2.1. The setup

Consider two regions, labelled by $i{\in }\left\{1,2\right\} $, under a central government. In region $i$, a representative firm $i$ emits pollution when producing, where $i=1,2$. The actual level of firm $i$’s emissions is denoted by $e_{i}$. Following Sandmo (Reference Sandmo2002) and Macho-Stadler and Perez-Castrillo (Reference Macho-Stadler and Perez-Castrillo2006), we assume that firm $i$’s production function takes the form $\mu _{i}e_{i}^{1/2}$, where the coefficient $\mu _{i}$ denotes firm $i$’s marginal benefit of polluting. This function measures the direct revenue from production where the output is associated with the actual level of emissions $e_{i}$, or what firms can gain by avoiding the cost of reducing emissions.

To control emissions, firms are subject to environmental regulation and are required to self-report their levels of emissions.Footnote ³ Let $z_{i}$ denote firm $i$’s self-reported level of emissions, which may or may not coincide with $e_{i}$. As there is no reward for over-reporting emissions, the reported level should never exceed the actual level (i.e., $z_{i}\leq e_{i})$. Thus, $e_{i}-z_{i}$ denotes firm $i$’s level of environmental violation. The full set of environmental regulations consists of a predetermined environmental policy and an enforcement strategy decided by the corresponding authority. The policy levies a tax rate $t$ associated with the level of reported emissions and a penalty rate $r$ for the under-reported level of emissions $e_{i}-z_{i}$. The policy is enforced in order to identify firms’ actual levels of emissions and applies to firms with an inspection intensity or probability $\alpha {\in }[0,1]$. Upon being audited, a firm’s actual level of emissions may not be perfectly detected. Let $\theta $ denote the probability that an enforcement agency accurately identifies firms’ actual emissions and environmental deviations. These imperfections in detection may be due to pollution’s natural volatility, restrictions on inspection technology, or information manipulation or corruption. Correspondingly, the magnitude of imperfection in detection may vary across regions and agencies. Let $\theta _{i}^{C}$ and $\theta _{i}^{D}$ denote the inspection effectiveness of central agencies and regional agencies in region $i$, respectively. Differences in inspection effectiveness across levels of government may arise from institutional and informational advantages. For example, central environmental agencies may possess superior monitoring technologies, broader investigative authority or stronger insulation from local political influence, leading to higher detection effectiveness. In contrast, regional authorities may possess better local information regarding firms’ production processes and pollution patterns, which can improve the accuracy of inspections in specific jurisdictions. Such differences are consistent with the institutional settings discussed in the Introduction, including the EPA’s delegation arrangements across U.S. states and the coexistence of central supervision and local implementation in China’s environmental enforcement system.

Pollution is a public bad, resulting in environmental damages that can extend far beyond a given region. Assume that the pollutant is a global one (e.g., carbon dioxide), and that its damages to each region depend on the combined level of emissions $e_{1}+e_{2}$. The damage function associated with emissions is given by $(e_{1}+e_{2})^{2}$. The convex functional form indicates increasing marginal disutility of emissions, which is a common assumption in environmental economics (Hritonenko and Yatsenko, Reference Hritonenko and Yatsenko2013). In addition to environmental concerns, regulations also bring up concerns about business friendliness. Authorities thereby face a tradeoff between the environment and the economy. Let $\lambda$ denote the weight an authority assigns to the environment relative to the economy, which depends on political agendas, population size and other factors (Woods, Reference Woods2022). This weight may vary across different authorities; $\lambda ^{C}$ and $\lambda _{i}$ denote the central and local authorities’ environmental concerns, respectively. The central or regional authority chooses the inspection strategy to maximize social welfare, which in this model consists of firm profits and environmental damages (e.g., Glazer, Reference Glazer1999; Kunce and Shogren, Reference Kunce and Shogren2002; D’Amato and Valentini, Reference D’Amato and Valentini2011). If the inspection strategy is determined by the central authority, denoted by $\alpha^{C}$, it indiscriminately applies to both regions.Footnote ⁴ If the local authority of region $i$ determines its own inspection strategy, it chooses $\alpha _{i}^{D}$ to maximize social welfare of the region. In what follows, we use the generic notations $\alpha$ and $\theta$ when referring to the inspection intensity and effectiveness of any decision structure. Table 1 provides definitions for all parameters and decision variables.

Table 1.

