Comparing Environmental Policy Instruments in the Presence of ...

Environmental & Resource Economics (2005) 32: 337–365 DOI 10.1007/s10640-005-6646-6

Springer 2005

Comparing Environmental Policy Instruments in the Presence of Imperfect Compliance – a Case Study SANDRA ROUSSEAU* and STEF PROOST K.U.Leuven, Centre for Economic Studies, Naamsestraat 69, 3000 Leuven, Belgium; *Author for correspondence (e-mail: [email protected]) Accepted 14 April 2005 Abstract. In this paper, we aim to include rule making, implementation, monitoring and enforcement costs into the cost comparison of policy instruments. We use a simple partial equilibrium model and apply it to the textile industry. The model includes discrete abatement functions and costly monitoring and enforcement. The case study uses individual firm data to simulate the differences in abatement costs and compliance decisions between firms. We compare combinations of regulatory instruments (emission taxes, emission standards and technology standards) and enforcement instruments (criminal fines, civil fines and transaction offers). We show that the inclusion of information, monitoring and enforcement costs indeed alters the relative cost efficiency of the different instruments. Key words: efficiency, environmental management, government policy, illegal behaviour and the enforcement of law, transaction costs JEL classifications: D23, D61, K42, Q2, Q28

1. Introduction Traditionally1 market-based instruments, such as emission taxes, are assumed to be more cost efficient than command-and-control instruments. By design, market-based instruments equalise marginal abatement costs across firms and industries. Rule making costs and enforcement issues are, however, often left out of the comparison. Firms are assumed both to act in good faith and to have full control over their levels of discharges. Reality, however, differs from this idealised picture and firms tend to disobey environmental regulation quite frequently. In this paper we aim to include rule making, implementation, monitoring and enforcement costs for both firms and government into the cost comparison of policy instruments. We use a simple partial equilibrium model and apply it to the textile industry in Flanders. The model we use includes discrete abatement costs functions and costly monitoring and enforcement. The case study uses individual firm data to simulate the differences in abatement costs and compliance decisions between firms. We compare combinations of reg-

338

SANDRA ROUSSEAU AND STEF PROOST

ulatory instruments (emission taxes, emission standards and technology standards)2 and enforcement instruments (criminal fines, civil fines and transaction offers). We show that the addition of information, monitoring and enforcement costs to the model alters the relative cost efficiency of the different instruments under consideration. Monitoring and enforcement issues were largely ignored3 until the seminal contribution of Becker in 1968. Becker developed a normative economic theory of crime and punishment. Together with the paper of Stigler (1970) it contained several interesting insights that would challenge later researchers. The authors assumed rational behaviour among criminals, leading ultimately to an ‘‘efficient’’ level of crime, where the marginal cost of enforcement is equated to the marginal social benefit of the reduction in crime per unit of enforcement. Therefore, given individual preferences and enforcement technologies, both crime rate and level of monitoring and enforcement are endogenous in their model. The Becker model has been considerably expanded during the last three decades4. The first to develop a formal model and look at the impact of incomplete compliance on the firmÕs reaction to environmental policy instruments was Harford (1978). He derived some important results on the comparative statics of firm behaviour. However, his main focus was not on the efficiency properties of different environmental policy instruments. Malik (1992) already showed that the decision rules for choosing between commandand-control policies and incentive-based policies by minimising enforcement costs or minimising abatement costs are different. A recent paper by Sandmo (2002) provides us with a rigorous analytical formulation of the effects of tax evasion and quota violation. The paper explores under which circumstances the efficiency properties of taxes continue to hold even under imperfect compliance. Moreover, it studies to which extent the fine for quota violations plays the same role as an emission tax with similar efficiency properties. Even though the theoretical foundations of instrument choice under incomplete compliance are now well known, their empirical verification is lagging behind. In our paper we attempt to fill this gap. This allows us to better balance the efficiency benefit of taxes with the possible enforcement benefit of standards. In the following section we describe the theoretical framework. Next, we focus on the assumptions underlying the case study. In the fourth section we construct the welfare function for the different instrument combinations and discuss the results of the case study. 2. Theoretical Framework In the next section we define some terms which we will use in the remainder of the text. Then we describe the monitoring and enforcement policy and the behaviour of the firms under three different environmental policy

339

COMPARING ENVIRONMENTAL POLICY INSTRUMENTS

instruments: the emission standard, the emission tax and the technology standard. Finally, we look at the behaviour of the government. 2.1.

SOME DEFINITIONS

For the regulator there are three stages in selecting an environmental policy: the rule-making stage, the implementation stage and the enforcement stage (see Table I). In the rule-making stage the regulator chooses how to tackle the pollution problem. Discussions with administrations and interest groups are held to decide on the environmental goals and on the instruments used to attain those goals. Costs linked to this stage are called rule-making costs (CR). These costs are borne both by government and firms. In the implementation stage the environmental regulation is in force and in order to ensure its correct implementation some extra regulation is needed. Costs linked to this stage are abatement costs (AC) and administrative

Table I. Structure of the regulatory process

Stage 1 – Rule making stage

Stage 2 – Implementation stage

Stage 3 – Enforcement stage

Description

List of instruments

The regulator chooses the instrument to tackle pollution. Discussions with administrations and interest groups. Costs linked to this stage are called rule-making costs (CR) The environmental regulation is in force and to ensure its correct application some extra regulation is needed. Costs linked to this stage are abatement costs (AC) and administrative implementation costs (CI) The compliance with the regulation needs to be ensured. A monitoring and enforcement policy is needed. Costs linked to this stage are the monitoring (CM) and enforcement (CE) costs

Emission tax, emission standard, technology standard

Documentation duty, notification duty (1 and 2), inspection and maintenance duty

Criminal fine, civil fine, transaction offer

340


implementation costs (CI). In the enforcement stage compliance with the regulation is ensured. A monitoring and enforcement policy is developed. Costs linked to this stage are the monitoring (CM) and enforcement (CE) costs. For a more detailed study of the legal and administrative process we refer to Billiet et al. (2002). In each stage the government uses policy instruments. A selection of these instruments can be found in Table I. In our model we include the following rule making instruments: an emission tax, an emission standard and a technology standard. In the implementation stage the policy maker can choose among three instruments: a documentation duty5, a notification duty6 and an inspection and maintenance duty7. Finally, we also distinguish three enforcement instruments8: a criminal fine, a civil fine and a transaction offer9. Most studies do not take these three regulatory stages into account. They only consider the implementation stage and include abatement costs and environmental benefits. In this paper we take the three different stages into consideration. We include, be it in a simple way, the costs of administrative preparation of the regulation as well as the costs of implementation and measurement for firms and government. Finally, we also consider the costs of the inspection and enforcement process for firms and government. The core of the model is a representation of the behaviour of polluting firms10. The firms take the combination of rule making, implementation, monitoring and enforcement policy as given. Once we know the behaviour of the firm for a particular policy combination we can compute total welfare costs and benefits for this instrument combination by adding the environmental damage and the costs of the enforcement agency as well as the rule making and implementation costs for the government. The model is then used to explore a number of possible combinations of policy instruments. 2.2.

