Environmental and Resource Economics 28: 395–408, 2004.
2004 Kluwer Academic Publishers. Printed in the Netherlands.
395
Optimal Environmental Charges/Taxes: Easy to
Estimate and Surplus-yielding
YEW-KWANG NG
Department of Economics, Monash University, Clayton 3800, Australia (E-mail: kwang.ng@
buseco.monash.edu.au)
Accepted 19 September 2003
Abstract. The estimation of the optimal charges/taxes on environmental disruption and the
financing of the spending on the abatement of environmental disruption are important
practical problems. This paper shows that, for most cases where some abatement is desirable,
both the estimation and the financing problems may be easily solved. It is desirable to charge
disruption (at least) at the marginal cost of abatement (which is easier to estimate than the
marginal damage of disruption) and such a charge will normally yield total revenue in excess
of the amount of abatement spending.
Key words: abatement, damage estimation, double dividend, effluent charges, environmental
disruption, environmental protection, financing, optimal taxes, pollution
JEL classification: D62, H23, Q20
In environmental protection, two important issues are: 1. How to estimate
the optimal amounts of charges on environmental disruption? 2. How to
finance for the spending on the abatement of environmental disruption
(including the improvement of environmental quality and the prevention
from deterioration)? These may be briefly referred to as the estimation
and the financing problems. This paper shows that, for a very large class
of situations, both problems may be (at least relatively speaking) easily
solved.
As the problem of environmental disruption has become very serious and
even threatens our long-term survival and with the increase in our income
levels and advancement of our knowledge, there are more and more instances
where governments undertake measures and spend to abate environmental
disruption, to prevent serious deterioration, and even to restore environmental quality. In many cases, such abatement spending is clearly desirable
or even should be increased. If our health and/or survival is threatened, it is
clearly desirable to avoid such disasters even at great costs. This paper shows
that, for cases where some abatement spending is desirable, the optimal
charge on environmental disruption equals (at least) the marginal cost of
396
YEW-KWANG NG
abatement, which may be more objectively estimated. Moreover, the revenue
from the optimal environmental charge is likely to more than cover the
optimal amount of abatement spending. These two important points are
explained intuitively (Section 1), illustrated graphically (Section 2), and
demonstrated in a simple mathematical model (Appendix 1).1
1. The Intuition
While environmental quality may have different dimensions or aspects, for
simplicity, we suppose that it may be measured by a single composite index.
(Our analysis may also be used with respect to a particular environmental
aspect/variable, such as the amount of CFC or CO2 .) Similarly for environmental disruption. In fact, whatever decreases the environmental quality
(as measured by the composite index) by a unit may be defined as a unit of
disruption. If it is desirable to spend on abatement to prevent environmental
quality from deteriorating by too much by the disruption activities, it is
desirable to discourage (through a tax or charge) disruption by the marginal
cost of abatement (the costs, at the margin, of improving, or preventing from
deterioration, environmental quality by one unit through abatement spending). This is so since a decrease in disruption by one unit could save abatement spending by the marginal cost of abatement without changing the
environmental quality. Thus, unless a unit of disruption yields net marginal
benefits to the disrupters no less than this marginal cost of abatement, it is
socially undesirable to disrupt. However, assuming that the disrupters
maximize their self-interest, they will undertake disruption activities until the
net marginal benefits equals zero. Thus, to make the disrupters take account
of the social costs, a charge on disruption equaling the marginal cost of
abatement is socially efficient. (For cases where the disrupters may have some
conscience effect, i.e. they care about the social costs of disruption and hence
undertake less disruption than privately optimal in the absence of such
conscience effects, see Ng 2004, Section 7.4. It is shown that a tax/charge on
external costs like environmental disruption may yield benefits much higher
than the traditional triangle by relieving the bad conscience and that the
Coase theorem may be wrong even in the absence of any transaction costs,
because the conscience-relieving effect is not symmetrical.)
