Rethinking the optimal level of environmental quality: justifications for strict environmental...

Ecological Economics 36 (2001) 461–473

ANALYSIS

Rethinking the optimal level of environmental quality:justifications for strict environmental policy

Amy Farmer a, James R. Kahn b, Judith A. McDonald c,*, Robert O’Neill d

a Department of Economics, Uni6ersity of Arkansas, AR, USAb En6ironmental Studies Program and Department of Economics, Washington and Lee Uni6ersity, USA

c Department of Economics, Lehigh Uni6ersity, 621 Taylor Street, Bethlehem, PA 18015-3117, USAd En6ironmental Sciences Di6ision, Oak Ridge National Laboratory, USA

Received 25 October 1999; received in revised form 24 August 2000; accepted 5 September 2000

Abstract

Traditional environmental theory suggests that the optimal level of a pollution emission occurs when the marginaldamage created by the emissions is equal to the marginal cost of reducing the emissions. We argue that the benefitsfrom reducing pollution should be much more broadly defined to include at least three other sources of benefits. First,we develop a game-theoretic model in which firms may under-invest in cost-saving ‘green technologies’. Second, wedemonstrate that consideration of future damages and abatement costs leads to a lower current optimal pollutionlevel than that obtained in traditional models. Finally, we show that ecological complexity creates indirect pathwaysby which greater pollution increases the likelihood of generating irreversible environmental damage. This broaderdefinition of the benefits of pollution abatement yields an optimal level of pollution that may actually be less than thelevel at which conventionally-measured marginal damages are equal to marginal abatement costs. Thus, environmen-tal policy should be stricter. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Environmental policy; Game-theoretic model; Ecological complexity; Environmental benefits; Environmental damages;Ecological services

www.elsevier.com/locate/ecolecon

1. Introduction

Excessive levels of environmental degradationoccur when those making decisions about usingresources do not consider social costs. This omis-

sion generates a market failure and creates a needfor policy intervention to correct the market fail-ure. However, before policy-makers intervene toattempt to correct this externality, they must havea policy goal, which, in the case of pollution,should be the optimal level of pollution. In deter-mining policy addressing environmental degrada-tion such as pollution or land-use changes, it isessential to value the damage accurately and then

* Corresponding author. Tel.: +1-610-7585345; fax: +1-610-7584677.

E-mail address: [email protected] (J.A. McDonald).

0921-8009/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.

PII: S 0921 -8009 (00 )00244 -5

A. Farmer et al. / Ecological Economics 36 (2001) 461–473462

to compare it with the costs of preventing thedegradation. Measuring the damage carefully is acritical and very difficult part of this processbecause estimating economic value entails manypractical problems.1

Traditional environmental economic theorysuggests that the optimal level of a pollutionemission occurs when the marginal damage cre-ated by the emissions is equal to the marginal costof reducing the emissions. Lately, this result hasbeen challenged by those who postulate that theoptimal level of pollution may actually be lessthan the level at which conventionally measuredmarginal damages are equal to marginal abate-ment costs. This result is not due to an abandon-ment of optimization and marginal analysis, butto a broader definition of the benefits of pollutionabatement. In general, advocates of ‘strict envi-ronmental policy’ believe that the benefits fromreducing the level of emissions are greater thanthe economic value of the direct physical damagesthat are prevented by lowering emissions.

This paper focuses on this broader definition ofthe benefits of pollution abatement and presentstheoretical arguments to determine whether thereis reason to rethink what constitutes the optimallevel of pollution. After more explicitly discussingthe traditional view of the optimal level of pollu-tion (and associated policy making) in Section 2,three categories of potential benefits associatedwith stricter environmental policy are defined. InSection 3, we develop a model in which firms mayunder-invest in ‘green technologies’ due to theleader firm’s inability to prevent other firms fromimitating its innovation if successful, and notfollowing (and thus having lower costs and poten-tially useful information about the correct way toinnovate successfully) if the leader firm’s innova-tion fails. In this case, government intervention inthe form of stricter environmental regulationwould lower production costs in a fashion similarto what has been suggested in the ‘Porter Hypoth-esis’ (Porter, 1990, 1991). Section 4 examines theoptimal level of current pollution in a dynamic

context. Consideration of future damages andabatement costs also leads to a lower currentoptimal pollution level than that obtained in tra-ditional models. Section 5 looks at the importanceof ecosystem complexity and nonlinearities whichcan lead to thresholds and irreversibilities, intensi-fying the potential damages associated with agiven level of pollution emissions.

These three factors broaden our perceptions ofthe damages arising from pollution which leads usto argue that the optimal level of pollution islower than conventionally determined and thatenvironmental policy should be stricter than iscurrently the case.

2. The traditional view of the optimal level ofenvironmental quality

The optimal level of environmental policy isusually discussed in terms of the intersection ofthe marginal damage function and the marginal

Fig. 1. Schematic of marginal damage function.

1 See Georgiou et al. (1997), Bjornstad and Kahn (1996),Barbier (1998), World Bank (1992) and Freeman (1993)formore discussion on these practical problems.

