cJ

Public Disclosure Authorized

4
1 I'D9

POLICY INSTRUMENTS
FOR POLLUTION CONTROL IN
DEVELOPING COUNTRIES
Gunnar S. Eskeland
Public Disclosure Authorized

Emmanuel Jimenez

LIC
What are appropriatestrategiesfor protectingthe environmentin developing
countriesthat also seek to promotegrowthand reducepoverty?This article
reviewsthe literatureon cost-effectiveintervention,comparingregulatoryand
Public Disclosure Authorized

fiscalinstrumentsthat canbe attunedto the purposeof [Link]

authorslook at what happenswhen developingcountryingredientsare intro-
duced into the standardpolicyproblem,and show how indirectinstruments
canbe effectivewhen monitoringand enforcementare [Link] discussdis-
tributiveconcernsfor two reasons:the effecton the poor may needparticular
consideration for equityreasons,and the effecton groupswith vestedinterests
can determinewhetherreformsare likelyto stand or fall.

R igorous studiesof pollutioncontrol in developingcountriesdo not

exist,but thereis convincingcasualevidencethat regulationsto protect
the environmentare ineffectiveor unnecessarily
[Link],thereare
no regulationsor they are badlydesignedor enforced(seeBernstein1991).In
Public Disclosure Authorized

addition,economicpoliciesseeminglyunrelatedto the environment-subsidies

for water consumption,pesticides,fertilizers,and energy use-nevertheless
affect it, often for the worse (Schrammand Warford1989).Evidenceon the
damagethat pollutionis doingto humanhealthand productivityis startingto
accumulate(see Thomas 1981and 1985on Sao Paulo; Hertzman 1990on
Poland;Margulis1992on Mexico).

The World Bank Research Observer, vol. 7, no. 2 (July 1992), pp. 145-169
© 1992 The International Bank for Reconstruction and Development/THE WORLD BANK 145
 This article reviews the design of cost-effective interventions to protect the
environment from excessive pollution in developing countries. The focus is
deliberately limited to domestic policies to control pollution. We do not treat
policies to address other environmental problems, such as soil erosion, defor-
estation, biodiversity, or desertification; nor do we deal explicitly with transna-
tional problems (acid rain) or the global consequences of pollution (climate
change; ozone depletion). Many of the principles we present, however, relate
broadly to correcting for externalities, or spillover effects, and can be applied
to these other problems as well (see Binswanger 1989; Mahar 1989; Repetto
and Gillis 1988; Schramm and Warford 1989).
The analysis concentrates on policy instruments that have traditionally been
in the realm of public finance, such as taxes, prices, and subsidies, comparing
these with regulations and other instruments traditionally used to reduce pol-
lution or ameliorate its damage. The interventions can be categorized as
(a) market-based incentives (MBIs) that affect the incentives of private agents,
(b) command and control (CAC) instruments that regulate activity through con-
straints on the source of pollution, and (c) government spending on cleanup or
enforcement (table 1). In this article we focus on (a) and (b), because the prin-
ciples guiding the decisions regarding (c) are fairly well established in the cost-
benefit literature. We also distinguish between instruments directly associated
with the amount of damage created or pollutants emitted, and those addressing
pollution indirectly via related variables such as inputs and outputs.
The analysis begins by examining the rationale for government intervention
and goes on to look at the relative merits of the policy instruments traditionally
prescribed for that intervention in industrial countries. The arguments for these
standard policy prescriptions are based on several simplifying assumptions-

Table 1. A Taxonomy of Policy Instruments to Reduce Pollution

Policies Direct instruments Indirect instruments

Market-based Effluentcharges; Input/outputtaxes

incentives tradablepermits; and subsidies;
depositrefund subsidiesfor
systems substitutesand
abatementinputs
Commandand Emissionregulations Regulationof
controlmeasures (source-specific, equipment,processes,
nontransferable
quotas) inputs,and outputs
Government Regulatoryagency Developmentof "clean"
productionor expendituresfor technologies
expenditure purification,cleanup,
wastedisposal,and
enforcement

146 The WorldBankResearchObserver,vol.7,no. 2(July1992)

competitive markets, costless transfers, certainty, full information. These as-
sumptions may be less applicable in developing countries. The remainder of the
article investigates what happens to policy prescription when the standard as-
sumptions are relaxed to take conditions in developing countries into account.

The Rationale for Government Intervention

The efficiency argument for public intervention to mitigate pollution prob-
lems is well established in the theoretical literature (see Baumol and Oates
1979, 1988; Tietenberg 1988b). Pollution is an example of a negative external
effect; it imposes harmful effects and costs on people other than the polluters.
The free market offers the polluters no inducement to reduce the damage, since
the costs are largely paid by others. The market, if left to itself, is consequently
not the most effective mechanism for keeping pollution at reasonable levels.
Optimal control of pollution would occur if the marginal costs, including dam-
ages from pollution, were low enough to be balanced by the marginal benefits
from the activity.
Is public intervention in fact essential to correct for such externalities? Ac-
cording to Coase (1960), there is no efficiency reason for a government to be
involved except to help enforce property rights. Coase's proposition is that if
those affected by pollution hold the rights to an unpolluted environment, pol-
luters will "bribe" them to allow some level of pollution. (Member states of
the Organization for Economic Cooperation and Development [OECD] and
many other countries have, in principle, given property rights to victims
through the "polluter pays" principle.) Similarly, if polluters have the right to
pollute, victims will bribe them to pollute less. In either case, as long as nego-
tiations are not costly, the resulting amount of pollution would be optimal.
When polluters and victims are few and the number of beneficiaries from an
agreement is fixed, the Coase proposition may indeed be valid: negotiations can
provide for the internalization of externalities. Such a negotiated solution
seems feasible in a situation such as in the Philippines, where soil sediments
caused by a single logger threatened the development of tourism in a bay
(Dixon and Hodgson 1988). When rivers run from one jurisdiction to another
(as the Paraiba does between Sao Paulo and Rio de Janeiro), the parties in-
volved are easily identified, and they could agree on pollution loads through
negotiation. In Turkey, farmers have been awarded damages in court when
emissions from factories have hurt their crops; here the established right to an
unpolluted environment did provide incentives to abate. Thus, when stake-
holders are easily identified and law enforcement is assured, a case can be made
that government intervention is unnecessary for efficient outcomes.
But, for the Coase proposition to hold, the costs of negotiating and enforcing
agreements must be zero or negligible. In practice, the costs will increase with
the number of polluters and victims. In Mexico City, for example, there are

GunnarS. Eskelandand EmmanuelJimenez 147

20 million consumers, 2.5 million motor vehicles, and 30,000 industries; it
seems hardly feasible that the economic agents concerned will conduct efficient
negotiations without an intervening authority. In private negotiation, parties
have an incentive to free-ride, either by not revealing willingness to pay or by
breaking the agreement. Moreover, to be efficient in the long run, the agree-
ment must accommodate the entry of newcomers into the market and the exit
of others.' Public intervention will often be the only efficient solution to these
coordination problems.

Standard Policy Prescriptions

Given that intervention is required, what form should it take? Policies aim
either to regulate the level of pollution at the source or to change prices or reg-
ulations to increase the private costs of polluting. To facilitate comparison, dis-
cussion of the standard policy instruments that follows starts out with some
restrictive assumptions: (a) pollution is uniformly dispersed-that is, the exter-
nal costs to society are independent of the location of the source; (b) transfer-
ring revenue to or from the public sector is not in itself costly; (c) the costs of
monitoring damage and emissions are low; (d) there is no uncertainty about
the costs and benefits of pollution control; and (e) a competitive market struc-
ture prevails. Toward the end of this section we refine the comparison of in-
struments by relaxing the first assumption-uniform dispersion. The
subsequent section will relate the argument to developing country conditions
by relaxing each of the remaining assumptions in turn and adding the issue of
distributive objectives.