If we do spend on abatement but not by as much as the optimal amount, it
may be desirable to increase the disruption charge to an amount more than
the marginal cost of abatement. However, it is then also desirable to increase
the amount of abatement which may then increase the marginal cost of
abatement and the optimal charge on disruption to the same amount. In any
case, it is desirable to charge disruption by at least the marginal cost of
abatement whether the abatement spending is optimal or sub-optimal.
OPTIMAL ENVIRONMENTAL CHARGES
397
It should be emphasized that our argument is that it is desirable to charge
disruption by at least the prevailing marginal cost of abatement, not
necessarily the marginal cost of abatement at the optimal level. It is obvious
that, ignoring second-best issues, overall optimality requires the equality of
the marginal cost of abatement (undertaken by the government or the society), the marginal valuation of environmental quality, the marginal damage
of disruption (to the society), and the marginal benefits of disruption or the
marginal cost of reducing disruption (to the disrupters). Moreover, all these
values equal the optimal charge (assuming continuity in the relevant functions and that the optimality is achieved by charging disruption). Thus, if we
know the marginal cost of abatement at the optimal level, we also know
marginal damage of disruption and the marginal valuation of environmental
quality. Thus, if we require knowledge about the marginal cost of abatement
at the optimal level, there is no informational economy in our proposal.
The difficulties of knowing or even estimating marginal damage of disruption
or the marginal valuation of environmental quality to determine the optimal disruption charge are used by people against disruption charges. However, our proposal is that, even if it is not possible to estimate the marginal
damage of disruption, it is still desirable to charge disruption at least at the
prevailing marginal cost of abatement. Thus, given our argument, the
argument against charging disruption on the ground of the difficulties of
estimating the marginal damage of disruption simply is no longer valid. We
know that it is desirable to charge disruption at least at the prevailing
marginal cost of abatement and the latter is relatively easy to estimate. So let
us make an improvement by start charging disruption at the prevailing
marginal cost of abatement. Whether we should then go further to charge
disruption at a higher level closer to the optimal charge is then another
matter.
Given that disruption is charged uniformly at the prevailing marginal cost
of abatement and that the abatement spending is optimized, will the revenue
from the disruption charge be sufficient to cover the abatement spending?
Efficient abatement requires that we spend on those areas where the costs per
unit of environmental quality improvement (or prevention from deterioration) are low first before moving to the areas with higher and higher costs.
Thus, it is reasonable to assume that the marginal cost of abatement increases
or at least does not decrease with the amount of abatement undertaken.
(Where some fixed costs are involved, some range of decreasing average or
even marginal cost may apply at the very initial range of abatement, but
overall, non-decreasingness may be taken to be the rule. Moreover, we are
concerned with the long run where all costs are variable.) Given this reasonable assumption, the total amount of abatement spending is less than the
marginal cost of abatement times the (physical) amount of abatement
(measured by the effect on environmental quality). The revenue from our
398
YEW-KWANG NG
proposed disruption charge equals the marginal cost of abatement (=rate of
charge) times the amount of disruption (also measured by the effect on
environmental quality). Thus, the revenue is larger than the abatement
spending if (but not necessarily only if) the amount of disruption is larger
than the amount of abatement.
Now, the environment has some capability to clean itself. In the absence of
further disruption, the already disrupted environment will improve in quality
by itself, at least to some extent. The reason that environmental quality may
continue to deteriorate is due to continuing heavy disruption, usually even at
an increasing rate due to higher production and consumption. For any period,
suppose that, in the absence of disruption charges and abatement, the disruption in that period will decrease the environment quality by x units (or x%
however measured) in the absence of the self-maintaining ability of the
environment. Then the environment quality will only actually deteriorate by
x – y units due to the effect of self-maintenance. Thus, if we want to keep the
environment from deteriorating or even to improve a little, we only have to
reduce disruption (by charges and abatement combined) by less than x units.