A. Farmer et al. / Ecological Economics 36 (2001) 461–473 463

Fig. 2. The optimal level of pollution.

ally lower their production costs. The argument isthat stricter environmental standards force firmsto be more efficient in converting inputs to eco-nomic outputs as they seek methods to reducetheir production of waste outputs. Porter arguesthat this reduction in cost will increase the inter-national competitiveness of US firms and havepositive impacts on balance of payments, employ-ment, and other macroeconomic variables. Ofcourse, many economists question the underlyingpremise of these arguments; if cost savings oppor-tunities exist, why don’t firms take advantage ofthem without government intervention?

Palmer and Simpson (1993) and Oates et al.,(1993) examine this question and critically evalu-ate three possible explanations of why these un-seized opportunities exist. These are: (1) firmstupidity; (2) firm short-sightedness and the fixed-cost nature of abatement activities; and (3) thepublic good properties of research anddevelopment.

Although corporation-bashing is a popular ac-tivity and there are certainly a number of firmsthat make apparently stupid decisions, firm stu-pidity is not a particularly satisfying explanationfor the existence of these unseized opportunities.While a number of ‘stupid’ firms may exist at anypoint in time, Darwinian market forces shouldfavor the existence of smart firms over stupidfirms in the long run.

The second argument, which is related to thefirst, is that firms are short-sighted and will notincur short-term costs in order to achieve long-term gain. This point is articulated by Simpson(1993), who argues that the way green technolo-gies reduce costs is that the firm incurs initialincreases in fixed cost (e.g., more energy efficientand less polluting capital) that result in reductionsin variable cost over the lifetime of the capital(e.g., less energy consumption, less waste gener-ated, and less waste disposal cost). While thisargument has more appeal than the first, andthere is certainly much anecdotal evidence sug-gesting that many firms are myopic, Simpsonnotes that Darwinian market forces should favorthe survival of far-sighted firms over short-sightedfirms in the long run.

abatement cost function. The marginal damagefunction shows pollution as a function of emis-sions of a particular pollutant, and is actuallycomposed of a chain of functional relationships,as depicted in Fig. 1. The marginal abatementcost function shows the cost of reducing emissionsbelow the level that would take place in an unreg-ulated market economy (Eu in Fig. 2). The opti-mal level of pollution occurs at E* and minimizesthe total social damages from emissions (the areaOAEu of Fig. 2).

There is no question that E* is the optimal levelof emissions of pollution because it minimizes thesum of total abatement costs (area AEuE*) andtotal damages (area OAE*). Our argument doesnot conflict with this important principle. Rather,our argument is that as traditionally conceptual-ized (and sometimes measured), the marginaldamage function represented by the flow diagramin Fig. 1 and illustrated in Fig. 2 excludes impor-tant damages from emissions, or conversely, doesnot include important benefits arising from reduc-ing the level of pollution.

3. The effect of strict environmental policy onproduction costs

In a widely-cited book and article, MichaelPorter (1990, 1991) argues that forcing firms tomeet stricter environmental standards will actu-


The third explanation is that research and de-velopment into cost-saving green technologies haspublic good benefits since innovations are easy tocopy. Because firms only respond to privatebenefits, there is insufficient expenditure on re-search and development of green technologies.Porter goes further to argue that forcing the greentechnology will not only lower cost, but lowercost relative to foreign competitors. The secondpart of this argument follows from the first only ifit is assumed that innovations that are easilycopied by domestic rivals are not as easily copiedby foreign rivals.2 Thus, unless inter-countrycopying is difficult, domestic firms would not belikely to develop and maintain a competitive edgeover foreign rivals.

Although these three reasons provide insuffi-cient support for the Porter Hypothesis, we ad-vance an argument that may explain the potentialexistence of unseized cost-saving opportunities.This discussion, based on the strategic behavior offirms, follows below.

3.1. Insufficient in6estment in green technologiesas a game-theory problem

This part of our paper builds on the idea thatthere may be unseized cost-reducing opportunitiesby illustrating a situation in which individualfirms will choose not to adopt a potentially cost-saving green technology that would raise the ex-pected profits of all firms in the industry. Thissituation is generated by the existence of asym-metries in the realization of the potential gainsfrom innovation that are generated by strategic

behavior. Under such circumstances, we demon-strate a case in which intervention may be re-quired to encourage risk-neutral, profit-maximizing firms to adopt a technology that willraise expected profits.

Some researchers (see, e.g., Simpson 1993, p.26) claim that ‘strong’ environmental policy isinferior to both direct production subsidies andR&D subsidies in terms of advancing industrialpolicy objectives by inducing innovation expendi-tures. In the discussion that follows it is theexistence of strategic behavior rather than marketfailure (such as some flaw in the capital marketthat would better be fixed by governmental subsi-dies to R&D) that causes a sub-optimal outcome.In this case, the traditional solutions to marketfailure, such as ‘fixing’ capital markets or R&Dsubsidies, would not change the strategic behaviorunless they would address the asymmetry thatgives rise to the strategic behavior.3 Therefore,strict en6ironmental policy may be the first-bestpolicy because it impacts directly on strategicbeha6ior.