POLLUTING FIRMS

In order to concentrate on the choice of instruments and the role of monitoring and enforcement, we assume that firms minimise their costs for any level of output but we do not treat the output decision explicitly. There are n firms in the industry. Once the environmental regulation is implemented firms have to make at most two decisions. First, firms have to decide what abatement technology to use and this determines their emission level. Secondly, in the case of an emission tax, the firm also decides how many emissions they report to the administration. We successively discuss three different rule-making instruments: an emission standard, an emission tax and a technology standard. First, however, we look at the modelling of the monitoring and enforcement policy.


341

2.2.1. Monitoring and Enforcement The monitoring and enforcement policy is modelled in a simple way and is similar to the modelling of Harford (1978) and Malik (1992). The probability of inspection is: pi ¼ p þ /ðsize of violationÞ with 0 pi 1

ð1Þ

Every firm, whether it is violating the environmental regulation or not, will be inspected with a certain fixed probability p. A violator, however, faces an extra possibility of being inspected. This probability /(.) is a function of the level of violation. This does not imply that the agency knows the level of violation or even which firms are in violation. It simply implies that the agency receives a noisy signal of the size of the violation. This signal can, for example, consist of complaints issued by the neighbouring community, environmental pressure groups or civil servants when they notice something suspicious.11 Specifying the variable inspection function /(.), we have for the three types of instruments: Ei E pi;es ¼ min pes þ aes ;1 for an emission standard E Ei ER i ð2Þ pi;et ¼ min pet þ aet ;1 for an emission tax Ei h i for a technology standard pi;ts ¼ min pts þ ats 1 yij^ ; 1 where Ei represents actual emissions, E is the emission standard, EiR reported emissions, yij^ is a 0/1 variable, which is one when the mandated technology ^j is installed by the firm i, and 0 £ am £ 1 with m 2{es,et,ts} is the variable inspection parameter. In the remainder of the text we assume for the variable inspection parameter that aes=aet=ats=a and for the fixed inspection parameter that pes ¼ pet ¼ pts ¼ p. We assume that every violation that is detected leads to a sanction for the violator. This implies that no measurement or judicial errors occur. The three types of sanctions12 we use are a function of the degree of violation and of a penalty parameter pm that has different dimensions13 for each instrument m. We have: 0Þ for an emission standard Fi;es ¼ pes maxðEi E; Fi;et ¼ pet s Ei ER for an emission tax i h i Fi;ts ¼ pts 1 yi^j for a technology standard

ð3Þ

The following conditions hold for the penalty parameters: pes ‡ 0, pet ‡ 1 and pts ‡ 0. Firms violating an emission tax always have to pay at least the evaded amount of taxes.

342


Finally, we assume that firms and government know the relation between the level of violation, the probability of inspection and the sanction. 2.2.2. Emission Standard As output levels are fixed, the only decision to be taken by the firm is the level of abatement and the type of abatement technologies to be used. We model abatement decisions as discrete variables yij that equal one if abatement technology14 j is put into operation by firm i and zero otherwise. The emission abatement realised by installing technology j is represented by Eaij. Firm i minimises the expected costs associated with the regulation in force. ! P ACij yij , rule-making costs (CfR;es ), These costs include abatement costs j

administrative implementation costs (CfI;es ), expected monitoring and enforcement costs (pi,es[CfM;es +yFi,es CfE;es ])15 and the expected sanction (pi,esFi,es). Some of these costs are identical for all firms and are marked with the index f. We assume that firms always incur the rule making and implementation costs even if they decide not to comply. Firms make an informed decision and consequently have to gather information about the environmental regulation and abatement technologies. The difference in costs between a compliant and a noncompliant firm are the abatement costs, the expected enforcement costs and the expected sanction. Formally firm i minimises total costs TCi and faces the following optimisation problem16: X min ACij yij þ CfR;es þ CfI;es þ pi;es CfM;es þ yFi;es CfE;es þ pi;es Fi;es yij

j

with Ei ¼ Eoi

X

EAij yij

j

and yij ¼ 1

if technology j is chosen by firm i

¼0

otherwise

yFi;es ¼ 1

if Fi;es > 0

¼0

pi;es Fi;es

if Fi;es ¼ 0 Ei E ¼ min p þ a ; 1 E ¼ pes maxðEi E; 0Þ

ð4Þ

Total abateThe firm emits Ei and is subject to an emission standard E. ment costs equal the sum of costs ACij of installing technology j in firm i


343

if that particular technology is selected to be implemented (yij=1). When the firm is violating the environmental policy it faces an expected sanction pi,esFi,es. The fine depends on the size of the violation and the penalty parameter pes. The rule-making and implementation costs include, among others, the firmsÕ additional administrative costs. Managers need to be informed about their legal obligations and the implications for their company. They may need to apply for a licence. Moreover they need to collect information about the technological possibilities to comply with the standard. Measurement of emissions is necessary to evaluate the current compliance status. The monitoring costs are incurred every time an inspection is performed on the firmsÕ premises. Examples of these costs are the costs of having to accompany the inspector on site and to perform a second test if necessary. Enforcement costs are only relevant if a firm is actually fined. Examples are costs of legal representation and court costs. We cannot derive general first-order conditions from problem (4) since our decision variables are discrete. Firms decide which abatement technology to install based on a very simple cost minimisation decision rule: they install technology j if total costs are smaller with that technology than without. Once the abatement decision has been taken, actual emissions are determined and also the degree of violation. Note that due to the indivisibilities in the abatement technologies, firms can overcomply with the regulation. The extra emission reductions benefit society but not the firms. 2.2.3. Emission Tax For an emission tax s the firmÕs minimisation problem is: h i X f f f f ACij yij þ s ER þ C þ C þ p C þ y C minR i;et F i;et i R;et I;et M;et E;et þ pi;et Fi;et yij ;Ei

j

with yij ¼ 1 if technology j is chosen by firm i ¼ 0 otherwise yFi;et ¼ 1 if Fi;et > 0 ¼ 0 if Fi;et ¼ 0 Ei ER i ;1 pi;et ¼ min p þ a Ei

Fi;et ¼ pet s max Ei ER i ;0

ð5Þ

Firms that are subject to an emission tax report a certain amount of emissions EiR to the government. They pay taxes on these reported emissions.