Parameters and decision variables

Table 1 Long description

To facilitate readability, Table 1 presents summary information for all variables, encompassing actual and reported pollution, marginal benefit, tax and penalty rates, inspection intensity, detection probability and environmental weight.

We propose a two-stage game, and the timing of the game is as follows. In stage $1$, given the predetermined environmental policy $(t,r)$, either the central or regional authority decides on the inspection strategy. If the strategy is determined at the regional level, the authorities of both regions determine inspection intensities simultaneously. The inspection strategies are publicly observable. In stage $2$, after observing the environmental policy and the inspection strategy, firms choose their production plan $e_{i}$ and the level of emissions $z_{i}$ to report. We solve the game by backward induction.

2.2. The firm’s decision

In stage $2$, after observing the environmental policy $(t,r)$ and the inspection strategy $\alpha ^{C}$ or $\alpha _{i}^{D}$, firm $i$ chooses $e_{i}$ and $z_{i}$ to maximize its profit. For each chosen pair $(e_{i},z_{i})$, firm $i$ pays an environmental tax $tz_{i}$. With probability $\alpha $, firm $i$ will be inspected by the corresponding authority, and its actual level of emissions will be detected with probability $\theta$. In the event of being caught underreporting, firm $i$ faces the consequences of paying back the amount of its tax evasion $t(e_{i}-z_{i})$ plus a fine $r(e_{i}-z_{i})$ for noncompliance. Assume that the following condition holds throughout the paper:

Assumption 1 means that the monitoring is never perfect. It follows that even if the firm is inspected with certainty – i.e., $\alpha=1$, it is still not optimal for the firm to report honestly. Firm $i$’s profit maximization problem reads:

\begin{equation*} \begin{aligned} & \underset{e_i, z_i}{\text{max}} & & \mu_ie_i^{1/2}-tz_i-\alpha\theta_i(t+r)(e_i-z_i)\\ & \text{subject to} & & z_i{\;\geq\;}0,\\ & & & z_i{\;\leq\;}e_i.\\ \end{aligned} \end{equation*}

The following lemma characterizes firm $i$’s decisions in stage $2$.Footnote ⁵

Lemma 1. Given policy parameters $(t,r)$, the inspection intensity and effectiveness $\alpha $ and $\theta $, firm $i$’s optimal strategy is defined as:

\begin{equation*} e_{i}^{\ast }=\left( \frac{\mu _{i}}{2\alpha \theta (t+r)}\right) ^{2}\ \ \ \text{and}\ \ \ z_{i}^{\ast }=0, \end{equation*}

resulting in firm profit:

\begin{equation*} \pi _{i}^{\ast }=\frac{\mu _{i}^{2}}{4\alpha \theta (t+r)}. \end{equation*}

Lemma 1 shows that firm $i$’s actual level of emissions $e_{i}$ decreases in the inspection intensity $\alpha $. That is, the stricter the enforcement, the more effectively it deters pollution. The same effect occurs with the environmental policy $(t,r)$ and the inspection effectiveness $\theta $, which decreases the equilibrium emission level and the firm’s profit. A higher $\mu _{i},$ suggesting higher benefits from pollution or a higher abatement cost, encourages the firm to pollute more. Regarding the optimal self-reported emissions level, the firm claims it is clean. Because the violation detection $\theta$ is not always efficient, the firm chooses to report the minimal possible level, which is $z=0$ in this model and could be positive in reality. According to the Toxic Release Inventory, though U.S. regulators added a large number of plants to their reporting programme, these facilities generally reported low pollution levels that accounted for only a small percentage (6 per cent) of the total releases (De Marchi and Hamilton, Reference De Marchi and Hamilton2006). Moreover, a self-reported level of zero could also be due to failure to file a report, which is prevalent in environmental compliance (Brehm and Hamilton, Reference Brehm and Hamilton1996).