The intuition driving the possibility that suchan inefficient outcome will occur without inter-vention is as follows. Suppose that a particularabatement technology exists, but the potentialimpact of its adoption on production costs isunknown ex ante. If one firm takes the lead inadoption, the remaining firms make their decisionwith superior information. Whichever firm makesthe initial investment bears the sole cost of gather-ing the information. If the new technology lowersproduction costs, the remaining firms in the in-dustry will follow the original firm’s adoptiondecision and the benefits of the cost savings areshared. However, if the technology has no impactor possibly even a negative impact on productioncosts, then no other firms will choose to followand adopt the technology; thus, the leading firm

2 There are a variety of differences in culture, economicinstitutions, social institutions, human capital and technologythat can impact the process of diffusion and adoption oftechnology. Therefore, the speed at which an innovation thatoccurs in one country is adopted by another country will varyby technology and country. For example, many Pacific rimcountries are noted for the rapid introduction of new technolo-gies and, in fact, some of these countries do not enforcecopyrights and intellectual property rights that originate inother countries. Thus, one cannot make a general propositionthat innovations are or are not easily copied across countries.However, if only a subset of countries copy the innovation, thecompetitive advantage will be lost.

3 R&D subsidies would shift the distribution of costs andbenefits faced by individual firms, but they would not changethe asymmetry of costs and benefits between firms that ismodeled below and that leads to the strategic behavior. How-ever, if R&D subsidies took the form of rebates or rewardssolely available to the firm that develops and implements theinnovation, a direct subsidy could accomplish the same result.


bears all of the costs of acquiring the information.This asymmetry implies that it may be possiblefor the expected benefits of the technology toexceed the expected costs for every firm in theindustry if they were all to simultaneously adopt,but still no firm has the incentive to be the first. Inshort, it pays to be a follower rather than a leaderand the strategic role of timing can result in aPareto inferior solution for the industry.

It is straightforward to illustrate a Cournotduopoly example in which both firms’ expectedprofits rise if they are required to adopt thetechnology, but, independently, neither will makethe choice to adopt. The illustration presentedbelow is not intended to show that firms willnever take the lead and adopt new potentiallycost-saving technologies. Without explicit data, itis impossible to make general statements concern-ing the prevalence of this problem across indus-tries. It is our goal to demonstrate through theuse of an example the potential need for interven-tion to achieve optimality. Specifically, what thisexample does show, however, is that situations doexist in which a technology that will raise firms’expected profits is not adopted because of thistype of strategic behavior. Since standard eco-nomic intuition suggests that regulation is notneeded to induce firms to undertake ventures thatare profitable, we use a set of specific (but reason-able) assumptions to develop a counter-exampleillustrating that this intuition is not necessarilycorrect and should be re-evaluated. Of course,further research is needed to determine particularindustries to which these difficulties might apply.Through our counter-example, we are simply sug-gesting that strategic behavior might imply theneed for stricter environmental regulation andthat this issue should be further investigated.

Consider a simple example in which two identi-cal, risk-neutral, Cournot competing firms have aconstant marginal cost of production equal to k.They face a market demand for their product ofP=A−BQ, where Q is their total output. Atechnology exists and costs C to adopt and, ifadopted, this technology results in a new marginalcost equal to k %, which may be larger or smallerthan k. Although k % is unknown, its distribution iscommon knowledge. If the distribution were un-

known, it would further strengthen our argumentsas the additional uncertainty would create greaterincentive to be a follower.

It is obvious that each firm prefers to follow.But, if a firm’s opponent refuses to lead, thedecision must be whether to lead or accept thestatus quo. In order to reach a decision, the firmmust consider the potential equilibrium outcomesand profits that may result in either case, in orderto reach a decision. Consider first the decision toaccept the status quo.

3.1.1. Firm 1 does not adoptIf neither firm takes action, then each simply

chooses output to maximize profit of

p1=P(Q)q1−kq1 (1)

where Q=q1+q2. After finding standardCournot reaction functions and solving for theequilibrium, this strategy yields each firm the fol-lowing profit level:

p1=p2=(A−k)2

9B(2)

3.1.2. Firm 1 chooses to leadIf instead firm 1 chooses to be a leader and

adopt the technology at a cost of C, the resultingmarginal cost is k % (which may be higher or lowerthan k). The important question is whether or notfirm 2 will follow. Prior to computing firm 1’sexpected profit from leading, it is necessary toisolate firm 2’s reaction strategy and the equi-librium profits that will result in either case.

In the first case, when firm 1 chooses to lead,firm 2 follows. In this case each firm spends thefixed adoption cost and then chooses the optimalCournot output given its new marginal cost k %.Final equilibrium profit for each firm is then

p1=p2=(A−k %)2

9B−C. (3)

In the second case, when firm 1 chooses to lead,firm 2 chooses not to follow. In this case, eachfirm will have different marginal costs. As a con-sequence, equilibrium profits resulting fromCournot competition are no longer identical.Note that each firm’s profits depend upon the


marginal cost of both firms because each firm isreacting to the other’s strategy, and equilibriumoutput is simultaneously determined. Profits arefound in Eqs. (4) and (5) below:

p1=(A−2k %+k)2

9B−C (4)

p2=(A+k %−2k)2

9B. (5)

Comparison of Eq. (3) and (5) reveal that firm2 will follow firm 1 and adopt the technology iff

k−k %]9BC

4(A−k). (6)

Note that if adoption is costless (C=0) then firm2 chooses to adopt as long as the new technologyreduces costs (k\k %). Otherwise, if adoption costsare positive, then the cost savings must be suffi-ciently large to induce adoption. Note also that thelarger the demand, the smaller this per unit savingsmust be to encourage adoption by firm 2.