344


However, if a firm reports less than the actual amount of emissions, it is in violation and faces a penalty. The difference between actual and reported emissions is never negative if the firm behaves rationally. In analogy to the emission standard, the rule making, implementation and expected enforcement costs are identical for all firms but may be different from those for an emission standard. Firms now not only face information costs but also the costs for the yearly tax report. Data must be collected and reported. Calculations must be made. Moreover, the firm also has to perform measurements to know its actual emissions. These costs can be substantial. P EAij yij ; Actual emissions Ei of firm i are equal to the difference EO i j

with EiO equal to the firmÕs initial emissions before any abatement technology P EAij yij equal to the amount of emissions reduced has been installed and j

by abatement. The first-order condition that determines how many emissions (EiR) a firm will report is: ! f f C þ y C 1 p p F i;et et M;et E;et ; Ei 1 Ei þ ER i ¼ min 2a pet 2pet s

ð6Þ

Firms can either report truthfully or they can report less than their actual emissions. The amount they will report consists of two terms. The first term is a fraction of actual emissions and depends on the monitoring and enforcement policy parameters but not on the emission tax rate. The second term is a correction for the monitoring and enforcement costs the firm bears. This term, however, does depend on the tax rate and on the fine parameter but not on the inspection parameters. The monitoring and enforcement costs borne by firms cause firms to report their emissions more truthfully in order to avoid some of these costs. Firms minimise the costs associated with the environmental policy. They will invest in abatement if the costs of doing so are smaller than the corresponding decrease in taxes paid and expected fine. 2.2.4. Technology Standard A technology standard forces the firm to use a particular abatement technology or production process. The firmÕs choice space is therefore limited. Either they comply with regulations and install the technology ^j or they are in violation. Abatement costs can differ between firms since we allow for firm heterogeneity. For each firm these costs are either zero or ACi^j since it is never optimal to use other abatement technologies. The firmÕs objective function is:


345

min ACi ^j yi ^j þ CfR;ts þ CfI;ts þ pi;ts CfM;ts þ yFi;ts CfE;ts þ pi;ts Fi;ts yi ^j

with yi^j ¼ 1 ¼0

if techn j^ is chosen by firm i otherwise

yFi;ts ¼ 1 ¼0 pi;ts Fi;ts

if Fi;ts > 0 if Fi;ts ¼ 0 ¼ min p þ a 1 yi^j ; 1 ¼ pts 1 yi^j

The standard requires the installation of one particular technology. Implementing this technology can lead to different costs for each firm. Each firm has two options: either it complies with the standard and installs the technology at cost ACi^j or it does not install the technology and incurs no costs. A firm will comply with the technology standard if it costs less than the expected fine and expected enforcement costs. This decision implies a corner solution for the firm. Again rule making, implementation and expected enforcement costs are identical for all firms. Costs for information acquisition are limited in size since the regulation already indicates which technology must be used. There is no need to know alternatives or even actual emissions. 2.3.

GOVERNMENT

Even though in reality the environmental inspection agency, the legislator and the judicial institutions are separate branches within the governmental institution, in this model we aggregate their objective functions into one social welfare function. We therefore assume that these agents all share the same objective, i.e. maximising social welfare. This is a rather limiting assumption. After all, enforcement authorities can have different goals than the legislator. Keeler (1995) believes that in practice achieving compliance will be a primary goal of the enforcer. However, empirical evidence on the agencyÕs objective function is in short supply. Dion et al. (1998) have, for example, analysed the regulatorÕs monitoring decision and found evidence supporting both the public interest theory of regulation and the economic theory of regulation. The public interest (or normative) theory would explain environmental regulation as an instrument that corrects market failure and increases social welfare (Posner 1974). The economic (or positive) theory of regulation, on the other hand, assumes that the regulator wants to maximise net political support. Within such model of political support maximisation

346


different interest groups compete in order to influence environmental regulation17. We follow the public interest theory and assume that the government (and all its components) wants to maximise social welfare. We do not model interest group behaviour although we take some of the costs associated with lobbying into account when defining rule making, implementation, monitoring and enforcement costs. Formally, a benevolent government looks for a combination of instruments that maximises social welfare (SW) and this is equal to: i i1 Ph h f pi;m CM;m þ yFi;m CfE;m þ pi;m Fi;m EQ þ PS n:CfR;m n:CfI;m B i h i C maxSW ¼ max@ A P R sEi þ pi;m Fi;m CgR;m CgI;m pi;m CgM;m þ yFi;m CgE;m þMCPF 0

ð8Þ

i

for each m 2{es,et,ts}. Social welfare consists of environmental quality (EQ), producer surplus

PS

n:CfR;m

n:CfI;m

i i Ph h f f pi;m CM;m þ yFi;m CE;m þ pi;m Fi;m ,18 i

and governmental budget surplus corrected with the marginal cost of public funds (MCPF). The government budget surplus is a consolidated budget that includes pollution tax revenues (s EiR), the enforcement agency costs P g g pi;m CM;m þ yFi;m CE;m as well as the revenues from fines (pi,m Fi,m) and i

g

g

other environmental administrative costs (CR,m and CI,m ). In the social welfare function, we include not only all rule-making, implementation and enforcement costs associated with a particular set of instruments but also environmental benefits. Environmental benefits are included to allow us to deal with the indivisibilities of the abatement technologies that may cause overcompliance and make comparisons across instruments more difficult (Oates et al. 1989). Rule making costs for the government result from meetings within the administration and with interest groups and experts. Governmental implementation costs have to do with, for instance, distributing regulatory information through official publication of laws and statutes. Enforcement costs include inspection and prosecution costs.

3. Empirical Illustration In order to illustrate our theoretical model we decided to focus on the textile industry in Flanders (one of three regions in Belgium). In the first section we describe the construction of the benchmark. In the second section we select and specify the regulatory combinations. Thirdly, we estimate the monitoring and enforcement parameters. Next, we look at the estimation of the abatement cost


347

functions and finally we describe the rule-making, implementation, monitoring and enforcement costs.

3.1.

CONSTRUCTION OF THE BENCHMARK

We concentrate on water pollution caused by textile improvement and carpet production. These two subsectors are, after all, responsible for most of the water pollution attributable to the textile industry. Several sector studies (Centexbel 1996; Jacobs et al. 1998; OVAM 1996; PRESTI 1994–1997) provide us with useful information. For reasons of tractability we limit our study to water pollution caused by BOD19 emissions and we only consider point sources. In our benchmark scenario there is no environmental regulation in place. We do, however, assume that all necessary legal and economic institutions are already in place; such as the environmental inspection agency, courts, parliament... Finally, the marginal cost of public funds20 is set to 1 and the willingness to pay for an improvement in water quality equals e 31 per year for each tonne of BOD removed (Billiet et al. 2002). We will provide a sensitivity analysis of this estimate. 3.2.

SELECTION AND SPECIFICATION OF THE REGULATORY COMBINATION

When we combine all possible instruments in the three stages (rule making, implementation and enforcement) we obtain nine combinations that could be of interest (Billiet et al. 2002). The instruments used in the implementation stage can only be used in particular combinations with rule-making instruments. We analyse, for each of the three sanctions, – an emission tax with a documentation duty and a notification duty (type 2)21, – an emission standard combined with a maintenance duty, and – a technology standard with a maintenance duty and a notification duty (type 1). For each of these combinations we will consider several values for the instruments so that we can select the best combination of instruments. 3.3.

MONITORING AND ENFORCEMENT PARAMETERS

For the region and sector considered it is reasonable to take the fixed inspection probability p equal to 0.122. Next we assume that the variable inspection parameter a is equal to 0.5. The probability of inspection is

348


therefore proportional to the size of the violation. For the a-parameter we have no reality check. This means that we will need a sensitivity analysis. Finally, we assume that the penalty parameter for the emission tax pet is equal to 2. In Billiet et al. (2002), we find that in Belgium the penalty for evading an emission tax is typically twice the evaded amount. Since we have no information on the other instruments, we assume for the penalty parameter for the emission standard that pes=2 e/kg BOD. The penalty parameter pts for a technology standard has a different dimension and is assumed to equal 50,000 euro. In section IV we perform a sensitivity analysis on these parameters. 3.4.