Uniform Dispersion of Pollutants

When pollutants are uniformly dispersed, ambient quality can be improved
or protected only by curbing emissions overall. Command and control policies
constrain emissions from each source and do not allow sources to trade the
right to pollute. Most industrial countries have relied predominantly on CAC
methods, setting and enforcing standards for equipment, processes, or emis-
sions (Opschoor and Vos 1989; Bernstein 1991). Emerging experience from de-
veloping countries does not appear to break this trend. With market-based
incentives, by contrast, constraints are not source-specific; they provide equal
incentives to all by increasing the marginal costs of polluting. Tradable permits,
subsidies for abatement, and emission taxes are examples of such instruments.
All let the market distribute abatement to where it is cheapest.

CAC VERSUS MBIs IN GENERAL. Command and control and market-based incen-
tives can achieve the same ambient quality, but market-based incentives are
generally more cost-effective. CAC can possibly minimize the cost to society if

148 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
the regulator tailors the abatement costs to each firm, so that no polluter is
asked to reduce emissions if another can do so at lower costs. The policy is
feasible if polluters are few and readily identified-for example, power plants
emitting sulfur oxides in some cities. CAC by a strong regulatory agency may
then be quite cost-effective. But generally, and especially in situations with
many heterogenous polluters, a large informal sector, and weak public admin-
istration, command and control policies will not work so well. Furthermore, in
practice, source-specific constraints allow marginal costs to vary among pollut-
ers, so that total costs of abatement are not minimized.
In contrast, market-based incentives generally require the regulator to esti-
mate only aggregate (rather than individual) costs of abatement to minimize
cost. A regulator would, for instance, issue enough tradable permits for the
marginal benefits and costs of abatement to be equated overall; a polluter
whose abatement costs are high would purchase permits, whereas one with
cheaper abatement options would prefer to reduce emissions (Dales 1968 is an
early proponent of this point). A pollution tax can also achieve cost-effective
abatement. It allows each polluting source to decide whether to pay the tax or
to undertake additional abatement, with the result that low-cost abatement is
selected since each source will abate only if the marginal costs of abatement
do not exceed the tax rate.
Empirical investigations have strongly supported the theoretical case for
market-based incentives: cost saving is reported to be significant over the com-
mand and control alternative. Tietenberg (1988b) reviews nine studies in which
market-based incentives are calibrated to reach the same ambience level as ap-
plied command and control methods. In seven of these studies, the ratio of MBI
to CAC costs is 1:4 or lower; in two of them, the ratio is 1:14 or lower.
Thomas's (1981) findings from Sao Paulo, with vastly differing abatement costs
between firms and sectors, indicate that savings from market-based incentives
would be high.
Market-based incentives are sometimes used in combination with command
and control instruments to obtain some of the advantages of a more flexible
approach without going to a full MBI system. In the United States, for example,
some limited opportunities to "trade" the right to pollute within a command
and control framework have been allowed in an attempt to gain some of the
savings possible with market-based incentives (see Hahn 1989; Opschoor and
Vos 1989). Expanded provisions for emission trading are among the amend-
ments to the United States Clean Air Act (U.S. Government 1990). Some of the
manifestations of emission trading have been netting, offsets, bubbles, banking,
and lead trading.
* Netting allows for internal trades within a firm: a firm can avoid the strin-
gent emission requirements for a new source if it reduces emissions from
existing sources.
* Offsets are used in areas where the establishment of new polluting activi-
ties is banned because air quality standards are not met. Through the offset

Gunnar S. Eskeland and Emmanuel Jimenez 149

 provision, a new source may be created if it will reduce more emissions
from another source than it will create itself. Germany and the United
States have offset provisions.
* Bubbles place an imaginary "bubble" over a factory. This allows a firm to
combine emissions from its various sources and comply with the general
rather than the particular requirements.
* Banking allows a firm to earn credits if its emissions are under the legal
limit; credits can be used later or sold to others.
* Lead trading between refineries in the United States was allowed in order
to reduce the costs of a major program to phase out lead in gasoline
(1982-87).
The savings from limited opportunities to trade within bounds like these are
substantial if the trade allowed is between sources that would otherwise abate
at vastly differing marginal costs. Hahn (1989) estimates that the savings relat-
ed to netting, offsets, bubbles, and lead trading have been considerable.

PRICE-BASED VERSUS QUANTITY-BASED INCENTIVES. As long as there is no un-

certainty about abatement costs, price-based incentives (such as taxes) and
quantity-based incentives (such as tradable permits) have exactly the same ef-
fect. The same level of emissions and economic costs should result. A uniform
emission tax will have the same incentive effects as emission permits, because
the market will distribute them within the industry according to willingness to
pay. Both minimize abatement costs overall, because high-cost abaters will ei-
ther pay the tax or outbid low-cost abaters for permits.
But although taxes and permits that can be traded freely are conceptually
equivalent, some analysts have argued that, for administrative reasons, tradable
quotas may be preferable to tax or price instruments (Baumol and Oates 1988).
First, adjustment of the tax rate to reach appropriate environmental goals may
be costly, particularly in inflationary environments. Second, permits may be
easy to implement, because they make it possible to introduce controls without
increasing the costs for existing firms. But in some cases, tax-like instruments
may be easier to administer than quotas. For example, an indirect pollution tax
on fuel consumption may fit more easily into existing administrative processes
than a quota (Anderson 1990). In most developing countries, input taxes to cur-
tail emissions would be easier to implement than a completely new scheme such
as permit trading.
Unless permits are auctioned, their distributive implications differ from
those of taxes. Nonmarket distribution of permits, as well as other compensa-
tion schemes, must be designed with caution so as to avoid creating undesirable
incentives.

SETTING PRICE-BASED INCENTIVES. Governments using a pollution tax to pro-

tect ambient quality should select a base and rate so that the external cost of
the activity is internalized. Such an instrument is often called a Pigouvian tax. 2

150 The World Bank Researcb Observer, vol. 7, no. 2 (July 1992)
The appropriatebase for the tax should be the damage causedor a closeproxy
for damage, such as the volume of [Link] examplemight be a carbon
tax, which has lately been proposedas a tool for efficientreductionof the emis-
sions that cause global warming. Other examplesare taxes applied on air pol-
lution in France, on dischargesinto water in Germany,and on solid waste in
Denmark (Whalleyand Wigle 1991;Opschoor and Vos 1989).
The Pigouvian incentive can be either a tax on pollution or a subsidy for
abatement. In the short term, the incentiveeffectscan be the same. In the long
term, when market entry and exit can be affected, a tax is normally preferable
because it does not give firms incentivesto enter a subsidized polluting in-
dustry.
What determinesthe rate of the tax? The general rule is to set taxes or pub-
lic pricesof commoditiesthat produceexternalitiesat a rate equivalentto mar-
ginal cost of production plus the incremental value of the externality (see
Eskeland and Jimenez 1991for a full discussion).There is no need to tax com-
plements or to subsidizesubstitutes when the tax on the polluting good fully
internalizesthe [Link] practice,many polluting activitiesare subsidized,
being priced at less than marginal cost (Schrammand Warford 1989;Repetto
and Gillis 1988;Anderson 1990).In such cases, raising prices or taxes closer to
marginalcost will already be an improvementover the present situation; emis-
sions can be reducedwith no cost at all.