3.1.3. Firm 1’s initial decisionGiven that firm 2 will adopt only if Eq. (6) is

satisfied, firm 1 must decide whether or not tochoose to be a leader. This decision depends uponthe probability that Eq. (6) will ultimately hold, aswell as on the profits firm 1 can expect in eachinstance. Firm 1 faces three possibilities. The firstpossibility is to maintain the status quo and receiveprofits found in Eq. (2), while the second is to adoptthe technology and firm 2 will follow. Profits arefound in Eq. (3) where k % is small enough to haveencouraged following. The third possibility for firm1 is to adopt the technology, firm 2 does not follow,and profits are found in Eq. (4).

Let us consider this problem first for somegeneral distribution of the final marginal costs.Define o to be k %−k ; note that this will be negativeif the new technology reduces costs. Then if oB−9BC/4(A−k), adoption will be followed by firm 2.Given some known distribution of o, firm 2 willfollow with probability F(−9BC/4(A−k)). Firm1’s expected profit from adoption is then theexpected profit in each case conditional on thevalues of o that would induce each case, weightedby the probability that o falls in that range. Firm

1 will take the lead if this expected profit exceedsthe status quo. More specifically, firm 1 will adoptthe technology iff

F� −9BC

4(A−k)n

E�(A−k %)2

9B−C �oB −9BC

4(A−k)n

(7)

+�

1−F� −9BC

4(A−k)nn

E�(A−2k %+k)2

9B−C �oB −9BC

4(A−k)n

](A−k)2

9B.

Note that the first term is the expected profit tofirm 1 when firm 2 follows (Eq. (3)) conditional ono being small enough to induce firm 2 to follow,weighted by the probability that o fits that con-straint. The second term is a similar expressionwhen following does not occur (Eq. (4)) and theright-hand side is the equilibrium expected profitsif the status quo is maintained (Eq. (2)).

3.1.4. Efficient simultaneous adoptionEq. (7) identifies the necessary condition for

either firm to be willing to assume a leadership roleand adopt the technology. However, if both firmscould simply agree to simultaneously adopt thetechnology prior to any knowledge of its costeffects, then they would both benefit from doing soif expected Cournot profits given the distributionof k % exceed profits given in Eq. (2). This occurs iff

E�(A−k %)2

9B−C

n\E

�(A−k)2

9Bn

. (8)

After some algebraic manipulation, this conditionsimplifies to

2(k−A)E(o)+E(o2)\9BC. (9)

If Eq. (9) holds, then the firms would expect tobe better off if they both adopted the technology.However, without the possibility for binding con-tracts between both firms, this is not enforceable.If Eq. (7) does not hold then each firm has theindividual incentive to wait and hope the other willlead. Even if the other will not lead, if Eq. (7) failsthen each firm will prefer the status quo overadoption. Thus, it is Eq. (7) rather than (9) that is


text. It will be shown that current decisions aboutthe target level of pollution can have beneficialeffects on future periods and actually influence theoptimal level of pollution. Two types of dynamicthe relevant condition from the standpoint of theindividual firm.

The obvious question then becomes whetherthere are situations in which Eq. (9) holds (i.e., bothfirms would be better off if they were somehowforced to simultaneously adopt this technology),but Eq. (7) does not hold (i.e., neither firm willmake that choice independent of regulation).

To answer this question, we use a very simpledistribution for o to illustrate a case in which thiswill occur. Suppose k % is either k+1 or k−1 eachwhich occurs with probability 0.5. In other words,o is either +1 or −1 with probability 0.5. Tosimplify the example further, let us assume thatadoption costs are zero. Since C=0 and E(o)=0,Eq. (9) implies that adoption would be optimal ifE(o2)\0. In this example, E(o2)=1, thus adoptionis optimal for every value of the parameters.

However, we can illustrate that neither firm hasthe incentive to take the lead independently. FromEq. (6) we know that firm 2 will choose to followif it turns out that k %Bk (i.e, if oB0). Substitutingk %=k+o into Eq. (7), we know that firm 1 will takethe lead and adopt the technology if

F(0)E�(A−k−o)2

9B�oB0

n(10)

+ (1−F(0))E�(A−k−2o)2

9B�oB0

n\

(A−k−o)2

9B.

Note first that for this specific distributionF(0)=0.5 and 1−F(0)=0.5. Computing the ex-pected profits conditional on o (whether or not firm2 follows) is straightforward in this simple example.In the case that firm 2 chooses not to follow theno \0; the distribution on o implies that o =1 andtherefore firm 1’s expected equilibrium profits willultimately be

E�(A−k−2o)2

9B�oB0

n=u (11)

(A−k)2−4(A−k)E(o)+4E(o2)9B

=(A−k−2)2

9B.