ABATEMENT COST FUNCTION

Explicitly modelling the firmsÕ heterogeneity is important to capture the advantages of market-based versus command-and-control instruments. Therefore we made use of a survey at firm level on abatement costs. We first contacted by mail 106 Flemish companies active in textile improvement and carpet production. Then we conducted a follow-up interview on site. We obtained useful cost estimates from 20 firms23. We asked firms to state the costs of presently installed abatement technologies and of planned investments in the next two years. These data were used to estimate abatement cost functions for each company in order to represent firmsÕ heterogeneity. The cost estimates take both fixed and variable costs into account. We include initial investment costs, subsidies, personnel, energy and other costs. The life span of an investment is assumed to be 20 years. We assign all costs to only one pollutant, i.e. BOD, and therefore assume that the sole purpose of the investment is to reduce BOD emissions24. After calculating the net present value (NPV) of each technology we derive the associated annuities and use these in the model. An extensive range of technologies was reported including filters, good housekeeping, the use of different inputs and wastewater treatment. Cost differences of abatement technologies between firms turn out to be large indeed; cost estimates (NPV) for building a water purification station ranged from 1 million euro to 4.7 million euro. 3.5.

RULE-MAKING, IMPLEMENTATION AND ENFORCEMENT COSTS

The enforcement, inspection and information costs were estimated using the firm survey, by checking court rulings and by interviewing experts in the administration and in the legal profession25. We identify cost factors that result from the legal context and from the instrument itself (see Billiet et al. 2002). For each of these cost factors we performed a relative valuation per instrument and per agent; and we took into account the different stages of the

349


Table II. Rule-making, implementation, monitoring and enforcement costs (in man-days) Instrument

Emission standard Emission tax Technology standard Documentation duty Notification duty 1 Notification duty 2 Inspection and maintenance duty Criminal fine Civil fine Transaction offer

Government

Firms

CR

CI

CM

CR

CI

21

0.4 + print/5

2.5

0.2

133 18.75

5 2.5

1 0.25

3.4

6 + print 0.5 + print/4 0.1

1.5

–

1.4

–

1.55

0.05

–

–

0.025

–

7.4 + print*2/5 5.1

1.8 + print*2/5 –

1.25

–

1.7

0.75

–

–

–

–

33 62.5 63.5

34 65 2

– – –

– 0.25 –

3+ lab/5 20 2.125

39 13 1.25

CM 3.5 8 1

– – –

Print = printing costs of new laws or rule books Lab = costs of having samples tested in a laboratory

regulatory combination. We include a wide variety of costs: the costs of lobbying, filling in forms, communicating with the administration, performing inspections, internal meetings, legal counselling... Details on the construction of this table can be found in Rousseau (2005). We have estimated and taken into account how often a particular instrument is used or changed by the government, etc. The legal framework is such that the level of the emission tax has to be approved by the parliament each year. This explains the higher rule making costs for an emission tax compared to an emission standard (holds on average four years) and a technology standard (holds on average 5 years). The results are summarised in Table II. In Table II the rule making and implementation costs (CR and CI) are presented on a yearly basis. The monitoring costs CM represent the costs of one audit to the firm or government. Since the amount of inspections performed depends on the compliance decision of the firm we could not calculate these costs on a yearly basis. The enforcement costs CE are derived from the different types of costs associated with an enforcement instrument. These costs are again endogenous as they depend on the firmsÕ compliance decisions.

350


The relative cost differences between the instruments are the core of this analysis. We do not want to stress the absolute values of these cost estimates. However, in order to calculate welfare effects we will need a monetary estimate of these costs26. 4. Results First, we compare the costs associated with three environmental policy instruments in a base scenario holding constant all other dimensions of instrument choice. We continue by examining the importance of the willingness to pay for water quality improvements and by testing the sensitivity to monitoring and enforcement parameters and to the rule making, implementation, monitoring and enforcement costs. 4.1.

COMPARING THE COSTS OF USING THREE ENVIRONMENTAL POLICY INSTRUMENTS

Next, we consider the welfare effects and emission levels for different emission tax rates. Secondly, we illustrate the costs associated with the different policy instruments. Finally, we compare the welfare effects of the emission standard with the emission tax and the technology standard.

2 000 000 1 800 000 1 600 000 1 400 000 1 200 000

1 117 752 1 115 541

1 000 000

1 117 209

800 000 600 000 400 000 295 618

200 000

294 432

0 0

0.01

0.02

0.03

0.04

0.05

tax rate ( /kg BOD) Welfare (in K Euro)

BOD emissions (in 20 tonne)

Figure 1. Emission tax (civil fine and WTP = 31e).

0.06

351


4.1.1. Welfare Effects and Emission Levels for Different Emission Tax Rates First we plot in Figure 1 the welfare effects and emission levels for one instrument combination, i.e. an emission tax combined with a documentation duty, a notification duty and a civil fine. We consider the effect of implementing different tax rates for given inspection probabilities and fine parameters. BOD emissions (measured on the vertical axis in multiples of 20 ton) reduce stepwise with an increasing tax rate because of the technological indivisibilities included via the abatement cost curve. It is only profitable to invest in a particular abatement technology when the tax rate exceeds a certain level. The abatement cost function is therefore not continuous. Due to the same reason the welfare level (measured on the vertical axis in KEuro) changes stepwise. Once the tax rate exceeds a threshold value, the welfare level jumps up only to slowly decline until another threshold is exceeded. This slow decline in welfare is caused by the increase in expected monitoring and enforcement costs without a compensating decrease in emissions because one has reached the region of strongly increasing abatement costs. 4.1.2. Cost Estimates for the Environmental Policy Instruments In order to provide some insight into the relative impact of the regulatory costs on social welfare, we give in Table III the regulatory costs for firms and government. The rule-making and implementation costs (CR and CI) do not vary with the specification of the policy instruments; they are fixed costs associated with the use of a particular environmental policy instrument. The Table III. The importance of the different costs (in KEuro) Emission tax 0.00074 e/kg BOD discharged Emissions = 22.5 mio tonnes BOD Social welfare = 182 mio e Emission standard = 293 mg BOD/litre waste water Emissions = 22.4 mio tonnes BOD Social welfare = 165 mio e Technology standard = heat exchange technology Emissions = 22.1 mio tonnes BOD Social welfare = 190 mio e