Nonuniform Dispersion of Pollutants

The precedingdiscussionof policy instruments,for the purposes of compar-
ison, has assumed uniform dispersal of pollutants. Greenhousegases, such as
carbon dioxide, are uniformlydispersed. But often pollutants are concentrated
in some pattern around the source and downwind or downstream. Removing
the assumptionof uniformdispersaldoes not affect the general argument about
the relativemerits of command and control policiesversus market-basedincen-
tives, but it does have implicationsfor the details of policy design.
The damage (per unit of emission)caused by a polluting activity will vary,
depending on location, on the dispersion characteristics of emissions (deter-
mined, for air pollution, by such variables as stack height or speed and tem-
perature of flow), and on the site being polluted. For many major air
pollutants, reducing emissions has few benefits over vast rural areas, so that
applying uniform emission charges (or one-for-onetradable emission permits)
in zones comprisingboth urban and rural areas would not be cost-effective.
The rural/urban dimensionillustratesdifferencesin marginaldamages per unit
of [Link] principle,and the need for differentiatedinstruments,is valid
for any pattern of nonuniformdamages (or, equivalently,benefits).Area A may
be vulnerable,but less important to protect than area B if higher present pol-
lution loads in B make the marginal damage there [Link],for water
pollutants such as those characterizedby biologicaloxygen demand, marginal

Gunnar S. Eskeland and Emmanuel Jimenez 151

 damage from discharges may be high where discharges are high but lower both
upstream and downstream. Uniform emission charges might require unneces-
sary abatement from many sources whose emissions do not pollute the "hot
spots." In these cases, the most cost-effective program would require abatement
for each source according to whether its emissions pollute vulnerable locations
or not. For example, emission charges could vary by location to reflect the ratio
of damages to emissions.
In essence, differences in damages per unit of emissions mean that unlimited
trading of emission permits is not a good idea. If zoning is in effect, permits
need to be distributed carefully among zones, since sources are barred from
solving problems of misallocation through trades.
If the polluters in a region display very different ratios of damages to emis-
sions, differentiating instruments accordingly can yield significant cost savings.
By the same token, the costs of applying uniform emission charges or permits
without zoning will also be high. Atkinson and Tietenberg (1982) calculate that
a market-based scheme for particulate emissions in St. Louis would save 83
percent of the costs of a command and control scheme if charges were fully
differentiated. The savings would be lower if geographical differentiation of
emission charges were limited, and would be only 50 percent if charges were
uniform throughout. Seskin, Anderson, and Reid (1983), modeling control
strategies for nitrogen oxides in Chicago, find that savings of 93 percent relative
to command and control fall to 50 percent if instruments have to be uniform
within industries, whereas a scheme with completely uniform instruments
would cost twice as much as the command and control strategy. The regula-
tors, not surprisingly, have had an eye on the geographical dimension when de-
signing the command and control scheme, and thus it did better than a market-
based scheme that ignored geography.

Efficient Pollution Control Policy in Developing Countries

How are the standard results just discussed affected when we change the
standard assumptions? Let us assume a public revenue constraint, inability to
monitor emissions and damages, uncertainty, and a market structure that is
not competitive-constraints that confront policymakers in many developing
countries.

Pigouvian Taxes under a Public Sector Revenue Constraint

Public budgets are often tight in developing countries. Raising additional
revenue through existing tax structures can injure resource allocation, as firms
and households adapt to a distorted price regime. And such distortions can be
immense: estimates from the United States assess costs at 17 to 56 cents for
every additional dollar of tax collected (Ballard, Shoven, and Whalley 1985). In

152 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
developing countries, where the tax base is often narrow and rates high, dis-
tortionary costs are likely to be higher (World Bank 1991).
One attraction of pollution taxes is that they can raise revenue while im-
proving efficiency, by persuading firms and households to reduce negative ex-
ternalities. Taxing commodities with negative externalities will thus reduce not
only the efficiency losses arising from the externality itself (say, damage from
pollution) but also the efficiency losses related to generating revenue, since the
proceeds may allow other rates to be reduced.
How much can be raised by such efficient taxes? Sandmo (1975) shows the-
oretically that pollution taxes belong in an optimal tax structure, adding to the
rates on polluting goods. Some empirical studies indicate that revenues from
efficient pollution control policies will be of the same order of magnitude as
total control costs. In a simulation of particulate control in St. Louis, Atkinson
and Tietenberg (1982) found that proceeds from permit charges would be in
the same range as abatement costs. The study of control strategies for nitrous
oxides in Chicago (Seskin, Anderson, and Reid 1983) also found charges in the
same range as control costs.
So far, charges have not made an impressive contribution to general revenue
in OECD countries. At less than a third of 1 percent of gross national product
(GNP) in the Netherlands and at 0.04 percent or less of GNP in the other
countries surveyed, the revenues were found to be of no importance for the gen-
eral budget (Opschoor and Vos 1989). The OECD study further showed that
charges were rarely effective in changing behavior because rates were too low
and the base was usually insufficiently responsive to individual behavior. But
proceeds from pollution charges were an important mechanism for funding
selective environmental expenditures in countries where they were earmarked
for that purpose. Indirect taxes, such as fuel taxes, could be important in gen-
erating revenue: for countries such as Pakistan, the Republic of Korea, and
Turkey, a 20 percent increase in the tax on fossil fuels could raise an additional
3-5 percent of revenue. Gil Diaz (1987) found that energy subsidies in Mexico
from 1977 to 1984 amounted to $100 billion-equal to the total amount of
foreign public debt. Clearly, the budgetary implications of energy pricing can
be immense.
Pollution charges may yield more or less than what is needed for environ-
mental expenditures, so the benefits of earmarking should be examined in the
broader context of public expenditure analysis (see McCleary 1991 on the pros
and cons of earmarking).

Monitoring Damages or Emissions: A Role for Indirect Instruments

The theory reviewed earlier implies that the efficient economic policy
(whether market-based or command and control) is to address the external ef-
fect directly-for instance, by taxing a polluter or regulating emissions accord-
ing to the environmental damage caused. In practice, monitoring damages or

Gunnar S. Eskeland and Emmanuel Jimenez 153

even emissions at the source may be costly, particularly in developing countries,
for technological and institutional reasons. The emissions of many pollutants
can only be imperfectly monitored, and the fact thatthe results of monitoring
may cost the polluter money does not make the task easier. In the United States
the Environmental Protection Agency (EPA) has 14,000 federal employees, and
even so, it monitors very few sources continuously. Russell (1990) notes that
the EPA largely trusts corporations in matters of compliance and self-reporting,
even though experience with motor vehicles has shown that active tampering
(not only inadequate operation and maintenance) is prevalent (see also Hamrin
1991). The monitoring problem is likely to be worse for regulatory agencies in
developing countries, since the agencies are often inadequately funded and have
less access to technology and trained labor. In Mexico only three plants are
scheduled to install equipment for continuous monitoring of emissions; the rest
of the 300 most-polluting firms report their own emissions.
To be credible, regulatory agencies must be able to design and administer
new schemes; behavior will change only if a threat to penalize noncompliance
is seen to have teeth. Even in industrial countries, sophisticated schemes, such
as permit trading, have run afoul of implementation issues (Hahn 1989). And
even when monitoring is technically feasible, institutions may be too weak to
enforce the taxes or regulations based on the monitoring.
When environmental damage or emissions cannot be tackled directly be-
cause monitoring and enforcement costs are too high, the regulator will use
indirect instruments aimed at the outputs and inputs of the polluting industry
or substitutes and complements to its outputs. Indirect pollution taxes applied
to fuels such as coal and gasoline-presumptive Pigouvian taxes-are an
example. How would the use of such instruments affect the policy choices dis-
cussed earlier? With indirect instruments, the "tradability" results apply, in that
actions that reduce (increase) emissions by an equal amount should receive an
equal subsidy (tax). Indirect instruments, however, typically specify physical ac-
tions (such as installation of catalytic converters) rather than emission reduc-
tions, and they often yield more emission reductions at one place than another.
Investments that reduce emission coefficents for vehicles, for instance, yield
highest benefits for the vehicles that are used most intensively. For this reason,
indirect instruments will often need to be applied with stronger inducements
to specific polluters and, consequently, will have aspects of command and con-
trol. An example in which this principle is applied: In Mexico City, high-use
vehicles such as taxis and minibuses are required to be cleaner than vehicles in
general. Presumptive emission taxes on fuel will thus be effective if emission
coefficients for one fuel are the same across users, but they should otherwise
be differentiated, if possible. The cement industry, for instance, which does not
discharge the sulfur of its fuels, should ideally be refunded presumptive sulfur
taxes on fuels.
In assessing the desirability of indirect instruments, the reduced cost of the
externality must be compared with the distortions the instruments themselves

154 The WorldBankResearchObserver,vol.7, no. 2 (July1992)

create through their effect on other choices. A fuel tax, for instance, is efficient
if the relationship between consumption and emissions is fixed, if nonpolluting
fuel use will be unaffected, and if fuel consumption can be monitored relatively
easily. A fuel tax is relatively inefficient, however, for fuel used in polluting
as well as nonpolluting activities. Furthermore, other effective abatement
measures (such as catalytic converters and scrubbers) are not triggered by fuel
economy and thus require instruments other than fuel taxes. Deposit refund
systems, similarly, can be attractive when unsafe disposal, rather than use,
causes damages (Bohm 1981).
Indirect instruments are blunter tools for tackling pollution in that they
may also have undesirable effects on behavior. How can such side effects be
minimized?