If the environment has been abused for many years without much abatement,
it is possible that for the very initial periods, it may be optimal to spend on
abatement by such a big amount that more than x units of disruption should
be abated. However, after such initial abnormal or emergency periods and
certainly for normal on-going periods, it is usually optimal to abate less than x
units of disruption. Thus, since the amount of disruption is normally larger
than the amount of abatement, the revenue from the optimal charge on disruption is larger than the optimal abatement spending, as follows from the last
sentence in the previous paragraph.
The discussion above ignores the second-best issue of pre-existing distortions in the economy. In fact, the appendix shows that, taking account of
the disincentive effect of the revenue-raising system (usually income or general consumption taxes) to finance for public expenditures, the charge on
disruption should be larger than the marginal cost of abatement as it then
yields a double dividend (of correcting for the disruption as well as reducing
the amount of the revenue-raising taxation). Our arguments may be summarized into the following proposition. (See the Appendix 1 for a more
rigorous demonstration.)
Proposition: Provided that some abatement is efficient or is being undertaken, it is efficient to charge disruption at a rate no less than the marginal cost of abatement. The revenue from this charge normally
exceeds the optimal amount of abatement spending. The rate of charge
should be even higher if it yields a ‘‘double dividend’’ through the disincentive effect of the normal revenue-raising system to finance for public
expenditures.
399
OPTIMAL ENVIRONMENTAL CHARGES
$
MB
MCA
L
MV’
MCB
M
J
I
Q
G
H
MV
E
C
A
D
N
F
B
Figure 1. Optimal charge exceeds optimal abatement spending.
Our two basic points are illustrated graphically in the next section
where the second-best (double dividend) issue (discussed in the appendix) is
ignored.
2. Graphical Analysis
Denote our measure of environmental quality as E. The amounts of both
disruption and abatement are measured with respect to the effects on E,
measured along the horizontal axis in Figure 1. Let the value of environmental quality at the beginning of the period under analysis be at A. Without
further disruption, the environment would clean itself and settle at the point
B at the end of this period or the beginning of next period. However, production and consumption activities generate disruption (measured leftward
from the point B). In the absence of disruption charges and abatement
spending, disruption will be undertaken until C where the (net) marginal
benefit of disruption (to the disrupters) is zero. This marginal benefit of
disruption (the curve MB) is measured using the point B as origin and with
the amount of disruption measured leftward from B. It is associated with the
gain of consumers from being able to consume goods produced at low costs
plus the gain of producers (if the low costs are not fully passed on as low
prices) if producers/consumers are free to use the lowest cost (not counting
disruption costs) methods of production/consumption. Since some con-
400
YEW-KWANG NG
sumption is necessary for survival and since some disruption is unavoidable
in production and consumption, the marginal benefit of disruption (i.e. the
height of the curve MB) is very high at low level of disruption (close to point
B). It decreases to reach zero at the point C where producers/consumers
derive no further (net) benefits from disruption.
At the point C, if the marginal cost of abatement by the government or
society (CG) is lower than the marginal valuation of environmental quality
(CL), it is desirable to undertake some abatement. In the absence of disruption charges, the optimal amount of abatement is determined by the
intersection of the marginal cost of abatement MCA (where the superscript A
indicates abatement only) with the marginal valuation of environmental
quality MV at M, giving the optimal amount of abatement CN and optimal
amount of abatement spending GCNM. However, if an optimal charge is
placed on disruption, disruption (measured leftward from B) will be reduced
from BC to BD, making the relevant marginal cost of abatement MCB
(where the superscript B indicates that both disruption charges and abatement are used; MCB may not or may be a horizontal displacement of MCA
depending on whether the disruption charges and the corresponding reduction in disruption affect the marginal cost of abatement or not). The intersection of MCB and MV at J determines the optimal rate of disruption charge
FJ (=DI), the amount of discouraged (by the charge) disruption DC, the
optimal amount of abatement DF, and the optimal amount of abatement
spending HDFJ. The reason for the above mentioned determination of the
optimal disruption charge and optimal abatement is due to the fact that
optimality requires both the marginal cost of abatement (FJ at the optimal
solution) and the marginal benefit of disruption (DI or BQ at the optimal
solution;) to be equal to the marginal value of environmental quality (FJ at
the optimal solution). It may also be noted that the marginal benefit (to the
disrupters) of disruption also equals the marginal cost of reducing disruption
on the part of the disrupters. Assuming self-interest behaviour, this also
equals the marginal charge on disruption. In other words, the disrupters
undertake disruption until the marginal (net) benefit is equal to the marginal
charge (additional charge per unit of disruption).