Firm Government

AC

CR

6 111.3 –

CI

119.9 3 569.7 = 4 228.0 84.2 29.2

CM + CE 638.4 547.9

= 661.3 Firm Government

26 983.3 –

28.6 342.7 = 630.4 14.6 2.6

259.1 211.3

= 228.5 Firm Government

2 546.4 –

34.3 245.6 = 365.1 14.3 3.3 = 117.2

85.2 99.6

352


abatement, monitoring and enforcement costs (AC, CM and CE) are, however, susceptible to changes in the policy specification. We, therefore, want to stress that the results for these costs change substantially according to the specification of the instrument under consideration. The numbers in Table III present only one possible specification for three instruments27. They are far from general. First of all, we notice that total costs for firms, excluding abatement costs, are more than twice those for the government for all three instruments. In the case of the emission tax, costs for the firm are even more than six times those for the government. The fixed rule making, implementation, monitoring and enforcement costs borne by the government differ over instruments. Emission taxes cost the government more or less three times what an emission standard would cost. This observation may point to one possible reason why standards are preferred by governments. From the governmentÕs administrative point of view it is often cheaper to issue an emission standard than to use an emission tax28. Although it performs quite well in Table III, the technology standard is less interesting, because it does not allow one to choose the resulting emissions. We will return to this point later. The emission tax leads to higher welfare than the emission standard even though the costs of an emission tax for the government are twice those of an emission standard. This result is due to the fact that costs for firms are much lower with an emission tax than with an emission standard. This is something we expected since taxes allow firms to equalise marginal abatement costs (and minimise total costs) while standards do not. The reason why technology standards lead to such low abatement costs is due to the model formulation29. As in Table II, we also find that the governmentÕs rule making costs for emission taxes are much higher than those for the other instruments. This is due to the Belgian constitution, which requires that all taxes are approved by parliament each year. For firms, however, emission taxes are by far the most expensive instrument if we do not consider abatement costs. The high implementation costs associated with emission taxes are clearly to blame. Tax forms have to be filled in every year. Gathering all necessary information and processing the data requires a lot of administrative resources. It is also important to notice that the emission standard is considered as such and not included in a licensing scheme. However, if we include abatement costs, we immediately see that efficiency gains from using emission taxes more than compensate for the higher costs for firms. In effect social welfare with the emission tax is higher than with the emission standard for the same reduction in emissions (see Table III).

353


We also want to remark that the monitoring and enforcement advantage of a technology standard over the other instruments can also be observed in Table III. Both firms and government have considerably lower costs associated with a technology standard. It is after all fairly easy to observe whether a firm complies with a technology standard or not. Finally, we compare our estimates with the prior estimates of regulatory costs. Probst et al. (1995) and Dixon (1995) looked at the administrative and compliance costs connected to the US Superfund program.30 Since Superfund introduced several special taxes with different tax rates rather than an increase of an existing tax (such as corporate or personal income taxes), the program is financed in a very expensive way. Slemrod and Sorum (1984) estimated the taxpayersÕ compliance costs with the individual income tax were 5 to 7 percent of collected tax revenues. In Table III, the sum of CR, CI, CM and CE is 36 percent of environmental tax revenue. This higher cost estimate makes sense since the environmental tax is a more inefficient tax than personal income tax and it also collects only a fraction of the latterÕs revenues. Hopkins (1996) estimated the costs of paperwork associated with US federal regulations at 2000$ per employee in 1992. He also found that smaller firms have relatively higher regulatory costs. Average employment in the Flemish textile improvement industry during the nineties was 4500 employees. This allows us to calculate an average implementation cost of 790 e per employee per year.

200 000

180 000 170 000 160 000 150 000 140 000 130 000 40 000 000 35 000 000 30 000 000 25 000 000 20 000 000 15 000 000 10 000 000

5 000 000

BOD emissions (in tonne)

Emission standard

Emission tax

Technology standard

Figure 2. Combinations with civil fine (no valuation of env. benefits).

0

Wel fare (i n K Euro)

190 000

354

SANDRA ROUSSEAU AND STEF PROOST 196 000

192 000 190 000 188 000 186 000 184 000

Welfar e (in K Euro)

194 000

182 000 180 000 39 000 000 37 000 000 35 000 000 33 000 000 31 000 000 29 000 000 27 000 000 25 000 000

BOD em issions (in tonne) Emission standard

Emission tax

Technology standard

Figure 3. Combinations with civil fine: detail (no valuation of env. benefits).

4.1.3. Welfare Comparison of the Different Instruments In the base scenario we decided to initially exclude environmental benefits. A sensitivity analysis will later check the influence of environmental benefits on the results. In Figure 2 we plot the change in welfare for the three regulatory instruments combined with a civil fine compared to the benchmark scenario with no environmental regulation in place. In Figure 3 we give a detail of Figure 2. In these graphs we measure the emission level on the horizontal axis and the welfare level on the vertical axis. We immediately see that a certain amount of rest emissions (namely 5563 Kton BOD) will persist whatever firms or government do. This is due to our assumptions that there is no perfectly clean technology and that the industryÕs output remains fixed. Our results indicate that in most cases emission taxes are the cheapest instrument to use. Only for minor reductions an emission standard outperforms an emission tax because the gains in abatement cost efficiency are less important and the fixed set-up costs weigh more heavily on an emission tax. We can nevertheless say that an emission tax is less costly than an emission standard for most of the larger emission reduction levels. These results are sensitive to changes in the monitoring and enforcement parameters and to legal procedures. Both emission taxes and emission standards can be useful since they can each reach their own specific set of emission reductions. The costs curves are after all discontinuous. Technology standards prove to be a very limited instrument since they can only reach as many points as there are possible abatement technologies. We have also compared the different sanctioning instruments for an emission tax. The criminal fine turns out to be the most expensive instrument

355


to use and, as could be expected, the transaction offer is the cheapest to use. However, in reality these three instruments are often used as complements. For minor violations or first time offenders, a transaction offer will often suffice. A criminal fine will be used for serious violations or extremely uncooperative firms. The administrative fine also has its specific use. Using an administrative fine avoids the social stigma associated with criminal fines. Therefore, we cannot a priori choose one of the enforcement instruments as being Ôthe bestÕ. We need a dynamic model to study the power of this type of strategies more closely. An example of a model that punishes repeat offenders more severely is Harrington (1988). 4.2.

THE ROLE OF THE WILLINGNESS TO PAY FOR WATER QUALITY IMPROVEMENTS

In the previous analysis we have excluded environmental benefits. The willingness to pay (WTP) for quality improvements matters for two reasons. First, the environmental benefit has to be high enough to implement an environmental policy as there are important fixed costs to implement a regulation or a tax. Second, the environmental benefits may reward the overcompliance associated with certain policy instruments. Our sensitivity analysis shows that the results are – in certain aspects – highly sensitive to the differences in the WTP for water quality improvements. As expected, the higher the WTP for the improvement in environmental quality, the more it 1 200 000

800 000

600 000

400 000

200 000

0 40 000 000

35 000 000

30 000 000

25 000 000

20 000 000

15 000 000

10 000 000

5 000 000

BOD emissions (in tonne) WTP not taken into account

Figure 4. Emission tax (civil fine).