2.3. Centralized enforcement

Anticipating firms’ reactions to certain inspection strategies, responsible authorities determine the inspection intensity in stage $1$. We first discuss the scenario in which the central authority determines the inspection strategy. The central authority weighs firm profits and environmental damages of both regions, and its welfare-maximizing problem reads:

(1)\begin{equation} \begin{aligned} & \underset{\alpha^C}{\text{max}} & & \sum\limits_{i=1}^{2}\pi_i(\alpha^C)-\lambda^C(\sum\limits_{i=1}^{2}e_i(\alpha^C))^2.\\ \end{aligned} \end{equation}

Substituting $e_{i}^{\ast }=\left( \frac{\mu _{i}}{2\alpha \theta_{i}(t+r)}\right) ^{2}$, $z_{i}^{\ast }=0$ and $\pi _{i}^{\ast }=\frac{\mu _{i}^{2}}{ 4\alpha \theta_{i}(t+r)}$ obtained in Lemma 1 into maximization problem 1, the central government’s problem becomes:

\begin{equation*} \begin{aligned} & \underset{\alpha^C}{\text{max}} & & \sum\limits_{i=1}^{2}\frac{\mu_i^{2}}{4\alpha ^{C}\theta^C_i(t+r)}-\frac{\lambda ^{C}}{[2\alpha ^{C}(t+r)]^{4}}(\sum\limits_{i=1}^{2}\frac{\mu_i^{2}}{\theta_i^{C^2}})^2.\\ \end{aligned} \end{equation*}

Taking the first-order condition and solving for $\alpha ^{C}$, we obtain the following lemma that characterizes the optimal centralized inspection strategy and the corresponding total surplus.

Lemma 2. The optimal centralized inspection intensity is given by

\begin{equation*} \alpha^{C^\ast}=\frac{1}{t+r}\left( \frac{\lambda^{C}\left( \frac{\mu _{1}^{2}}{(\theta_1^C)^2}+\frac{\mu_2^2}{(\theta_2^C)^2}\right) ^{2}}{\frac{\mu _{1}^{2}}{\theta_1^C}+\frac{\mu_{2}^{2}}{\theta_2^C}}\right) ^{1/3}, \end{equation*}

and the total surplus of both regions under centralized enforcement, denoted by $W^{C}$, is given by:

\begin{equation*} W^{C^\ast}=\frac{3}{16}\left( \frac{\left( \frac{\mu _{1}^{2}}{\theta_1^C}+\frac{\mu_2^2}{\theta_2^C}\right) ^{4}}{\lambda ^{C}\left( \frac{\mu_1^2}{ (\theta_1^C)^2}+\frac{\mu_2^2}{(\theta_2^C)^{2}}\right) ^{2}}\right) ^{1/3}. \end{equation*}

The optimal centralized inspection intensity decreases in both $t$ ( $\frac{\partial \alpha ^{C^\ast}}{\partial t} \lt 0$) and $r$ ( $\frac{\partial \alpha ^{C^\ast}}{\partial r} \lt 0$), suggesting that the inspection intensity and stringency of environmental policies are substitutes.

2.4. Decentralized enforcement

In a decentralized framework, each region is responsible for its own enforcement strategy. In stage 1, the regional authority chooses its inspection intensity to maximize the regional surplus. Region $i$’s welfare-maximizing problem reads:

\begin{equation*} \begin{aligned} & \underset{\alpha_i^D}{\text{max}} & &\pi_i(\alpha_i^D)-\lambda_i^D(e_1(\alpha_1^D)+e_2(\alpha_2^D))^2.\\ \end{aligned} \end{equation*}

Substituting $e_{i}^{\ast }=\left( \frac{\mu _{i}}{2\alpha \theta_{i}(t+r)}\right) ^{2}$, $z_{i}^{\ast }=0$, and $\pi _{i}^{\ast }=\frac{\mu _{i}^{2}}{ 4\alpha \theta_{i}(t+r)}$ obtained in Lemma 1 into the welfare maximization problem of region $i$, the local government’s problem can be written as:

\begin{equation*} \begin{aligned} & \underset{\alpha_i^D}{\text{max}} & &\frac{\mu _{i}^{2}}{4\alpha _{i}^{D}\theta _{i}^{D}(t+r)}-\frac{\lambda _{i}}{[2(t+r)]^{4}}\left( \frac{\mu _{1}^{2}}{(\alpha_{1}^{D})^{2}(\theta _{1}^{D})^{2}}+\frac{\mu _{2}^{2}}{(\alpha_{2}^{D})^{2}(\theta _{2}^{D})^{2}}\right) ^{2}.\\ \end{aligned} \end{equation*}

Taking the first-order condition with respect to $\alpha_i^D$, we obtain the best response function $\alpha_i^D(\alpha_j^D)$, where $i,j=1,2$ and $i\neq j$. Solving the system of best response functions, the following lemma characterizes the optimal decentralized inspection strategies and the corresponding total surplus of each region.