The simultaneous determination of the amount of optimal abatement and
optimal charge on disruption still requires the estimation of the marginal
valuation of environmental quality. The point here is that, even if we have no
reliable estimate of this at all, we may still charge disruption at least at the
marginal cost of abatement, provided that some abatement is desirable and/
or that some abatement is being undertaken. For example, suppose that we
do not know the MV curve which may well be MV’ instead of MV. We
undertake DF amount of abatement which may well be sub-optimal. However, it is still desirable to charge disruption (at least) at the rate equaling the
marginal cost of abatement FJ (=DI). It may be desirable to charge at even a
OPTIMAL ENVIRONMENTAL CHARGES
401
higher rate, but charging at FJ is better than charging at a lower rate or not
charging at all. Thus, if we find it too difficult to estimate the MV of environmental quality, we know at least that it is desirable to charge disruption at
the rate no lower than the marginal cost of abatement, which can be more
objectively estimated. It is still open to environmentalists to argue that both
the existing abatement and the disruption charge should be increased. On the
other hand, in the less likely case where the amount of abatement is excessive
(the MC of abatement is larger than the MV of environmental quality), it is
still desirable (in comparison to no charge) to charge disruption at the marginal cost of abatement and reduce abatement by the amount of discouraged
disruption. (Due to the global public good and long-term nature of environmental quality, decisions by myopic national governments are extremely
unlikely to make this excessive abatement case relevant.) Thus, provided that
either if some abatement is desirable or if some abatement is being undertaken, it is always desirable to charge disruption at least at the marginal cost
of abatement. Whether it is desirable to decrease or increase abatement depends on whether the amount of abatement is excessive or deficient.
Our second point that the revenue from the disruption charge normally
exceeds the abatement spending may also be seen in Figure 1. With DF
amount of abatement, the total abatement spending is HDFJ. With the rate
of disruption charge being equal to the marginal cost of abatement FJ, the
amount of revenue from the disruption charge equals IDBQ which is larger
than the abatement spending HDFJ if, but not only if, the amount of disruption BD is larger than the amount of abatement DF. To see that this is
normally the case, note that, in the case illustrated in Figure 1 where BD is
significantly larger than DF, the final value of environmental quality (at the
end of the period under analysis or the beginning of next period) is at the
point F which is higher than that at A, the corresponding value of last period.
If (but not only if) the amount of improvement in environmental quality is
not significantly larger than the amount of environmental self-cleaning AB,
the revenue from the disruption charge will be larger than the abatement
spending. (This is so since abatement–disruption = improvement–selfcleaning.) As noted in the previous section, this may be expected to be the
case except possibly during emergency abatement after a long period of
environmental abuse.
3. Concluding Remarks
This paper demonstrates two simple but important points. First, where some
abatement is desirable or is being undertaken (widely applicable) and where it
is difficult to estimate the marginal valuation of environmental quality (and
hence the marginal damage of environmental disruption, also widely appli-
402
YEW-KWANG NG
cable), it is desirable to charge disruption at least at the marginal cost of
abatement, which is easier to estimate. Secondly, the total revenue from this
disruption charge normally exceeds the total cost of abatement. Both
these simple points have significant implications on the practicability of
environmental protection.