If instead firm 2 chooses to follow (oB0) then it

must be the case that o= −1 and firm 1 can expectequilibrium profits of

E�(A−k−o)2

9B�oB0

n= (12)

(A−k)2−2(A−k)E(o)+E(o2)9B

=(A−k+1)2

9B.

Substituting both the probabilities and the con-ditional profits into Eq. (7), we find that firm 1 willadopt the new technology if A−kB5/2. In otherwords, if demand is sufficiently large relative tocosts, firm 1 will not choose to lead and adopt thetechnology first. Of course, since firms are identical,if firm 1 is willing to accept the status quo ratherthan adopt the technology, so too is firm 2. Recall,however, that for every value of the parameters,simultaneous adoption raises both firms’ expectedprofits.

Although this basic example does not prove ageneral result, it does illustrate a simple point.Circumstances can exist in which the welfare ofeach and every firm will be improved if they areforced to implement a technology that they wouldnot otherwise choose to implement. Individualstrategic behavior can generate a Pareto inferiorequilibrium in which some form of binding contractor outside intervention is needed to assist firms inmoving to a superior equilibrium. Although theidea that firms may need external coordination tohelp them achieve the Pareto efficient solution isnot new, the implications of this for environmentalpolicy are new and important: under certain cir-cumstances firms would not adopt cost-savinggreen technologies without government interven-tion. Thus, stricter environmental policy could raisesocial welfare, not only through improving envi-ronmental quality, but also through decreasing thecosts of production.

4. Dynamic considerations and the currentbenefits of emissions abatement

Further insight into whether additional benefitsexist from abatement can be gained by looking atthe pollution control problem in a dynamic con-


interaction are examined. First, the impact ofcurrent target levels of pollution on technologicalinnovation and future abatement costs is ana-lyzed. Second, the impact of prior period emis-sions on later period damages is examined.

Before examining the dynamic considerations,it is useful to derive the optimal level of pollutionin the static case. This is shown below in Eq. (13),where the total social costs of pollution (TSC) arethe sum of abatement costs (TAC) and the dam-ages from the remaining emissions (TD). SolvingEq. (13) gives the familiar result that the level ofemissions is chosen so that marginal abatementcosts (MAC) and marginal damages (MD) areequal in each period:

Min TSC=TAC(Q)+TD(Q) wrtQ

dTACdQ

B0,dTDdQ\0. (13)

This model can be made slightly more sophisti-cated by allowing the preference for environmen-tal quality to be increasing over time, as in Eq.(14),

Min TSC=TAC(Q)+TD(Q,t) wrtQ

dTACdQ

B0,(TD(Q\0,

(TD(t\0. (14)

Although the optimal level of pollution will bedeclining over time, optimality still requires MACto equal MD and the model is still basically astatic model as each period’s decision is indepen-dent of every other period.

Similarly, adding exogenously-determined tech-nological innovation (which is represented bymaking TAC a decreasing function of time in Eq.(15)) creates an optimal pollution path that isdeclining over time. However, as in the abovecase, each period’s decision is independent of thedecision in every other time period.

Min&�

t=0

TSC dt

=&�

t=0

[TAC(Q,t)+TD(Q,t)] dt wrtQ (15)

(TAC(Q

B0,(TD(Q\0,

(TD(t\0,

(TAC(tB0.

However, if technological innovation is notonly related to the passage of time, but also tocurrent incentives to innovate, a very differentpicture emerges. In Eq. (16), an optimizationproblem is considered in which the costs of abate-ment are a function of the state of technology,(TECH), and the current rate of technologicalinnovation (d) is a function of the state of tech-nology and current incentives for technologicalinnovation (current total abatement costs):

Min&�

t=0

TSC dt=&�

t=0

[TAC(Q,TECH(t),t)

+TD(Q,t)] dt (16)

(TECH(t

=dt(TECH, TAC, t)

(d

(TECH\B

0,(d

(TAC\0

Eq. (17) contains the Hamiltonian for this opti-mization problem and the optimality conditionand co-state equation are contained in Eq. (18).Eq. (19) contains the solution to the optimalitycondition and shows that when technological in-novation is endogenous, optimality is not deter-mined by an emission level where currentmarginal damages equal current marginal abate-ment costs, but one where current marginal abate-ment costs are greater than current marginaldamages. If marginal abatement costs are a de-clining function of emissions and marginal dam-ages are an increasing function of emissions, theoptimal level of emissions in a given period will beless than the level at which that period’s MACequals that period’s MD.

H= (TAC(Q, TECH)+TD(Q) e−rt)

+ldt(TECH, TAC(Q, TECH)) (17)

(H(Q

=�(TAC(Q

+(TD(Q

�e−rt+l

(d

(TAC(TAC(Q

=0

(18)

−(H(TECH

=dl

dt= −

� (TAC(TECH

e−rtn−l

� (d

(TECH+(d

(TAC(TAC(TECH

n(TAC(Q

�1+lert (d

(TAC�

= −(TD(Q

(19)


This is an extremely important result because itillustrates that even if emissions at time t onlycause damages at time t, the determination of theoptimal level of pollution at any point in time is adynamic problem, which must take into accountthe impact of the current level of regulation onfuture abatement costs. In other words, today’sstrictness improves future welfare by generatingreduced costs of abatement in the future. Thesereduced costs mean that the future can obtainbetter environmental quality at a lower abatementcost. Strict environmental policy cannot only beviewed as a potentially more efficient policy, but itcan also be viewed as a form of investment,allowing future periods to obtain higher levels ofboth environmental quality and GDP. This doesnot mean we should implement the most strictpolicies possible, as cost savings in the futuremust be balanced against the costs imposed at thepresent time.