Lemma 3. The optimal decentralized inspection intensity of region $i$ is given by:

\begin{equation*} \alpha _{i}^{D^{\ast }}=\frac{1}{\theta _{i}^{D}(t+r)}\left[ \lambda _{i}\left( \mu _{i}^{2}+\frac{\mu _{j}^{2}\lambda _{i}^{2}}{\lambda _{j}^{2}}\right) \right] ^{1/3}, \end{equation*}

and the total surplus of region $i$, denoted by $W_{i}^{D}$, is given by:

\begin{equation*} W_{i}^{D^{\ast }}=\frac{3\mu _{i}^{2}-\frac{\mu _{j}^{2}\lambda _{i}^{2}}{\lambda _{j}^{2}}}{16\left[ \lambda _{i}\left( \mu _{i}^{2}+\frac{\mu _{j}^{2}\lambda _{i}^{2}}{\lambda _{j}^{2}}\right) \right] ^{1/3}}, \end{equation*}

where $i,j=1,2$ and $i{\;\neq \;}j$.

The optimal decentralized inspection intensity of region $i$ decreases in both $t$ ( $\frac{\partial \alpha_i ^{D^\ast}}{\partial t} \lt 0$) and $r$ ( $\frac{\partial \alpha_i^{D^\ast}}{\partial r} \lt 0$). In the decentralized framework of enforcement, the inspection intensity and the stringency of environmental policies are substitutes as well.

If region $j$ cares more about the environment, indicating a larger $\lambda_j$, region $i$ lowers its inspection intensity ( $\frac{\partial \alpha_{i}^{D}}{\partial \lambda _{j}} \lt 0$). The inter-region externalities provide region $i$ with an incentive to free ride on region $j$’s environmental efforts. As a consequence, region $j$ suffers a welfare loss due to region $i$’s increased pollution. Moreover, a stringent enforcement strategy harms both the firm profits and business attractiveness of region $j$. Both effects drive region $j$ to implement a less stringent enforcement strategy.

By the same token, region $i$’s total surplus increases in region $j$’s level of environmental concern ( $\frac{\partial W_{i}^{D}}{\partial \lambda _{j}} \gt 0$). In the extreme case in which region $j$ cares exclusively about firm profits ( $\lambda _{j}\rightarrow 0$), it chooses a minimum level of inspection intensity ( $\alpha _{j}\rightarrow 0$). Consequently, firm $j$ chooses the maximum level of emissions, causing region $i$ a considerable welfare loss. Moreover, if firm $i$ has a larger private benefit from polluting, it harms the social welfare of the other region ( $\frac{\partial W_{j}^{D}}{\partial \mu _{i}} \lt 0$) but not itself ( $\frac{\partial W_{i}^{D}}{\partial \mu _{i}} \gt 0$).

3.1. $\lambda^{C}=\lambda_1+\lambda_2$

We proceed by comparing centralized and decentralized enforcement setups along three dimensions: optimal enforcement policy, the total surplus of the country and the total surplus of the region. As local enforcement of environmental legislation has long been associated with the fear of environmental free riding, we first answer the question of whether (and under what conditions) local agencies put more or less effort into environmental protection compared to the central agency. The following proposition compares the optimal inspection intensity of the centralized strategy with that of the decentralized one.

Proposition 1 shows that fully symmetric regions impose less stringent enforcement strategies than the centralized one, which consequently results in higher pollution emissions. This illustrates that the negative externalities of the global pollutant cannot be fully internalized within a decentralized framework. If regional environmental concerns are highly heterogeneous ( $\frac{\lambda _{i}}{\lambda _{j}} \gt 1.5$), region $i$, which values the environment more, imposes a more stringent enforcement strategy relative to the centralized one, whereas region $j$ is more lenient in enforcement relative to the centralized one. Similar conclusions can also be obtained if inspection effectiveness across regions is highly heterogeneous. The decentralized strategy can better account for regional heterogeneities. A region can compensate for its low inspection efficiency or volatile pollution rate with a higher level of inspection intensity. Other than policy-related parameters, the characteristic heterogeneity in the marginal benefit of polluting $\mu $ across firms does not affect the comparison, so the centralized enforcement is always more stringent. Therefore, non-cooperative decisions by individual regions that do not account for interregional externalities provide inadequate enforcement from a coordinated point of view. In contrast, nondiscriminatory enforcement imposed by the central authority can result in too much stringency in regions with high inspection effectiveness and too much leniency in regions with low inspection effectiveness.