OUTPUTS AND INPUTS AS A BASE. If a polluter's emissions are fully determined

by the consumption of one good, then taxing the good according to marginal
external costs is equivalent to an emission tax. Carbon taxes on fuels are an
example, because the external effects are independent of both source location
and combustion process. But if a polluting good cannot be fully taxed, a related
good should be taxed if it is a complement to the polluting good, and subsi-
dized if it is a substitute good. For example, suppose that private vehicle use
in urban areas is polluting but cannot be taxed sufficiently (or only at a pro-
hibitive cost). A clean substitute such as a subway should be subsidized, but a
clean complement such as central parking facilities should be taxed. This holds
unambiguously as long as demand for subway service is unrelated to demand
for parking space.
Wijkander (1985) shows that taxation of complements and subsidization of
substitutes are efficient under fairly general assumptions, but counterintuitive
results can take place depending on own-price and cross-price elasticities of
demand. These occur when the indirect instruments bring unintended distor-
tions. In our motor vehicle example, if public transport and central parking
facilities are sufficiently strong substitutes for each other, subsidizing subways
and taxing parking spaces may lead to overutilization of subways. If this prob-
lem arises, parking space should not be taxed so much and may even have to
be subsidized.
Sandmo (1976) presents another situation in which indirect instruments are
useful. A commodity is used by consumers for two purposes, only one of which
has negative external effects. Gasoline use, for example, can be said to be "in-
nocent" when used for countryside driving, but it has negative external effects
in terms of pollution and congestion when used in cities. A commodity tax to
address the externality is then itself distortionary in relation to the innocent
use, and the question is whether an additional instrument on a related good
can reduce the costs of that distortion. Sandmo concludes that a related
good should be taxed if it is a complement to the polluting activity and a sub-
stitute for the "innocent" activity, and subsidized if it is a complement to the

GunnarS. Eskelandand EmmanuelJimenez 155

"innocent" activity while a substitute for the polluting activity. Intuitively,
supplementinggasoline taxes with taxes on central parking and central road
use, and subsidizingparking at peripheral metrorailand bus stations, would fit
in this picture. If the related good is a complementto both uses, it should be
taxed if it is more complementaryto the polluting use, but subsidizedif it is
more complementaryto the "innocent" use. Likewisefor a substitute good: it
should be subsidizedif it is more substitutable to the polluting use, but taxed
if it is more substitutable for the "innocent" use. For instance, if bus services
are substitutes for polluting cars, but also for "innocent" bicycles, then one
would subsidizebusesonly to the extent that the formereffect is stronger. (See
Sandmo 1975and Balcer 1980for further illustration of this point.)
There are severalpossible bases for pollution taxes:
*Taxing the damagecreated makes it possible to differentiate betweenpol-
luters according to the amount of damage caused per unit of emission.
Each source balances marginal abatement costs equally against individual
[Link] such taxes have yet beenapplied. Oth-
er damage-relatedinstruments are used, [Link] rules hold pol-
luters accountable for accidents, such as oil spillage (a market-based
incentive). Offsets, bubbles (market-based),and zoning policies (market-
based and command and control) give some considerationto the location
of the source and thereby to the amount of damage it creates. Ambience
permit systemsare markets in damage quotas and have been simulatedbut
not tried in practice.
* Taxing emissionsminimizesthe costs of abatement by equalizingmarginal
abatement costs across sources but does not differentiate between sources
according to damages. Emissioncharges thus fail to provide incentivesto
relocatewithin a [Link]'s tax on emissionsto air,
water chargesin Germany,waste chargesin Denmark, manure taxes in the
Netherlands, and carbon taxes in Norway.
* Taxing inputs and outputs of polluting activities mimics an emission or
damage tax (imperfectly)but fails to giveincentivesto minimizeemissions
or damages for a given level of inputs or outputs. As an example, lead in
gasoline is taxed in Norway and [Link] countries also tax fossil
fuels in general.
* Taxing (or subsidizing) fixed inputs of polluting activities can provide
some abatement incentive, but fails to influencehow carefully and fre-
quently the equipment is maintained and used. Tax differentiationis ap-
plied according to emission characteristics of cars in Germany, the
Netherlands, Norway, and [Link] countries tax noise characteris-
tics of aircraft and subsidizeor mandate the installation of "clean" equip-
ment or processes.
* Taxing complementsand subsidizingsubstitutesare alternativesif the pol-
luting activity is untaxable; they can also be valuable supplementaryin-
struments. Apart from complementaryinputs (vehicles),complements are

156 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
 not otherwise known to have been taxed. As a substitute for private trans-
port, urban mass transport is subsidized almost everywhere.
Taxing unreturned items. Depending on cost relationships, many materials
will be recycled without government intervention, particularly if labor is
cheap. When additional incentives to recycling are desirable because of ex-
ternal costs, a deposit refund system may be efficient. Such systems have
been proposed for batteries and are in place for car hulks and beverage
containers in Scandinavia.

FIXED INPUTS. If one cannot monitor emissions or variable inputs and outputs,
one can still achieve something by targeting fixed inputs, such as equipment
and installations. When it is difficult to observe what people do and earn, gov-
ernments have often resorted to presumptive taxation to raise revenue, basing
the taxes on observable proxies for income such as land ownership or house
size. To correct for externalities, the analogous approach would be to tax pol-
lution-generating equipment as if it were used (when use is unobservable) and
to tax cleaner equipment at a lower rate. Unlike presumptive taxes used for
revenue collection, presumptive taxes on pollution make sense only if they
affect behavior.
Technical standards can thus be regarded as indirect instruments under mon-
itoring costs. In the United States the costs of monitoring emissions continu-
ously are prohibitive not only for mobile sources, but often for stationary
sources as well (Hamrin 1991). In the United States, Mexico, and many other
countries, emissions from mobile sources are controlled through testing of new
sources (and in some places through annual testing), even though the results
are imperfectly correlated with actual emissions.
Technical standards have been heavily criticized by economists because they
tend to be applied in a mandatory, uniform, and thus excessively costly way-
and generally in the form of command and control policies, instead of through
selective taxes or subsidies on equipment that are differentiated according to
presumed emissions. Often, regulations are applied only to new sources, failing
to exploit abatement opportunities among other sources. Another problem
with this "new-source bias" is that it may increase the market power of incum-
bents.
Many developing countries now follow the United States's example in low-
ering vehicle emissions mainly by setting strict tailpipe standards for new vehi-
cles. Thus the major instrument for control is a source-specific constraint
applied to equipment, as opposed to emission taxes or taxes on variable inputs
such as fuels and road use. The strategy certainly reduces emissions per kilo-
meter traveled. It would be more productive, however, when combined with
fuel taxes or other instruments that encourage people to buy smaller vehicles
or use their vehicles less.
The effects of indirect instruments are often subtle, because they are
frequently a stopgap device, instituted to compensate for the lack of first best

Gunnar S. Eskeland and Emmanuel Jimenez 157

instruments. Take, for example, the interaction betweenstandards for new ve-
hicles and a proposed presumptivePigouviantax on fuels in Mexico [Link]
new cars are purchased, the vehicle fleet will become cleaner and larger.
Whether the Pigouvian tax should decline over time as the fleet becomes
cleaner on average depends on whether the air pollution situation overall
deteriorates or not.