Previously, dynamic considerations in the deter-mination of optimal emissions levels were onlythought to occur when the emissions from oneperiod accumulated into future periods andcaused damages in future periods (such as emis-sions of DDT, heavy metal, CFCs, or CO2). Thisintertemporal dependency is well discussed in theliterature4, but may be even more important thansuggested because this type of dynamic consider-ation would also occur even if the emissions donot accumulate, but if ecological stresses do. Forexample, emissions of nitrogen oxides, sulfur ox-ides, and volatile organic compounds have rela-tively short residence periods in the atmosphere,as does tropospheric ozone, the by-product oftheir interaction. However, the stress that thisperiod’s ozone generates for ridge-top forests maymake them more susceptible to damage fromozone in future periods5.

5. The importance of ecological complexity

Ecological services are an important source of

social benefits that are not often explicitly consid-ered when measuring the damages associated withenvironmental degradation. Ecological services in-clude outputs of ecosystems, such as nutrientcycling, waste assimilation, maintenance of localand global climate, carbon sequestration, biodi-versity, watershed protection and soil formation.Ecological services are important at many levels.At the most basic level, they provide life supportservices for humans and other species. They alsoprovide basic inputs to economic processes ascrops grow better in a clean air environment, andbeer and other products are cheaper to produce ifyou start with clean water. At a more complexlevel, the economic and social systems are con-tained within the larger environmental system.The more productive the ecosystem in terms ofthe production of ecological services, the greaterthe prospects for economic productivity and thequality of life in general [see Kahn and O’Neill(1999) for further discussion].

Another justification of the importance of eco-logical services (in an area where they have beentraditionally excluded from economic consider-ation) can be found in a criticism of the econom-ics literature’s discussion of sustainability. Forexample, Hartwick (1977) examines the prospectsfor sustainability when output is produced byhuman-made capital, labor and exhaustible re-sources. He finds that sustainability is feasible ifthe rents from the exhaustible resource are re-in-vested back into human-made capital. In the nar-row context of this model, he is correct. Animportant, but justifiable assumption of thismodel is that human-made capital and labor aregood substitutes for exhaustible resources (such asoil or iron).

However, let us consider the case where, inaddition to labor, human-made capital and ex-haustible resources, productive inputs include en-vironmental resources that provide ecologicalservices. In this case, the Hartwick rule will breakdown (see Franceschi and Kahn, 1999a andFranceschi and Kahn, 1999b) as the depletion ofenvironmental resources cannot be compensatedfor by investing in additional human-made capi-tal, as human-made capital is not a good substi-tute for environmental resources. This assertionmay at first seem absurd, as one might expect that

4 See, for example, Ko et al., (1992), Brito and Intriligator(1987) and Benford (1998).

5 See Kahn and O’Neill (1999) for a discussion of theaccumulation of ecological stresses.


ecological services such as nutrient cycling, wasteprocessing, and soil formation can be provided bythe application of labor and human-made capitalin human-engineered systems. However, at thescale that the ecosystems provide these services, itis completely infeasible to produce these ecologicalservices in human-engineered systems. For exam-ple, Costanza et al. (1997) estimated the ecosystemservices provided by the world’s oceans would costabout US $20 trillion per year to replace withhuman technology.

The argument above suggests that ecologicalservices are important. The value of ecologicalservices has traditionally been excluded from themeasurement of the damages from environmentaldegradation. Indeed, this exclusion itself supportsthe view that the traditional conception of theoptimal level of emissions is too high and that ahigher level of environmental quality would be apotential Pareto improvement. The existence ofnon-linearities, complexity and irreversibilities inecological relationships strengthens thesearguments.6

In economics, all functional relationships tend tobe viewed as marginal. A marginal change invariable X generally leads to a marginal change invariable Y. However, in both ecological and eco-nomic relationships, this is often not true.7 It isparticularly true when marginality is not viewed inthe strict mathematical sense (a change of arbitrar-ily small nature), but in a policy sense wheremarginal would refer to the smallest types ofadjustments that are possible with policy or thatwould take place in the absence of policy. Thisnon-marginality from a policy perspective may takethe form of a nonlinear, but continuous damagefunction, where a small (but discrete) change in onevariable leads to an extremely large change in other

Fig. 3. Nonlinear damage function.

variable(s). For example, initially, marginalchanges in carbon dioxide emissions will lead tomarginal changes in atmospheric concentrations ofcarbon dioxide, which will lead to marginal changesin global temperature and marginal changes insocial welfare. However, global temperature willeventually rise to the point where positive feed-backs occur (such as the melting of the tundra andthe massive release of methane, a powerful green-house gas), causing an accelerating increase in bothconcentrations of greenhouse gases and globaltemperature. Thus, thresholds may be crossed(O’Neill et al., 1989) leading to marginal damagefunctions such as the one in Fig. 3, where emissionsin excess of E1 are associated with very largedamages. The threshold may also take the form ofa discontinuity as shown in Fig. 4.