If the central government is much more effective in detecting environmental violations than regional governments ( $\frac{\theta ^{C}}{\theta ^{D}} \gt \sqrt[3]{2}$), the agency that is less effective at detecting environmental violations imposes a more stringent enforcement strategy as compensation. Conversely, local governments may decrease their inspection frequency in the decentralized structure if they have more information on their own firms. In sum, only the heterogeneity between the two levels of government can possibly induce both regions to implement stricter enforcement than the centralized enforcement, while the heterogeneity across regions leads to one or both local agencies being more lenient in inspection.

Based on optimal enforcement strategies, we then evaluate centralized and decentralized enforcement from the welfare perspective.

Delegating enforcement powers to local authorities is superior from the welfare perspective when inspection effectiveness is highly heterogeneous across regions. The gains from addressing regional heterogeneity outweigh the welfare loss stemming from not fully internalizing the negative externalities of the pollutant. Comparing this to Proposition 1, though the heterogeneity in the level of environmental concern $\lambda _{i}$ makes each region adapt its enforcement policy accordingly, the decentralized structure does not yield higher welfare as long as $\lambda ^{C}=\lambda _{1}+\lambda _{2}$. In addition, centralized enforcement is always preferred, regardless of the heterogeneity across levels of government.

Given that decentralized enforcement can yield a higher total surplus under certain conditions, we ask whether both regions can benefit from the decentralized structure. Alternatively, we ask whether both regions have an incentive to apply for enforcement delegation rights.

Without compensation or subsidy across regions, shifting from centralized to decentralized enforcement makes one region better off and the other worse off, even if decentralization yields a higher total surplus. This cannot be a Pareto improvement. Therefore, even if decentralization is preferable from the perspective of aggregate social welfare, it may not be feasible to implement.

3.2. λ^C>λ₁+λ₂

In this section, we consider the scenario in which the central authority’s level of environmental concern exceeds that of the regional authorities combined – i.e., $\lambda^{C} \gt \lambda _{1}+\lambda _{2}$. This might be due to pressure from the international community or a ruling party’s political agenda. We focus on whether the divergence of concern between central and regional authorities affects previous conclusions.

We conduct the comparison in two ways. First, from the central authority’s perspective: $W^{C}$ versus $W^{D}(\lambda ^{C})$. The former denotes the total surplus under centralized enforcement defined in Section 2.3, and the latter denotes the total surplus under decentralized enforcement associated with the central authority’s environmental concerns $\lambda ^{C}$. Specifically, the latter is given by

(2)\begin{equation} W^{D}(\lambda ^{C})=\underset{i=1}{\overset{2}{\Sigma }}\pi _{i}(\alpha _{i}^{D})-\lambda ^{C}(e_{1}(\alpha _{1}^{D})+e_{2}(\alpha _{2}^{D}))^{2}. \end{equation}

This comparison corresponds to the central authority’s evaluation of both centralized and decentralized enforcement, with the same level of environmental concern $\lambda ^{C}$. Second, from local authorities’ perspective: $\underset{i=1}{\overset{2}{\Sigma }}W_{i}^{D}$ versus $\underset{i=1}{\overset{2}{\Sigma }}W_{i}^{C}(\lambda _{i})$. The former denotes the aggregate total surplus of both regions under decentralized enforcement defined in Section 2.4, and the latter denotes the aggregate total surplus of both regions under centralized enforcement associated with local authorities’ concerns $\lambda _{i}$ – that is,

(3)\begin{equation} \underset{i=1}{\overset{2}{\Sigma }}W_{i}^{C}(\lambda _{i})=\underset{i=1}{\overset{2}{\Sigma }}\pi _{i}(\alpha ^{C})-\underset{i=1}{\overset{2}{\Sigma }}\lambda _{i}(e_{1}(\alpha ^{C})+e_{2}(\alpha ^{C}))^{2}. \end{equation}

This comparison reflects regional governments’ preferences for centralized and decentralized enforcement according to their own levels of environmental concern.