WTP=31

WTP=3.1

0

Welfare (in K Euro)

1 000 000

356 238 000

210 000

237 000

205 000

236 000

200 000

235 000

195 000

234 000

190 000

233 000

185 000

232 000

180 000

231 000

175 000

230 000

170 000 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

BOD em issions (in 100 tonne)

Social welfare (in K Euro)


1

fixed inspection frequency (pbar) Social welfare (in K Euro)


Figure 5. Emission tax rate = 0.0012 e/kg BOD (WTP = 3.1e).

pays to pursue an environmental policy even if the emissions are only minimally reduced. See, for example, Figure 4 for the welfare curves associated with three different levels of WTP for the emission tax combined with a transaction offer31. The changes in WTP, however, do not influence the relative position of the different instruments since the differences in rankings are driven largely by differences in policy related costs. This implies that the overcompliance benefits associated to standards (Oates et al. 1989) are insufficient to overrule the cost advantage of emission taxes. 4.3.

THE IMPORTANCE OF THE MONITORING AND ENFORCEMENT POLICY

Sensitivity analysis with respect to the enforcement parameters (fixed inspection probability, coefficient of the variable inspection probability and the penalty coefficient) shows their immense importance. In Figure 5 we have plotted the change in welfare and emission reduction for one particular tax rate (s=0.0012 e/kg BOD) when the fixed inspection frequency pi changes from zero to one. An optimum is reached for a fixed inspection frequency equal to 0.5. Reducing the parameter from 50% to 40% decreases welfare by approximately 615,000 euro. It is important to note that the optimum is not found at one of the extremes and that not all increases in inspection probabilities affect emissions. It is not optimal to inspect all firms every year. The improvement in environmental quality would not counterbalance the additional costs. However, we also see that even a small increase (e.g. from 0.1 to 0.2) in the inspection probability can significantly increase social welfare.

357


240 000

350 000

300 000

230 000 225 000

250 000

220 000 200 000

215 000 210 000

150 000

BOD em issions (in 100 tonne)


235 000

205 000 200 000

100 000 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

variable inspection parameter (alfa) Social welfare (in K Euro)



In Figure 6 we perform the same exercise for the variable inspection parameter a. The higher the variable inspection parameter, the higher social welfare will be. However, increasing the variable inspection parameter from 0.4 to 0.5 improves social welfare by 5 million euro, while increasing a from 0.5 to 1 additionally increases welfare by only 92,000 euro. We also see that improving the parameter a from 0.1 to 0.2 increases social welfare by 300 000 250 000 200 000 150 000 100 000 50 000 0 0

2

4

6

8

10

Penalty param eter (pi)




12

358


approximately 31.5 million euro. We observe again that also a less than optimal effort can greatly improve social welfare. Tests have shown us that both inspection parameters (fixed inspection frequency and variable inspection parameter) influence each other. In order to optimise the inspection policy, government will have to fix both parameters simultaneously. This surely is an exercise worth doing since welfare can increase considerably as shown in Rousseau and Billiet (2003). Finally, we perform the same exercise for the penalty parameter pet (Figure 7). One immediately sees that the penalty greatly influences the outcome: resulting emissions range from 37.1 million to 18.1 million tonnes BOD. Welfare levels range from 5 million to 237 million euro. The penalty can be used as substitute for a higher tax up to a certain point. Manipulating the penalty cannot achieve the same emission reduction as manipulating the tax rate beyond the limit of 18.1 million tonnes BOD for this specification. We also find that it does not pay to increase the penalty parameter above 10. Social welfare has reached its optimum and does not change anymore. All firms have decided to report their emissions truthfully and pay the required tax rather than risking the fine. In Rousseau and Billiet (2003) we show that, when the enforcement parameters are set at a suboptimal level, it may be more cost effective not to have environmental regulation at all. 4.4.

THE IMPORTANCE OF THE RULE MAKING, IMPLEMENTATION AND ENFORCEMENT COSTS

In order to investigate the influence of the rule making, implementation, monitoring and enforcement costs on our results, we now assume them to be 200 000

180 000 170 000 160 000 150 000 140 000 40 000 000 35 000 000 30 000 000 25 000 000 20 000 000 15 000 000 10 000 000 5 000 000

0

BOD em issions (in tonne)

Emission standard

Emission tax

Technology standard

Figure 8. Combinations with civil fine (CR = CI = CM = CE = 0 = WTP).

Welf are ( in K Euro)

190 000

359

COMPARING ENVIRONMENTAL POLICY INSTRUMENTS 200 000

180 000 170 000 160 000 150 000

Welfare (in K Euro)

190 000

140 000 130 000 40 000 000 35 000 000 30 000 000 25 000 000 20 000 000 15 000 000 10 000 000

5 000 000

0

BOD emissions (in tonne)

Emission tax no costs

emission tax with costs

Figure 9. Emission tax with civil fine (CR = CI = CM = CE = 0 = WTP).

zero and redo the analysis. Looking at the relative cost comparison between the emission standard, emission tax and technology standard, we find in Figure 8 that the ranking has become more straightforward. Emission taxes outperform emission standards for all possible emission reductions. Taxes even outperform technology standards except for one specification. These results are as expected since we replicate the classical model without rule making, implementation, monitoring and enforcement costs. The cost efficiency of taxes is clear. In Figure 9 we show how social welfare changes for the emission tax with or without rule making, implementation, monitoring and enforcement costs. The curve shifts upwards when those costs are not included in the analysis. Moreover, the optimal tax rate to obtain a certain emission reduction changes. Under the zero-cost regime a slightly higher tax rate is needed to obtain the same emission reduction than under the positive-cost regime. The higher tax rates are necessary to compensate the effect of firmsÕ costs associated with the environmental regulation.

5. Conclusion A more realistic cost comparison of different environmental policy instruments is obtained by including rule making, information, monitoring and enforcement costs for both firms and government into the framework. Using a simple partial equilibrium model and applying it to the textile industry in Flanders allows us to empirically validate our results. We focus especially on

360


the formulation of firm behaviour and concentrate on a more realistic presentation of the firmÕs decision-making process. The model contains discrete abatement functions and costly monitoring and enforcement. The case study uses individual firm data to simulate the differences in abatement costs and compliance decisions between firms. We compare combinations of regulatory instruments (emission taxes, emission standards and technology standards) and enforcement instruments (criminal fines, civil fines and transaction offers). We show that this extended modelling of the costs associated with environmental policies indeed alters the relative cost efficiency of the different instruments. Identifying the policy instrument that leads to the highest welfare level for a given emission reduction is not clear-cut. We find that, for small reductions, an emission standard outperforms an emission tax because the gains in abatement cost efficiency are less important and the fixed set-up costs are more important for an emission tax. Nevertheless, an emission tax is less costly than an emission standard for most of the larger emission reduction levels. These results are sensitive to changes in the monitoring and enforcement parameters and to the legal procedures. Ignoring all policy costs in the model, allows us to replicate the classical result that emission taxes outperform emission standards for all possible emission reduction levels. The technology standard turns out to be a limited instrument. This instrument can only reach as many levels of emission reductions as there are abatement technologies and the obtained result is highly sensitive to changes in the monitoring and enforcement parameters. When it comes to sanctioning instruments, the criminal fine is the most expensive, and, as could be expected, the transaction offer is the cheapest. However, in reality these three instruments (transaction offer, civil fine and criminal fine) are often used as complements. For minor violations or first time offenders, a transaction offer will often suffice. Typical examples here are parking fines. A criminal fine will be used for serious violations or extremely uncooperative firms. The civil fine also has its specific uses. Therefore, we cannot a priori choose one of the enforcement instruments as being Õthe bestÔ. Changes in willingness to pay for water quality improvements do not influence the relative position of the different instruments. Sensitivity analysis shows that the results are, however, in certain aspects, highly sensitive to the differences in the willingness to pay. Higher environmental benefits are important to compensate the fixed costs associated with a policy. Sensitivity analysis with respect to the enforcement parameters (fixed inspection probability, the variable inspection probability and the penalty coefficient) shows their immense importance. Changing the level of the parameters can suddenly make a policy worthwhile pursuing. In Rousseau and Billiet (2003) we derive the optimal monitoring and enforcement strategy for an emission standard as a combination of fixed and variable inspections