Charges and Permits under Uncertainty

Evenunder perfect monitoring,the effects of environmentalpoliciesmay be
uncertain. The benefits from abatement may be subject to events that are in-
herentlydifficultto predict. For example,the effectsof air pollution can depend
on the health of the affected population-about which there is little informa-
tion in developingcountries. Furthermore, the costs of abatement depend on
how easily polluters adjust, which a regulator cannot know with certainty.
The standard recommendationsabout the choice of instruments are affected
by uncertaintyin three ways: (a) the equivalenceof price-basedversusquantity-
based instrumentsmay no longerhold; (b) in an uncertainenvironment,flexible
instruments are better; and (c) liability rules may be an attractive option.

PRICE-BASED VERSUS QUANTITY-BASED INSTRUMENTS UNDER UNCERTAINTY.

When the marginal costs of pollution abatement are known to the regulatory
agency,uncertainty about the benefits does not favor one type of instrument
over the other. Firms' abatement response depends only on costs and on the
policy instrument, which are both known to the firm. So, even if the benefits
deviate from expectedlevels,the abatement level and the efficiencylosseswill
be exactly the same whether the price or the quantity instrument is used.
When abatement costs are uncertain to the agency,producers are assumed
to have information that the agency does not have (Weitzman 1974). The
quantity instrument, which can guarantee an emissionlevel, is better if unex-
pected emissionloads are costly. This would occur if marginal damages from
unexpectedly high emissionsare steep (as with leaks from nuclear facilities).
Correspondingly,the price instrument, which guarantees that marginal abate-
ment costs do not exceed the tax rate, is better when marginal damages are
relativelyflat.
In an empirical study, Kolstad (1986)evaluated policies to control sulfur
emissionsfrom power plants, where abatement costs were uncertain because
of uncertainty about future electricitydemand. He found that if marginal ben-
efitswere constant, a price instrument would be slightly preferable, but that a
mild slope would be enough to make permits the more desirable option. Lyon
(1989) argues that tradable permits are particularly attractive to developing
countries because they provide certainty about ambient quality in a dynamic
context, and they also allow implicit property rights to be changed gradually.

158 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
However, if marginal benefits are seen as fairly flat, then price instruments do
have the attraction of placing a definite upper bound on control costs.
A combination of instruments may be the solution. Roberts and Spence
(1976) suggest a permit to be accompanied by a (high) fee for further emissions
and by a promise to repurchase unused parts of the permit (as a subsidy for
additional abatement) at a (low) price. As they explain, "The subsidy provides
a residual incentive for firms to clean up even more when costs are low. The
finite penalty provides an escape valve if costs are very high" (page 203).

FLEXIBILITY OF INSTRUMENTS. What if the planner can adjust policy instru-

ments when information is revealed? The results just mentioned may change if
some instruments are more easily adjusted than others. Bawa (1988) suggests
a mixed policy under the assumption that a regulatory policy (command and
control) can be put into effect faster and more flexibly than a charge (or a trad-
able permit).
Suppose, for example, that stochastic changes (such as weather) make the
ambient quality resulting from a steady flow of emissions worse in some
periods than others, but the periods are too short for the effluent charge to be
adjusted. An emission tax can then be complemented by a command and con-
trol instrument-for instance, factories are closed down under a "smog alert"
(see Plourde and Yeung 1989). In practice, pollution control authorities are of-
ten authorized to shut down polluting activities selectively on short notice. In
several cities in Brazil and also in Mexico City, industries are shut down during
crisis periods (Sebastian forthcoming). Both Mexico City and Santiago close
central areas to traffic when air pollution reaches threshold levels.

LIABILITY RULES. If monitoring of actions to avoid causing damage is

expensive but the source of discharges or spills can be identified, a liability rule
might usefully substitute for a regulation (Bohm and Russell 1985). This view
of liability is similar to the Coasean proposition about negotiation-that nego-
tiation between polluters and victims can take the place of government inter-
vention-but it explicitly requires support from a legal system. Liability rules
will usually distribute property rights, for instance, by postulating that every-
body has the right not to be harmed by others. Liability rules can be seen as a
complement to other regulations (Posner 1986; Farrel 1987) and will then un-
doubtedly allow for more flexible case-by-case damage assessment. Incidents
such as the Bhopal catastrophe and the Prince William's Sound oil spillage
(with sizable damage awards and losses of reputation for the responsible firms)
warn firms of their potential liability and thus induce them to take more pre-
cautions.
But there are many potential problems that might limit the usefulness of li-
ability rules: low likelihood that offenders will be detected, high costs to
victims for representation and litigation, the unpredictability of an underdevel-
oped judicial process, and the potential insolvency of the liable party. Ringleb

Gunnar S. Eskeland and Emmanuel Jimenez 159

and Wiggins (1990) find that in the United States, where liability is generally
unlimited, industries prone to litigation are increasingly being dominated by
small, independent firms. Shell, for instance, pulled out of oil transport in U.S.
waters in 1990. This pattern could indicate that less wealth is backing the po-
tential liabilities, in which case the incentives given may be less powerful, since
owners have little to lose. Bohm and Russell (1985)note that the liability in-
strument may encourage people not to protect themselves against pollution,
since the price paid by the polluter is actually passed on to the victims. Kolstad,
Ulen, and Johnson (1990) add that uncertainty about liability assessment gives
a rationale for supplementing ex-post liability with ex-ante regulatory stan-
dards.

Noncompetitive Market Structure

The rules of policy intervention generally assume that markets are compet-
itive. But often that assumption is untenable. In industrial countries, utilities
are prime examples of monopolies; they are often subject to controls on both
pricing and emissions. In developing countries, many markets may be small;
entry barriers, tariffs, and transportation costs high; and access to credit, tech-
nology, and law enforcement limited. Utilities and some industries are often
public and loss-making, and profit maximization-or even cost minimization-
may not be a primary objective.
How do the recommendations for policy intervention change when the pol-
luting firm is also a monopoly (such as a utility) whose market power permits
it to supply less than optimal output? In such a case there are two sources of
market failure-pollution externalities and market power. Indeed, it is theoret-
ically possible (since output is lower from a monopoly than from a competitive
firm) that the exercise of monopoly power may partially address the pollution
externality (Buchanan 1969). There would be efficiency gains with two instru-
ments to address each source of market failure. But, if a tax to address the
externality directly were infeasible, an output tax would address both problems
and would be the sum of a Pigouvian tax and a subsidy to output (Baumol and
Oates 1988).
Market power introduces a more serious problem if abatement is to be reg-
ulated within a market for pollution permits. If the permit market does not
result in competitive pricing, abatement will not be efficiently distributed, even
if the number of permits available is optimal. That permit markets will often
be fragmented and interdependent may be an additional argument for using
price instruments. If polluters behave strategically to manipulate the charge,
however, problems similar to those in permit markets will emerge (see Jack
1990).
Different policies may affect firms' entry into and exit from markets, and
thereby the extent of their market power (Dewees 1983; Kohn 1988; Spulber
1985). This intuitive conclusion runs counter to the finding in short-term

160 The World Bank Researcb Observer, vol. 7, no. 2 (July 1992)
static models that nonmarginal rewards are neutral. The most important
consideration is probably that quotas and standards, if they raise the costs
of entry, could provide for collusion among existing firms. Assuming com-
petitive behavior, Spulber (1985) shows that an optimal effluent charge (or
number of tradable permits) will yield the efficient number of firms and ef-
ficient output even if there are economies or diseconomies of scale in produc-
tion and abatement.
There are many claims that industries in industrial countries (paper and
pulp, for example, and copper smelting) have become more concentrated as a
result of environmental control policies. This may, however, be a response to
excessive concentration on mandated equipment, or a reflection of the fact that
abatement requirements usually favor incumbents, thereby discouraging new
entrants. (As noted earlier, abatement activities that rely on equipment and
fixed installations may justifiably be preferred to other, equally cheap
abatement options if they are less costly to monitor.) There are also examples,
such as fossil-fueled power generation and steel production in the United
States, where pollution control policies have eroded some economies of scale,
thus giving a boost to small plants (Gollop and Roberts 1983).
In developing countries, small firms in the informal sector are often major
polluters. Restructuring and concentration in an industry could lower the costs
of monitoring and enforcement, but using flexible instruments such as taxes
and regulation of inputs may also save on those costs. This may be a way to
curb emissions from small firms without forcing them underground or out of
business.