We assume that the regions are symmetric in their environmental concerns – i.e., $\lambda _{1}=\lambda _{2}=\lambda _{0}$, and in pollution benefit – i.e., $\mu _{1}=\mu _{2}$,Footnote ⁶ but asymmetric in inspection effectiveness – i.e., $\theta _{1}\neq \theta _{2}$ and $\theta _{i}^{C}=\theta _{i}^{D}$. In this way, the analysis focuses on the effects of central–regional environmental concerns and heterogeneity in inspection effectiveness across regions. Note that when $\lambda ^{C} \gt \lambda _{1}+\lambda _{2}$, the comparison outcome of inspection intensity is similar to the case where $\lambda ^{C}=\lambda_{1}+\lambda _{2}$. The following proposition characterizes the welfare comparison from the perspectives of both the central and regional authorities.

This proposition shows that regional authorities also prefer decentralized enforcement when there is high heterogeneity in inspection effectiveness across regions. Together with the condition that the central authority’s level of environmental concern is similar to that of the regional authorities combined, the central and the regional authorities may agree on the enforcement structure. However, the preferences of central and local agencies do not always coincide. Regional authorities prefer self-enforcement when their level of environmental concern is low, whereas the central authority desires centralization. By relaxing enforcement intensity, regional governments can partially ignore negative externalities while enjoying the economic benefits firms bring, an outcome that deviates from more sustainable development guidelines set by the central government.

Figure 1 depicts the central and regional authorities’ views on both enforcement structures. The dashed curve in Figure 1 corresponds to the central authority’s preference for enforcement structures, whereas the solid curve represents the regional authority’s. The region labelled as $C$ ( $D$) represents the set of parameters upon which both tiers of authorities prefer centralized (decentralized) enforcement. In these regions, a consensus regarding which enforcement strategy should be implemented is achieved. Similar to the results in Section 3.1, the enforcement structure that yields a higher total surplus will gain more political support. We are interested in whether the central authority and the regional authorities would reach a consensus if they voted on the enforcement structure. The striped area in the figure represents the set of parameters upon which the central authority prefers centralized enforcement, whereas the regional authority opts for enforcement delegation. This happens when the regional authority’s level of environmental concern is relatively low and when regional inspection effectiveness is highly heterogeneous with moderate relative environmental concern. When the regional environmental concern is relatively low, the regional authority favours decentralization to exercise local discretion, whereas the central authority wishes to impose centralization to prevent lax enforcement, leading to a preference-driven divergence. In addition, when regional inspection efficiency is highly heterogeneous, and the relative environmental concern is only moderate, the regional authority prioritizes the efficiency gains of delegation, while the central authority remains cautious and prefers to retain control to ensure national standards. Hence, in these cases, the parties would not agree upon an enforcement setting.

Figure 1.

Authorities’ preferences for enforcement structure.

Figure 1 Long description

Legend: Central Authority Preference shown with a dashed line. Regional Authority Preference shown with a solid line. Region C: Centralization. Region D: Decentralization. Striped Area: Conflict. The horizontal axis is labeled, Relative Environmental Concern, open parenthesis 2 lambda subscript 0 over lambda superscript C close parenthesis. The axis runs from 0.0 to 1.0 with labeled ticks at 0.0, 0.2, 0.4, 0.6, 0.8 and 1.0. The vertical axis is labeled, Relative Heterogeneity, open parenthesis theta bar superscript I over theta bar close parenthesis. The axis runs from 0.0 to 1.0 with labeled ticks at 0.0, 0.2, 0.4, 0.6, 0.8 and 1.0. A solid curved boundary starts near x equals 0.35 at y equals 1.0, then curves down and right, passing near x equals 0.40 at y equals 0.60, near x equals 0.50 at y equals 0.45, near x equals 0.60 at y equals 0.38, near x equals 0.80 at y equals 0.31 and reaching near x equals 1.0 at y equals 0.29. The area above and to the right of this solid curve is labeled with a large letter C. A dashed curve appears in the lower right. It begins near x equals 0.78 at y equals 0.0 and rises toward the right edge, meeting near x equals 1.0 at y equals 0.29. A region in the lower right corner is labeled with a large letter D. The striped area covers most of the left and lower portion of the plot, extending up to the solid curve and also occupying the area above the dashed curve and below the solid curve on the right side.

This result is in line with the existing literature that environmental policy is likely to create tensions between levels of government (Wälti, Reference Wälti2004). Naturally, local authorities desire the power that comes along with decentralized authorization, while the central authority tends to limit regional power expansion. The tensions between different levels of enforcement agencies may induce higher noncompliance rates, more counteractions or other forms of conflicts; thus, effective negotiation that allows for the expression of concerns, monitoring rules and incentives for compatibility between different levels of government are essential to the decision-making around enforcement structure.