361

and penalties. The discreteness of the abatement cost functions leads to optimal fines, which are lower than maximal fines. Underlying our model is the assumption that firms are risk neutral. This assumption is quite defendable when firms operate in a setting of welldeveloped risk markets. It also allows one to easily distinguish between production efficiency and consumption efficiency. However, in some instances risk averse behaviour of firms is clearly a realistic assumption. Sandmo (2002) provides a theoretical analysis of firm behaviour and instrument choice under the assumption that the firm has a utility function that is concave in profits, i.e. the firm is risk averse. He finds that results for the case of an emission tax remain virtually unchanged. However, risk aversion makes much more of a difference in the decision to violate an emission standard, where, as expected, risk aversion makes firms more cautious in exposing themselves to the risk of penalties. In conclusion, adding rule making, implementation, monitoring and enforcement costs of an environmental policy, can drastically change traditional results with respect to the relative efficiency of instruments. Our empirical illustration proves this point by showing how an emission tax can be the most expensive instrument to use in order to obtain a particular level of environmental quality. This result holds even though we include heterogeneity of the industry into our model and allow for the efficiency advantage of taxes over standards. The main disadvantage of emission taxes is the large administrative burden of setting up a tax system. Firms, for example, have to monitor emissions frequently and fill in a yearly – often quite complex – tax form. Also legally, it proves to be more difficult to prosecute tax evaders since taxes are levied on the previous fiscal year. At the moment of trial, it is very difficult to prove the exact amount of emissions during the last year unless continuous monitoring has been performed. This relative disadvantage of taxes over standard is certainly more general than our case study. Acknowledgements We would like to acknowledge the financial support of the Science Policy Office research program – Indicators for sustainable development – contract nr. HL/DD/015 (Law & economics of the choice of environmental policy instruments), the ÔSteunpunt MilieubeleidswetenschappenÕ funded by the Flemish government and the FWO project ÔAn economic approach to modelling enforcement of environmental regulationÕ. We would like to thank C.M Billiet for all the juridical advice and A. Ulph, E. Calthrop, A. Heyes, E. Schokkaert, D. Heremans, T. Ulen, B. De Borger and one anonymous referee as well as seminar participants at UIUC, Indiana University and the World Meeting of EAERE in Monterey for their useful comments on earlier drafts.

362


Notes 1. See, for example, Kolstad (2000). 2. We do not consider liability rules in this paper. 3. Note that economic analysis of public law enforcement dates from the eighteenth century contributions of Montesquieu (1748), Beccaria (1770) and, especially, Bentham (1789), whose analysis of deterrence was both sophisticated and expansive. However, after these contributions, the subject of law enforcement was largely ignored until the late 1960s, when Becker published his article. 4. For recent overviews see Cohen (2000), Polinsky and Shavell (2000) and Heyes (2001). 5. A documentation duty asks the firm to have documentation about, for example, its emissions. Nothing has to be done with the documents; firms just need to have them. 6. A notification duty asks the firm to communicate certain information to the administration; e.g. in order to pay taxes a firm has to notify the administration of its emissions. We distinguish two different types of these duties. The first is a simple and one-off notification duty e.g. the duty to notify the administration fourteen days beforehand that a licensed firm will start operating. The second notification duty is complex and repeating, e.g. a tax report. 7. An inspection and maintenance duty asks firms to maintain and test their installation on a regular basis and often, by an approved expert. 8. The three fines we consider only differ with respect to the associated administrative costs. The way they are modelled is equivalent for all three. 9. A transaction offer is an administrative fine used in Belgium with very low administrative costs. 10. It is possible to include consumers and environmental pressure groups but their detailed modelling falls beyond the scope of this paper. Aidt (1998), for instance, shows that political competition can be an important source of internalisation of economic externalities. Lobby groups give voice to different aspects of environmental policy making sure that all aspects are considered in the political trade off and reflected in the implemented policy. The emphasis in this paper is not on the outcome of this policy game but on the aggregate costs of using different instruments. 11. Internal regulations of the Flemish environmental inspection agency state that every complaint is followed by a site visit within three months. 12. These sanctions have different rule-making and implementation costs but we assume that they do not differ in the way they are modelled nor in the way they influence firm behaviour. 13. The parameter pes has the dimension Euro/kg pollution per year. The parameter pet has no dimensions. The parameter pts is expressed in Euros. 14. The technology j can represent a single technology as well as a combination of different technologies. 15. with yFi,es =1(0) if the fine Fi,es > (=)0. 16. We assume that firms are risk neutral. This is not a harmless assumption when it comes to enforcement. We discuss the implications of this assumption in our conclusions. 17. See, for example, Boyer and Laffont (1999); Aidt (1998); Ando (2003); and Damania (1999). 18. PS is our estimate of the producer surplus in the benchmark or before implementation of the environmental policy. 19. BOD or Biological Oxygen Demand represents the amount of oxygen (mg/l waste water) that certain bacteria use, during five days at 20 C, in order to oxidise organic carbon to carbonic acid. 20. We set the marginal cost of public funds equal to one, and, therefore, assume the presence of perfect lump-sum taxes, in order to be as close as possible to the traditional partial equilibrium comparison of environmental policy instruments. We are aware that the


21. 22.

23. 24.

25. 26.

27. 28. 29.

30.

31.