The types of damage functions presented inFigs. 3 and 4 become important in the context of

Fig. 4. Discontinuous damage function.

6 The argument is further strengthened by the existence ofcomplex relationships among pollutants, which often lead tonon-linear relationships between emissions (such as ozoneprecursors) and ambient concentrations of pollution (such astropospheric ozone).

7 An important exception to this generalization is the workthat has been done in economic and chaos theory. See Rosser(1991) for a discussion of the economic implications of chaostheory.


irreversibility. A conventional perspective of eco-nomics is that most activities are reversible. Whilethe existence of irreversibilities has long been rec-ognized in environmental economics8, irreversibil-ities have generally been viewed as the directconsequence of economic activity. For example,Krutilla and Fisher (1975) use examples involvingthe conversion of unique natural environments,such as daming a wild river or mining in awilderness area. Other examples of irreversibilitiesdiscussed in the environmental economics litera-ture involve the emissions of long-lived pollutantsthat accumulate in the environment (heavymetals, dioxins, DDT, carbon dioxide, PCBs,CFCs, etc.) and the extinction of species throughover-harvesting.

Kahn and O’Neill (1999) introduce the conceptof indirect irreversibilities, whereby economic ac-tivities affect the complex interactions of speciesand ecological communities and generate newequilibria which are substantially different thanthe existing equilibria. As these interactions areimpacted by economic activity, irreversible changeis generated as other species or other ecologicalcommunities have gained a competitive advan-tage, preventing the return of the initialequilibrium.

A good example of this can be found in theridge-top forests of the Southern AppalachianMountains. Ridge-top spruce and fir forests arestressed by the synergistic impacts of acid rain,tropospheric ozone, and non-native insects. Thesestresses are weakening the spruce and fir andallowing other species to become established. Ifsome time in the future the stresses are removed,the spruce and fir forests will not be able tore-establish themselves as the other species be-come established and maintain a competitive ad-vantage. This process is illustrated in Fig. 5,where the horizontal axis measures the level ofenvironmental degradation or stress and the verti-cal axis measures an ecological characteristicwhich is an indicator of the health of that particu-lar ecosystem. This ecological characteristic will

Fig. 5. Ecological instability and environmental change.

fluctuate with changes in environmental charac-teristics (such as climate), but as the environmen-tal degradation increases, the character of thefluctuations changes. The amplitude increases andthere may be a gradual downward trend. Eventu-ally, a bifurcation occurs, and the system will flopto a new equilibrium. Even if the environmentaldegradation is reduced (as shown by the fluctuat-ing dotted curve), the system will not restore theoriginal equilibrium.9

The above discussion suggests that the targetlevel of environmental quality should be moreambitious than the level that is traditionally ad-vanced in both the economics literature and in thepolicy process. First, ecological services have notbeen included in the quantitative measures of thebenefits of environmental improvement. Second,ecological complexities and non-linearities implyan additional source of irreversibility that has notbeen articulated in the economics literature. Kru-tilla and Fisher (1975) and Arrow and Fisher(1974) argue very persuasively that the presence ofirreversibility and uncertainty generates the needfor more caution when considering the target levelof environmental quality or environmental preser-vation. The existence of this additional source ofirreversibility (indirect irreversibility) will implythe need for even greater caution.

8 See, for example, Krutilla and Fisher (1975) and Arrowand Fisher (1974).

9 See O’Neill (1999) and Kahn and O’Neill (1999) for fur-ther discussion of these issues.


6. Conclusions

The typical approach to defining the optimallevel of pollution has been to equate the marginaldamages of an additional unit of pollution withthe marginal costs of reducing the pollution.Damages are typically measured as impacts tohuman health, recreational activity, materials andother physical impacts of pollution. Often at-tempts are made to measure existence values andother indirect use values (also called non-use andpassive use) through contingent valuation andother techniques.

We argue that the damages from pollution (orthe benefits from reducing pollution) should bemuch more broadly defined to include at leastthree other sources of benefits. First, we show thatthere may be obstacles associated with strategicbehavior that reduce the incentives for firms toadopt cleaner cost-saving technology. A morestringent environmental policy would reduce theincentives that underlie this strategic behavior.Second, we demonstrate that the dynamic consid-eration of the effect of environmental strictness ontechnological innovation implies future benefits ofstricter environmental policy today. Other dynam-ical factors (such as the intertemporal accumula-tion of ecological stresses) make the intertemporalconsideration of pollution even more important.Finally, we show that ecological complexity cre-ates indirect pathways by which increasing thelevel of pollution increases damages by increasingthe likelihood of crossing a threshold, generatingirreversible environmental change.10

The implications of this paper are associatedwith the way we define our goals for environmen-tal quality. Our arguments suggest that if we haveset our environmental goals based on a moretraditional measure of damages then social wel-fare would be increased by establishing more am-bitious policy goals. Of course, our paper has notestablished the magnitude by which we shouldbecome more ambitious nor has it addressed theadditional costs that would be incurred fromstricter policy. These costs must be balancedagainst the benefits we have enumerated. Ourpaper has focused on outlining those conceptualarguments that call for more ambitious environ-mental policy. Measures of the size and relativeimportance of the components of a more broadlydefined conceptualization of damages await theimplementation of empirical research in these im-portant areas.