Distributive Effects: Welfare and Political Economy

The distributive effects of alternative pollution control policies are important
for -two reasons. First, an increase in income may be valued more highly for
the poor than for others. Second, the government should know who gains and
who loses so that it can judge whether a particular reform will have sufficient
political support.3

Environmental Policies and the Poor

The distribution of the benefits of pollution control is an empirical issue,
and the evidence on the differential health effects of reducing pollution is
mixed. Several authors have noted that the poor are likely to benefit more
(see Anderson 1990; Asch and Seneca 1978), because they tend to live in un-
healthy, unsanitary conditions in polluted urban areas and cannot afford
to protect themselves or move. Some U.S. studies (see Christiansen and
Tietenberg 1985 for a review) support this view-they show that air pollution
is worse in cities with many poor residents and, within cities, in the areas

Gunnar S. Eskeland and Emmanuel Jimenez 161

 where the poor live. Theoretical arguments and some empirical evidence,
however, indicate that wealthier people may be more willing to pay for envi-
ronmental improvement than the poor. This could make the wealthy the prin-
cipal beneficiaries of control policies (Johansson 1987; Christiansen and
Tietenberg 1985).
Most of the researchers comparing the incidence of costs assume the distri-
bution of benefits to be uniform. An exception is Harrison (1975), who notes
that the policies to control air pollution from motor vehicles in the United
States have affected the rural poor badly-their car ownership rates (and thus
control costs) are necessarily high because they have no public transportation
alternatives, and the environmental benefits in their areas are modest. In the
1990 Clean Air Act Amendments (U.S. Congress 1990), emission standards for
vehicles were made stricter only in the most polluting cities, and this could
somewhat reduce the costs to households that derive only minor benefits. Such
differentiation makes pollution control harder to administer but perhaps more
efficient, if damages per unit of emissions are lower in less polluted areas. As
for water pollution in developing countries, it is clear that the wealthy can pro-
tect themselves from exposure. Private wells, piped water, and bottled water
are available to households that can afford them, leaving the poorer families
most vulnerable to surface water conditions.
The incidence of the costs of alternative pollution policies depends on the
ability of polluters to pass on the costs to customers, to other producers (if
intermediate goods are produced), and to workers. Thus, the elasticities of de-
mand and supply in production and consumption must be analyzed along with
the amount that poor households spend on the polluting goods. A higher gas-
oline tax, for example, may have very little effect on slum dwellers who own
no cars and do not use public transport. A diesel price hike may affect com-
muters and rural communities; vulnerability will depend on the availability of
substitutes. Regulatory intervention can also have distributive effects because
it affects the cost of doing business, which can be shifted forward to prices or
backward to labor and capital. The poor may be especially vulnerable to
selective price increases and at greater risk of unemployment if an industry
becomes uncompetitive (Yu and Ingene 1982; Yohe 1979).
In industries for which environmental regulations are very costly, changes
in costs and practices may affect incidence. For example, if labor is a substi-
tute for polluting inputs (labor and energy are often substitutable), abatement
policies can lead to increased labor-intensity in production and can thus in-
crease employment or remuneration of labor. Another situation arises when
abatement does not change input combination but simply raises costs in the
industry. Assuming flexible prices, Forster (1983) uses a simple theoretical
general equilibrium framework to show that the factor most intensively used
in the polluting sector will lose remuneration, while the other factors will
earn more. Alternative assumptions, such as the introduction of price/wage

162 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
rigidities (and possible unemployment) or immobile factors would affect the
results.
Most empirical studies incorporating the incidence of costs do not explicitly
compare alternative instruments. Pearson and Smith (1990) find that carbon
taxes sufficient to reduce carbon emissions in the United Kingdom by 20 per-
cent would raise £8 billion in the short run and would be highly regressive,
reducing the real income of the very poor by up to 2.7 percent compared with
only 0.4 percent for the very rich.4 They do find, naturally, that if all the pro-
ceeds from the carbon tax were redistributed to the poor, the scheme could end
up being progressive.
Because the poor most likely spend more of their income, price increases
generally tend to have a regressive effect (Gianessi, Peskin, and Wolff 1979;
Dorfman 1975), although this may be seen as a transitory phenomenon. Over
and above this, Yan and others (1975) found no pattern of regressiveness or
progressiveness arising from the specific prices that increased as a result of
environmental policies. Water pollution treatment, which is typically partially
financed through charges and taxes, appears to have been progressive in some
cases and regressive in others.
Lessons about presumptive Pigouvian taxes on marketed goods and services
can be derived from empirical studies in developing countries. An internal
World Bank study on Indonesia found that the average household spends 4.2
to 4.5 percent of its income on fuels and 0.2 to 0.3 on transport. Among man-
ufacturing industries, no industry had an elasticity to costs of fuel prices higher
than 9 percent (most are in the range of 1-3 percent), and labor was assumed
to be a substitute for energy. Another internal World Bank study, on Mexico,
found that wealthy households spend a higher share of their budgets on trans-
port and communication, suggesting a good tax base for distributive purposes.

Political Economy and Implementation

If some instruments are so much better than others (for instance, market-
based incentives rather than command and control, direct rather than indirect),
why are they not applied more in practice? Many researchers have shown that
policy outcomes are influenced not so much by net gains to society as by who
gains and who bears the burden of different strategies. (See Hahn 1989 for a
brief review and interpretation; see also Baumol and Oates 1979, 1988; Dewees
1983; Buchanan and Tullock 1975.) If a group that prefers one instrument over
another can influence policy decisions, it is likely that a policy will be chosen
that does not minimize costs. For example, Buchanan and Tullock (1975,
pp. 141-42) compare regulation to an effluent tax, and note:
Regulation is less desirable on efficiency grounds ... but this instru-
ment will be preferred by those whose behavior is subjected to either
one or the other of the two policy instruments.... [In] their own

Gunnar S. Eskeland and Emmanuel Jimenez 163

 private interests, owners of firms in the industry along with employ-
ees will oppose the (effluent) tax. By contrast, under regulation firms
may well secure pecuniary gains from the imposition of direct con-
trols that reduce total industry output.... This political choice set-
ting is... the familiar one in which a small, concentrated, identifiable
and intensely interested group may exert more influence on political
choice making than the much larger majority.
Thus, a tax on emissions is unpopular among influential polluting indus-
tries that, for any given level of abatement effort, strongly prefer not to pay
for the remaining emissions. The regulatory agency, for its part, is likely to
settle for any solution that is consistent with its ambient quality goals. Those
who pay the additional costs of inefficient intervention are citizens at large,
who are often badly organized compared to the industry in question. In addi-
tion to the inefficiency problems already mentioned, real-world command and
control policies tend to be selective and thus very susceptible to rent-seeking
behavior.
Political economy-the study of how distributive effects affect decisions-
can also be important in setting priorities among environmental problems.
Dixon (1991) argues that the wealthy can privately avoid the consequences of
water pollution to a greater extent than they can avoid the consequences
of air pollution, and that as a result, water pollution control will lag in the
policy process.

Implications for Policy Reform

Which policy instruments are more efficient, practical, and equitable under
the constraints that developing country governments confront? What research
must be done to provide better answers to the questions we have raised?
The stringent budgetary restrictions in developing countries probably
strengthen considerably the case for price-based versus quantity-based interven-
tions, particularly if the instruments are related only indirectly to damages or
emissions.
Even though quantity-based interventions (such as tradable permits) can
incorporate aspects of market-based incentives, they will in most cases re-
quire new administrative systems. For indirect instruments, many govern-
ments already have ways of charging for or taxing relevant commodities, and
the administrative capacity to manipulate domestic relative prices is generally
well established. What price-based interventions and permits have in common
is that they can generate revenue. But the potential revenues from pollution
control instruments are likely to be small in relation to the overall needs
of the treasury-except in the case of fuel taxes, which can generate vast
revenues.

