MMP program easier to evaluate. We also stress that the lack of reliable, granular data on service-area populations hinders the Boards’ ability to consistently implement and evaluate penalty conversions, necessitating detailed research to determine a facility’s eligibility for each proposed project. High-quality data collection and availability at the state and federal levels are invaluable for both enforcement and program evaluation. ICIS data incompleteness has broader ramifications for design, implementation, and evaluation of federal environmental policy: for example, EPA’s EJScreen uses ICIS data.

Although our findings are broadly consistent with rational polluter models that suggest either increases to the scale of penalties or reductions to the cost of compliance are likely to be met with proportionate response in levels of compliance, our case study of the California MMP program also highlights an important gap between theory and practice: these models provide little to no guidance on the systems-level implications of the various tweaks regulators can make to the different sides of this Beckerian equation. For instance, expanding penalty conversion eligibility criteria might help a wider set of facilities come into compliance more quickly, thereby reducing the burdens on surrounding communities. However, such a policy could also be self-defeating to the degree that it enables more “free riding” by facilities that can actually afford to make upgrades without the incentives of compliance assistance (recall cases of individual jurisdictions even seeking to directly manipulate eligibility for themselves through legislative strategies). Perhaps such facilities should remain solely subject to the threat of monetary sanctions – which, in turn, fund the mitigation of other environmental harms across the state.

Indeed, the mixed evidence for MMP impacts on effluent violations also suggests that California policymakers may want to reconsider the scale of the base penalties. Not only is $3,000 very small relative to typical compliance costs, but the amount has been unchanged despite 81% cumulative inflation since 2000 [117]. Furthermore, the minimum penalty amount could be increased for larger single violations or more toxic pollutants in order to better reflect the true social cost of the violation. While accurately measuring such costs is difficult, it seems implausible that they have remained both uniform and unchanged since 2000. To be fair, achieving complete compliance is both impractical and prohibitively expensive for POTWs: systems fail; trying to fix a small issue can cause a larger one; and upstream actors make bad decisions. Nevertheless, to the extent regulators, policymakers, and the public believe that environmental compliance is too low, and that those affected are insufficiently compensated via penalties, it would seem that the cost of a violation may be too low.

Finally, while we believe our mixed-methods approach provides several insights for both researchers and regulators about the relative roles of assistance and penalties in promoting environmental compliance, a number of aspects of data quality and availability pose limitations that may suggest promising avenues for future work. In our analysis of effluent exceedances, we see a range of effect sizes that seem broadly consistent with deterrence theory, but those are the combined effects of the deterrence and compliance assistance components of the program. While we gain valuable insights about the interplay of components from analyzing this combined setting, it would, for instance, be further illuminating if we could isolate the deterrence impacts of penalties on facilities not eligible for penalty conversions by comparing similar facilities in donor-pool states, but the lack of service-area data for POTWs creates a significant barrier to such a study (one for which we hope our curated service-area dataset for CA facilities will serve as a model toward addressing). Future work can also more closely examine extended compliance schedules; while only 1.6% of MMP violations are associated with extended compliance schedules, and these violations are omitted in Figs 2 and 3, we cannot disentangle their influence in our synthetic control analyses, and our qualitative analyses suggest they may play an outsized role in shaping compliance behavior.

While the within-case comparisons and interviews added valuable context and depth to our understanding of the effects and limitations of the MMP program, the illustrations are not representative of the full population of POTWs. Illustrations with more news coverage and public records were more likely to be included. Of course, one key takeaway from this case study is the considerable heterogeneity in the impact of the MMP program, across types of violations as well as individual facilities and the various financial, political, and structural incentives they face. A more comprehensive qualitative study here could be worthwhile, as could further investigation of specific behaviors, motivations, and conditions conducive to compliance for different categories of facilities. In particular, our analysis focused on POTWs given they are the primary focus of NPDES-related enforcement, to strengthen the causal assumptions employed in our synthetic controls, and to focus on a relatively uniform population across our study of MMPs and penalty conversions. Future work that explores the responsiveness of non-POTW facilities (including manufacturers, commercial establishments, and other private facilities) to MMP enforcement would be of particular interest, especially given the strong reduction in reporting violations observed in this population after 2008.

Our ability to develop a robust inference strategy to measure the quantitative impact of the MMP program was limited also by data quality, which has been a major challenge under the CWA more generally. For example, while the introduction of the MMP program in 2000 may itself have changed compliance behavior despite minimal enforcement prior to 2008, historical missingness in both state and federal data preclude an analysis of the initial effect. In addition, we assume that self-reported DMR data are accurate, in spite of known data integrity issues. Our penalty conversion synthetic control analysis was also limited by historical ICIS data incompleteness and self-selection into electronic DMR reporting, which produced a small sample size and short pre and posttreatment periods. Given the continuous effort by the Boards to improve data collection and quality control practices [118], we encourage researchers to revisit this policy in the future to examine its long-term impacts on compliance and environmental justice.

7. Conclusion

Understanding the efficacy of different enforcement approaches is critical to achieving regulatory goals across diverse contexts, though modern deterrence theory offers little guidance on which elements of a rational polluter’s calculus may be best to focus on in practice. Here, we explored the role of both deterrence and compliance assistance tools through the lens of California’s MMP program. At a high level, our findings support rational polluter models of deterrence theory: the decline in violations of effluent reporting requirements, coupled with no clear effects on effluent water quality, highlights the need for costs of violations to proportionately align with compliance costs. This is further evidenced by some operators’ indifference to the minor penalties, often neither swift nor certain, in our qualitative analyses. At the same time, within-case comparisons did confirm that the MMP program, by increasing the perceived probability of sanctions among some facility operators and the political cost of noncompliance for some local governments (at least, after the initial enforcement delays that hampered the program’s early implementation), did influence compliance behavior. Similarly, although limited by data incompleteness, both our quantitative and qualitative analyses of penalty conversions suggest this assistance could nudge some facilities towards improved compliance, at least when sized appropriately relative to the cost of needed upgrades. Nonetheless, even where accumulated MMPs or a penalty conversion might be large enough to incentivize facility behavior, our qualitative work revealed additional important factors, such as the behavior of upstream actors, political support from the city council or public, and constraints on population growth.

This case study has implications for policymakers and regulators, not just those continuing to improve California’s MMP program but those in other states and countries considering similar approaches, who should carefully consider how the scale of penalties relates to both the social cost of the harms of violations and the cost of compliance for facilities. Likewise, challenges in implementing and evaluating California’s MMP program underline familiar lessons for regulatory program design: clear enforcement guidelines, sufficient enforcement resources, and high-quality data are critical for program efficacy and evaluation. Finally, our observation that the distribution of penalty conversion eligibility deviates from commonly understood environmental justice principles underscores how such normative considerations may be missing when rational choice theory is indiscriminately applied to policy design, and reflects a need for thoughtfulness and intentionality in how our social values and goals are encoded in the design and implementation of regulatory structures.

Supporting information

Acknowledgments

We thank the California wastewater facility operators we interviewed for their valuable insights; Nicole Lin, Cornelia Ilin, Nicolas Rothbacher, Dawson Verley, and Victor Wu for research and data assistance; and Arun Frey, Phil Bobel, and staff at the Palo Alto Regional Water Quality Control Plant for helpful comments and conversations.