Acknowledgements

The authors would like to thank the reviewersfor their comments and suggestions. Any errorsremain the responsibility of the authors.

References

Arrow, K., Fisher, A., 1974. Environmental preservation, un-certainty, and irreversibility. Q. J. Econ. 88, 312–319.

Barbier, E., 1998. The Economics of Environment and Devel-opment: Selected Essays. Edward Elgar, London, UK.

Benford, F., 1998. On the dynamics of the regulation ofpollution: incentive compatible regulation of a persistentpollutant. J. Environ. Econ. Manage. 36, 1–25.

Bjornstad, D., Kahn, J., 1996. The Contingent Valuation ofEnvironmental Resources: Methodological Issues and Re-search Needs. Edward Elgar, London, UK.

Brito, D., Intriligator, M., 1987. Stock externalities, Pigoviantaxation and dynamic stability. J. Pub. Econ. 33, 59–72.

Costanza, R., d’Arge, R., De Groot, R., Farber, S., Grasso,M., Hannon, B., Limburg, K., Naeem, S., O’Neill, R.,Paruelo, J., Raskin, R., Sutton, P., van den Belt, M., 1997.The value of the world’s ecosystem services and naturalcapital. Nature 387, 253–260.

Franceschi, D., Kahn, J., 1999a. The economic implicationsfor sustainable mining within the paradigm of sustainabledevelopment. Analise Economica. 17 (31), 141–154.

10 A fourth important factor lies in the potential existence ofa double dividend from environmental taxation. The argumentfor a second-benefit (the double dividend) from environmentaltaxation arises because a substitution of environmental taxesfor income taxes will reduce distortions in the labor market,increasing social welfare. Critics of the double dividend arguethat increases in product prices associated with abatementcosts will further distort the labor market by giving people agreater incentive for leisure. Kahn and Farmer (1999) arguethat these results are an artifact of the modeling efforts, andthe potential existence of the double dividend (and its relativemagnitude) are empirical issues. Because the existence of adouble dividend from environmental taxation is a controver-sial issue, we have not made it a central theme of this paper.


Franceschi, D. and Kahn, J., 1999b. Capital accumulationand sustainable development: the role of capital, mineralresources and environmental resources in the sustainabledevelopment of the Brazilian Amazon. Department ofEconomics, University of Tennessee, Knoxville (unpubl.).

Freeman, A., III, 1993. The measurement of environmentaland resource values: theory and methods. Resources forthe Future, Washington, DC.

Georgiou, S., Pearce, D., Whittington, D., Moran, D., 1997.Economic Values and the Environment in the Develop-ing World. Edward Elgar, London, UK.

Hartwick, J., 1977. Intergenerational equity and the invest-ing of rents from exhaustible resources. Am. Econ. Rev.67, 972–974.

Kahn, J., Farmer, A., 1999. The double dividend, second-best worlds, and real-world environmental policy. Eco-logical Economics 30, 433–439.

Kahn, J., O’Neill, R., 1999. Ecological interaction as asource of economic irreversibility. S. Econ. J. 66, 391–402.

Ko, II-Dong, Lapan, H., Sandler, T., 1992. Controllingstock externalities, flexible versus inflexible Pigovian cor-rections. Eur. Econ. Rev. 36, 1263–1276.

Krutilla, J., Fisher, A., 1975. The Economics of NaturalResources: Studies in the Valuation of Commodity and

Amenity Resources. Johns Hopkins University Press,Baltimore.

Oates, W., Palmer, K. and Portney, P., 1993. Environmentalregulation and competitiveness: thinking about the Porterhypothesis. Discussion Paper 94-02. Resources for theFuture, Washington, DC.

O’Neill, R., 1999. Recovery in complex ecosystems. J.Aquat. Ecosyst. Stress Recov. 6, 181–187.

O’Neill, R., Johnson, A., King, A., 1989. A hierarchicalframework for the analysis of scale. Landscape Ecol. 3,193–205.

Palmer, K., Simpson, D., 1993. Environmental policy as in-dustrial policy. Resources 112, 17–21.

Porter, M., 1990. The Competitive Advantage of Nations.Free Press, New York.

Porter, M., 1991. America’s green strategy, Scientific Ameri-can, April, 168.

Rosser, J., 1991. From Catastrophe to Chaos: a GeneralTheory of Economic Discontinuities. Kluwer Academic,Dordrecht/Boston.

Simpson, D., 1993. Taxing variable cost: environmental reg-ulation as industrial policy. Discussion Paper ENR 93–12. Resources for the Future, Washington, DC.

World Bank, 1992. World Development Report 1992: Devel-opment and the Environment. Oxford University Press:Oxford/New York.

.

Rethinking the optimal level of environmental quality: justifications for strict environmental...

Documents

Transcript of Rethinking the optimal level of environmental quality: justifications for strict environmental...