164 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
 In an uncertain world, price-based instruments provide greater certainty
about abatement costs. Such instruments are superior when there are concerns
that underestimatingcosts would yield controls that are "too strict" and envi-
ronmental quality that is "too high"-concerns that arguably trouble policy-
makers in many developingcountries.
The costs of monitoring individualemissionsand enforcingtheir abatement
can be high, particularly in developingcountries,for technologicaland institu-
tional [Link] taxes or subsidiesfor marketed inputs and outputs are
desirable alternativesbecause they depend less on vulnerableand costly moni-
toring and [Link] these instruments also imply additional
costs because they can only imperfectly mimic taxes on monitored emissions
and damages. The choice of which commodities to tax and at what rates
should be sensitiveto demand relationships,which play a role in emissionre-
duction and in the probable [Link] empirical issues should be the
subject of future research.
The considerationsraised in this article do not mean that instruments such
as tradable permits should not be tried out, but that less sophisticatedinstru-
ments should be consideredwhereverthey can potentially do much of the job
in an easy-to-implement,low-cost way. Then, monitoring and enforcement
capacity can be developedand priorities set for remaining problem sectors or
pollutants. Schemesthat encourageself-compliance,such as deposit refund sys-
tems, should also be considered.

Notes
Gunnar Eskeland is an economist in the Country Economics Department of the World Bank.
Emmanuel Jimenez is a division chief in the Bank's Population and Human Resources Depart-
ment.
1. On the long-term efficiency of the negotiated solution with well-defined property rights,
see Frech (1973) and Tybout (1972, 1973). Efficiency can be the result if firms that leave or en-
ter a market can charge or be charged for doing so. For a discussion of incentive-compatible
demand revelation, see Groves and Ledyard (1977) and Green and Laffont (1979). Farrel (1987)
uses a simple approach to show that an intervening bureaucrat may be more efficient than ne-
gotiations, even when the bureaucrat is limited by poor information and there are only two
agents.
2. Since A. C. Pigou's (1920) seminal contribution, the expression "Pigouvian taxes" has been
used for taxes intended to discourage activities with negative externalities. The "polluter pays"
principle, as defined by OECD guidelines (OECD 1975; Opschoor and Vos 1989) is not a "true"
Pigouvian instrument, because it usually requires payment for abatement but not for damages
from emissions.
3. The perspective of this article is policy analysis under a well-defined welfare objective. In
this context, insights from public choice models are relevant mostly for problems of policy adop-
tion and implementation. See Buchanan and Tullock (1975) and Hahn (1989) for applications to
environmental policies.
4. This calculation is a rough approximation using Roy's identity and a money measure of
utility.

GunnarS. Eskelandand EmmanuelJimenez 165

References
The word "processed" describes informally reproduced works that may not be commonly
available through libraries.

Anderson, Dennis. 1990. "Environmental Policy and the Public Revenue in Developing Coun-
tries." Environment Department Working Paper 36. World Bank, Washington, D.C. Processed.
Atkinson, Scott E., and T. H. Tietenberg. 1982. "The Empirical Properties of Two Classes of
Designs for Transferable Discharge Permit Markets." Journal of Environmental Economics
and Management 9: 101-21.
Asch, Peter, and Joseph J. Seneca. 1978. "Some Evidence on the Distribution of Air Quality."
Land Economics 54: 278-97.
Balcer, Yves. 1980. "Taxation of Externalities, Direct Versus Indirect." Journal of Public Eco-
nomics 13: 121-29.
Ballard, C. L., J. B. Shoven, and John Whalley. 1985. "General Equilibrium Computations of the
Marginal Welfare Costs of Taxes in the United States." American Economic Review 75: 128-38.
Bawa, V. S. 1988. "On Optimal Pollution Costs of Policies." In Baumol and Oates 1988.
Baumol, William J., and Wallace E. Oates. 1979. Economics, Environmental Policy and the Qual-
ity of Life. Englewood Cliffs, N.J.: Prentice Hall.
. 1988. The Theory of Environmental Policy. Cambridge and New York: Cambridge Uni-
versity Press.
Bernstein, Janis D. 1991. "Alternative Approaches to Pollution Control and Waste Management:
Regulatory and Economic Instruments." Urban Management Program Discussion Paper 3.
World Bank, Washington, D.C. Processed.
Binswanger, Hans P. 1989. "Brazilian Policies That Encourage Deforestation in the Amazon."
Environment Department Working Paper 16. World Bank, Washington, D.C. Processed.
Bohm, Peter. 1981. "Deposit-Refund Systems: Theory and Applications to Environmental, Con-
servation, and Consumer Policy." Published for Resources for the Future. Baltimore and
London: The Johns Hopkins University Press.
Bohm, Peter, and C. S. Russell. 1985. "Comparative Analysis of Alternative Policy Instruments."
In Allen V. Kneese and James L. Sweeney, eds., Handbook of Natural Resource and Energy
Economics. Vol. I. Amsterdam: North Holland.
Buchanan, James M. 1969. "External Diseconomies, Corrective Taxes and Market Structure."
American Economic Review 59, no. 7 (March): 174-77.
Buchanan, James M., and Gordon Tullock. 1975. "Polluters' Profit and Political Response: Direct
Controls Versus Taxes." American Economic Review 59, no. 7 (March): 139-47.
Christiansen, G. B., and T. H. Tietenberg. 1985. "Distributional and Macroeconomic Aspects of
Environmental Policy." In Allen V. Kneese and James L. Sweeney, eds., Handbook of Natural
Resource and Energy Economics. Vol I. Amsterdam: North Holland.
Coase, Ronald. 1960. "The Problems of Social Costs." Journal of Law and Economics 3: 1-44.
Dales, J. H. 1968. Pollution, Property and Prices. Toronto: University of Toronto Press.
Dewees, Donald N. 1983. "Instrument Choice in Environmental Policy." Economic Inquiry 21
(January): 53-71.
Dixon, J. A. 1991. "The Social Costs of Urban Pollution in Latin America: Incidence and Impli-
cations." World Bank, Latin America Technical Department, Washington, D.C. Processed.
Dixon, John, and G. Hodgson. 1988. "Logging Versus Fisheries and Tourism in Palawan: An
Environmental and Economic Analysis." EAPIOccasional Paper 7. East/West Center, Honolu-
lu, Hawaii. Processed.
Dorfman, N. S. 1975. "Who Will Pay for Pollution Control? The Distribution by Income of the
Burden of the National Environmental Protection Program, 1972-1980." National Tax Jour-
nal 28: 101-15.