363

MCPF is influenced by the use of the net tax revenues generated by the environmental policy. This will indeed also effect the position of the different instruments (see, for example, Goulder et al. 1999). Furthermore, it can be shown that the expected fines can function as a price instrument (see Rousseau and Proost 2004). See footnote 7 for the definition of the different types of notification duties. This value is based on a communication from the Ministry of the Flemish Community on 11 June 2001. We read Ô... that every class-I-firm is inspected thoroughly not even once every ten yearsÕ. Since we only consider class-I-firms we can assume that every firm regardless of its compliance status has a ten percent probability of being inspected each year. Our results are, therefore, likely to experience a self-selection bias. In reality investments in abatement technologies often serve multiple purposes and reduce the output of several pollutants. This means that firms can ÔoverachieveÕ and do better than legally required. One way to deal with this overachievement problem is to value the overachievement as suggested by Oates et al. (1989). In our paper, we follow this approach. The extra emission reductions increase environmental quality and, therefore, social welfare. Firms, however, do not value the overachievement. We would like to thank especially C.M. Billiet for the expert help on estimating the different costs. We have chosen an average gross wage level in the textile industry of 37 e/hour. The average gross wage in the civil sector is assumed to be 50 e/hour. The costs of analysing a sample in the lab are assumed to be e 372 on average. The costs of printing the regulatory information are assumed to be approximately e 12,395. We assume that the instruments are combined with a civil fine and that the willingness to pay for water quality improvement is zero. This assumes that the revenue from the emission tax does not flow back to the environmental agency. We assume that firms for which we had no cost estimate for the heat exchange technology and which were therefore not allowed to use this technology under an emission tax or emission standard, could install this technology under a technology standard but only at the highest available cost estimate. Since the heat exchange technology is a relatively cheap instrument compared to the other abatement technologies, many firms now have a cheaper abatement option available than under the emission tax or emission standard. We choose not to make all technologies available to all firms for the other policy instruments since the heterogeneity of the abatement cost functions is a crucial factor in our analysis. Allowing all firms to use all technologies would severely compromise this heterogeneity. The Superfund program was created in 1980 when congress enacted the Comprehensive Environmental Response, Compensation and Liability Act (CERCLA) to achieve two primary objectives: to identify and clean up sites contaminated with hazardous substances throughout the United States and to assign the costs of cleanup directly to the responsible parties. (Probst et al. 1995) Since the three different sanctioning instruments only differ with respect to the costs associated with them, we show the results for only one instrument.

References Aidt, T. S. (1998), ‘Political Internalization of Economic Externalities and Environmental Policy’, Journal of Public Economics 69, 1–16. Ando, A. W. (1998), Do Interest Groups Compete? RFF Discussion Paper 98–14.

364


Beccaria, C. (1770), ÔOn Crimes and Punishments, and Other WritingsÕ, in R. Bellamy and R. Davies et al. (eds.), trans., New York: Cambridge University Press. Becker, G. S. (1968), ‘Crime and Punishment: An Economic Approach’, Journal of Political Economy 76(2), 169–217. Bentham, J. (1789), ÔAn introduction to the principles of morals and legislationÕ, in The Utilitarians, Rept. Garden City, New York: Anchor Books. Billiet, C. M., S. Rousseau and S. Proost (2002), Law & Economics and the Choice of Environmental Policy Instruments. Final report HL/DD/14. http://www.belspo.be/belspo/ fedra/proj.asp?|=nl&COD=HL/DD/14. Boyer, M. and J. J. Laffont (1999), ‘Toward a Political Theory of the Emergence of Environmental Incentive Regulation’, RAND Journal of Economics 30(1), 137–157. Centexbel (1996), Haalbaarheid afvalwaternormen in de Vlaamse textielnijverheid. Report. Cohen, M. A. (2000), ÔMonitoring and Enforcement of Environmental PolicyÕ, in T. Tietenberg and H. Folmer (eds.), ÔInternational Year Book of Environmental and Resource EconomicsÕ, Edward Elgar Publishers. Damania, R. (1999), ‘Political Competition, Rent Seeking and the Choice of Environmental Policy Instruments’, Environmental and Resource Economics 13, 415–433. Dion, C., P. Lanoie and B. Laplante (1998), ‘Monitoring of Pollution Regulation: Do Local Conditions Matter?’ Journal of Regulatory Economics 13 (1), 5–18. Dixon, L. S. (1995), ÔThe Transaction Costs Generated by SuperfundÕs Liability ApproachÕ, in R. Revesz and R. B. Steward (eds.), Analyzing Superfund: Economics, Science and Law (pp. 171–185). Washington, DC: Resources for the Future. Goulder, L. H., I. Parry, R. C. Williams III and D. Burtraw (1999), ‘The Cost-Effectiveness of Alternative Instruments for Environmental Protection in a Second-Best Setting’, Journal of Public Economics 72, 329–360. Harford, J. D. (1978), ‘Firm Behavior under Imperfectly Enforceable Pollution Standards and Taxes’, Journal of Environmental Economics and Management 5, 26–43. Harrington, W. (1988), ‘Enforcement Leverage When Penalties are Restricted’, Journal of Public Economics 37, 39–53. Heyes, A. (2001), The Law and Economics of the Environment. Edward Elgar. Hopkins, T. D. (1996), Regulatory Costs in Profile. Center for the Study of American Business, Policy Study no. 132. Jacobs A., L. Bettens, A. De Grijse and R. Dijkmans (1998), Best beschikbare technieken voor de textielveredeling. VITO. Academia Press. Keeler, A. G. (1995), ‘Regulatory Objectives and Enforcement Behaviour’, Environmental and Resource Economics 6, 73–85. Kolstad, C. D. (2000), Environmental Economics. Oxford University Press. Malik, A. S. (1992), ‘Enforcement Costs and the Choice of Policy Instruments for Controlling Pollution’, Economic Inquiry 30, 714–721. Ministry of the Flemish Community (11 June 2001), Milieuhandhavingsrapport 2000 en het 10-jarig bestaan van de Vlaamse Milieu-inspectie. Press Release. Montesquieu, C. (1748), The Spirit of the Laws. Rept. edition. Berkeley: University of California Press. Oates, W. E., P. R. Portney and A. M. McGartland (1989), ‘The Net Benefits of Incentivebased Regulation: A Case Study of Environmental Standard Setting’, American Economic Review 79(5), 1233–1242. OVAM (1996), Sectorstudie bedrijfsafvalstoffen: loontextielveredeling. Report. Polinsky, A. M. and S. Shavell (2000), ‘The Economic Theory of Public Law Enforcement’, Journal of Economic Literature 37(March), 45–76.


365

Posner, R.A. (1974), ‘Theories of Economic Regulation’, Bell Journal of Economics and Management Science 4, 335–359. PRESTI project (1994–1997), Handleiding voor de modelaanpak van preventieve milieuzorg – Textielsector. Report. Probst, K. N., D. Fullerton, R. E. Litan and P. R. Portney (1995), Footing the Bill for Superfund Cleanups. Who Pays and How? Washington, DC: The Brookings Institute and Resources for the Future. Rousseau S. (2005), Selecting environmental policy instruments in the presence of incomplete compliance. Doctoral dissertation. Catholic University of Leuven. Rousseau S. and C. M. Billiet (2003), Using Emission Standards under Incomplete Compliance. ETE Discussion Paper 2003-03. Rousseau S. and S. Proost (2004), The Relative Cost Efficiency of Market-based Environmental Policy Instruments with Imperfect Compliance. ETE Discussion Paper 2004-15. Sandmo, A. (2002), ‘Efficient Environmental Policy with Imperfect Compliance’, Environmental and Resource Economics 23(1), 85–103. Slemrod, J. and N. Sorum (1984), ‘The Compliance Cost of the US Individual Tax System’, National Tax Journal 37(4), 461–474. Stigler, G. J. (1970), ‘The Optimum Enforcement of Laws’, Journal of Political Economy 78, 526–536.