166 The WorldBankResearchObserver,vol.7, no.2 (July1992)

Eskeland, Gunnar, and Emmanuel Jimenez. 1991. "Choosing Policy Instruments for Pollution
Control: A Review." Policy,Planning, and Research Working Paper 624. World Bank, Country
Economics Department, Washington, D.C. Processed.
Farrel, Joseph. 1987. "Information and the Coase Theorem." Journal of Economic Perspectives
1, no. 2: 113-29.
Forster, Bruce A. 1983. "The Backward Incidence of Pollution Control: A Dual Approach." Jour-
nal of Environmental Economics and Management 11: 1-17.
Frech, H. E., III. 1973. "Pricing of Pollution: The Coase Theorem in the Long Run." The Bell
Journal of Economics and Management Science 4, no. 1: 316-19.
Gianessi, L. P., Henry M. Peskin, and Edward Wolff. 1979. "The Distributional Effects of Uni-
form Air Pollution Policy in the United States." Quarterly Journal 91: 654-74.
Gil Diaz, Francisco. 1987. "Debt Accumulation and Distorted Growth through Subsidized Public
Sector Prices." Banco de Mexico. Processed.
Gollop, Frank M., and Mark J. Roberts. 1983. "Environmental Regulations and Productivity
Growth: The Case of Fossil-Fueled Electric Power Generation." Journal of Political Economy
91, no. 4: 654-74.
Green, Jerry, and J. J. Laffont. 1979. Individual Incentives in Public Decision Making. Amster-
dam: North Holland.
Groves, Theodore, and John 0. Ledyard. 1977. "Optimal Allocation of Public Goods: a Solution
to the 'Free Rider' Problem." Econometrica 4S: 783-809.
Hahn, Robert W 1989. "Economic Prescriptions for Environmental Problems: How the Patient
Followed the Doctor's Orders." Journal of Economic Perspectives 3, no. 2 (Spring): 95-114.
Hamrin, Robert. 1991. "Monitoring and Enforcement of Pollution Control-Lessons from the
U.S. Experience." World Bank, Country Economics Department, Washington, D.C. Processed.
Harrison, David, Jr. 1975. Who Pays for Clean Air: The Cost and Benefit Distribution of Auto-
mobile Emission Standards. Cambridge, Mass.: Ballinger.
Hertzman, Clyde. 1990. "Poland: Health and Environment in the Context of Socioeconomic
Decline." Internal Document 13. University of British Columbia, Vancouver. Processed.
Jack, W. G. 1990. "Power Sharing and Pollution Control: Coordinating Policies among Levels of
Government." Policy, Planning, and Research Working Paper 887. World Bank, Country Eco-
nomics Department, Washington, D.C. Processed.
Johansson, Per-Olov. 1987. The Economic Theory and Measurement of Environmental Benefits.
Cambridge: Cambridge University Press.
Kohn, Robert E. 1988. "Efficient Scale of the Pollution-Abating Firm." Land Economics 64, no. 1
(February): 53-61.
Kolstad, Charles. 1986. "Empirical Properties of Economic Incentives and Command-and-
Control Regulations for Air Pollution Control." Land Economics 62, no. 3 (August): 250-68.
Kolstad, C. D., T. S. Ulen, and G. [Link]. 1990. "Ex Post Liability for Harm Versus Ex Ante
Safety Regulation: Substitutes or Complements?" American Economic Review 80 (September):
888-901.
Lyon, Randolph M. 1989. "Transferable Discharge Permit Systems and Environmental Manage-
ment in Developing Countries." World Development 17, no. 8: 1299-1312.
Mahar, Dennis J. 1989. Government Policies and Deforestation in Brazil's Amazon Region.
Washington, D.C.: World Bank.
Margulis, Sergio. 1992. "Back-of-the-Envelope Estimates of Environmental Damage Costs in
Mexico." Working Paper 824. World Bank, Latin America and the Caribbean-Country De-
partment II, Agriculture Operations Division, Washington, D.C. Processed.
McCleary, William. 1991. "The Earmarking of Government Revenue: A Review of Some World
Bank Experience." World Bank Research Observer 6, no. 1: 81-104.
OECD (Organization for Economic Cooperation and Development). 1975. The Polluter Pays Prin-
ciple. Paris.

GunnarS. Eskelandand EmmanuelJimenez 167

Opschoor, J. P., and Hans Vos. 1989. The Application of Economic Instruments for Environ-
mental Protection in OECD Member Countries. Paris: Organization for Economic Cooperation
and Development.
Pearson, M., and Smith S. 1990. "Taxation and Environmental Policy: Some Empirical Evi-
dence." Commentary 19. Institute for Fiscal Studies, London. Processed.
Pigou, A. C. 1920. The Economics of Welfare. London: Macmillan.
Plourde, Charles, and David Yeung. 1989. "A Model of Industrial Pollution in a Stochastic En-
vironment." Journal of Environmental Economics and Management 16: 97-105.
Posner, Richard A. 1986. Economic Analysis of Law. 3rd ed. Boston: Little, Brown.
Repetto, Robert, and Malcolm Gillis, eds. 1988. Public Policies and the Misuse of Forest Re-
sources. Cambridge and New York: Cambridge University Press, with World Resources Insti-
tute.
Ringleb, Al H., and Steven N. Wiggins. 1990. "Liability and Large-Scale, Long-Term Hazards."
Journal of Political Economy 98, no. 3: 574-95.
Roberts, Marc J., and Michael Spence. 1976. "Effluent Charges and Licenses under Uncertainty."
Journal of Public Economics 5: 193-208.
Russell, Clifford S. 1990. "Monitoring and Enforcement." In Paul R. Portney, ed., Public Policies
for Environmental Protection, Washington, D.C.: Resources for the Future.
Sandmo, Agnar. 1975. "Optimal Taxation in the Presence of Externalities." Scandinavian Journal
of Economics 77: 86-98.
. 1976. "Direct versus Indirect Pigouvian Taxation." European Economic Review 7:
337-49.
Schramm, Gunter, and Jeremy J. Warford, eds. 1989. Environmental Management and Economic
Development. Baltimore: Johns Hopkins University Press.
Sebastian, lona. Forthcoming. "Issues in Urban Air Pollution: Mexico City Diagnostic Study."
Environment Department Working Paper. World Bank, Washington, D.C. Processed.
Seskin, Eugene P., Robert J. Anderson, Jr., and Robert 0. Reid. 1983. "An Empirical Analysis
of Economic Strategies for Controlling Air Pollution." Journal of Environmental Economics
and Management 10: 112-24.
Spulber, Daniel E 1985. "Effluent Regulation and Long-Run Optimality." Journal of Environ-
mental Economics and Management 12: 103-16.
Thomas, Vinod. 1981. Pollution Control in Sao Paulo, Brazil: Costs, Benefits and Effects on In-
dustrial Location. Staff Working Paper 501. Washington, D.C.: World Bank.
. 1985. "Evaluating Pollution Control: The Case of Sao Paulo, Brazil." Journal of Devel-
opment Economics 19: 133-46.
Tietenberg, T. H. 1988a. Emissions Trading: An Exercise in Reforming Pollution Policy.
Washington, D.C.: Resources for the Future.
. 1988b. Environmental and Natural Resource Economics. 2nd ed. Glenview, Ill.: Scott,
Foresman.
Tybout, R. A. 1972. "Pricing Pollution and Other Negative Externalities." The Bell Journal of
Economics and Management Science 3, no. 1 (Spring): 252-66.
.1973. "Pricing of Pollution: Reply." The Bell Journal of Economics and Management
Science 4, no. 7: 320-21.
U.S. Congress. 1990. Clean Air Act Amendments. Report S.1630. 101st Cong., 2nd sess.
Weitzman, Martin L. 1974. "Prices vs. Quantities." Review of Economic Studies 41: 477-91.
Whalley, John, and Randall Wigle. 1991. "Cutting CO2 Emissions: The Effects of Alternative
Policy Approaches." Energy Journal 12, no. 1 (January): 109-24.
Wiikander, Hans. 1985. "Correcting Externalities through Taxes on/Subsidies to Related
Goods." Journal of Public Economics 28: 111-25.
World Bank. 1991. Lessons of Tax Reform. Washington, D.C.

168 The World Bank Research Observer, vol. 7, no. 2 (July 1992)
Yan, Chiou-Shuang, Ah-Min Chung, Edward C. Kaziara, and Andrew G. Verzilli. 1975. "Air
Pollution Control Costs and Consumption Pattern Effects: A Regional Analysis." Journal of
Environmental Economics and Management 2: 60-68.
Yohe, Gary W. 1979. "The Backward Incidence of Pollution Control-Some Comparative Statics
in General Equilibrium." Journal of Environmental Economics and Management 6: 187-98.
Yu, E. S. H., and C. A. Ingene. 1982. "The Backward Incidence of Pollution Control in a Rigid
Wage Economy." Journal of Environmental Economics and Management 9: 304-10.

Gunnar S. Eskelandand EmmanuelJimenez 169