Using a representative individual approach, our analysis does not address
the issues of income distribution and individual differences. Elsewhere (Ng
1979/1983, 1984), I have argued that, for any specific issue of economic policy
or cost-benefit analysis, we should follow the principle of ‘‘a dollar is a
dollar’’ i.e. concentrating on efficiency only, without regard to distributional
effects, leaving the objective of equality to be achieved more efficiently
through the general tax/transfer system. This makes the ignoring of the
distributional issue less of a problem, to say the least. In any case, the distributional problem is a separate issue, which may be addressed separately if
desired.
Our analysis takes the case of a closed economy (or a world economy)
without the complication of a national government/country versus the rest of
the world. This complication is discussed in Ng and Liu (2004). (On the
economics of international environmental agreements, see papers collected in
Batabyal (2000).) While further considerations are generated by this complication, the basic message of this paper is not affected.
Appendix 1: Mathematical Appendix
We could consider a model in which there is a representative individual and a
representative firm, representing the consumption and production sides
respectively. This would be more realistic as much of environmental disruption is taken at the firm level. However, consumers also pollute and the
pollution (used interchangeably with ‘‘disruption’’) at the production level
may be included in a more general consumption/production sector, especially
where competition is assumed. Since the separate production decision and its
interaction with the consumption sector has been extensively analysed in
traditional analysis and is not the focus here, both sectors are included within
the single consumption/production sector here. This is done mainly for
simplicity and to focus on our basic points. We thus consider a model where
the utility of a representative individual U (assumed the same as the welfare
of that individual, for the case where there are divergences between the two,
see Ng 2003) is a function of her consumption c, leisure x, public goods G,
and environmental quality E. Taking G and E as beyond her control, the
individual maximizes with respect to leisure x and pollution p
U ¼ Uðc; x; G; EÞ
ðA1Þ
OPTIMAL ENVIRONMENTAL CHARGES
403
subject to
c ¼ cðs; pÞ;
cs > 0; cp > 0
ðA2Þ
s ¼ ð1 tÞy fp
ðA3Þ
y ¼ ð1 xÞw
ðA4Þ
where c = effective consumption, s = financial consumption, y = pre-tax
income, w = wage-rate (taken as exogenous), t = (proportional) tax-rate,
f = per-unit charge or fine on pollution. (We only consider a fixed charge; for
the use of non-linear charges, see Lee and Kim 2000.) The amount of effective
consumption is taken as an increasing function of financial consumption and
pollution. This is a simplified device to incorporate both consumer pollution
and producer pollution within an inclusive consumption/production sector.
The first-order conditions are
Ux ¼ ð1 tÞwcs Uc
ðA5Þ
cp ¼ fcs
ðA6Þ
where a subscript denotes partial differentiation, e.g. Ux ” oU/ox. It may be
noted that cs Uc is the marginal utility of financial consumption and is
equivalent to the traditional marginal utility of consumption. It consists of
two parts because of the combination of the production and consumption
into one effective consumption sector c. (See Equation (A2) above.)
The government chooses its income tax-rate t, abatement-spending
proportion a, and pollution charge rate f to maximize (A1) subject to the
above maximization by individuals in the economy and subject to the following constraints
G ¼ ð1 aÞtY þ fp
ðA7Þ
A ¼ atY
ðA8Þ
E ¼ EðP; AÞ;
Ep < 0; EA > 0
P ¼ p; Y ¼ y; etc:
ðA9Þ
ðA10Þ
where G = spending on public goods, excluding environmental disruption
abatement spending A, which is a fraction a of the income-tax revenue. The
environmental quality E is an increasing function of abatement spending A
and a decreasing function of pollution P. As the number of individuals in
each country is given and not the focus here, it has been normalized at unity
for simplicity, but the addition of this given variable will not affect the
analysis below in any substantial way. (Requires replacing MRS by SMRS
for the public good G.) The equality of the small case variables at the individual level with those at the economy level (capital case) in (A10) also
follows from the representativeness and the above normalization. Also,
partly for simplicity and mainly due to our concern with the appropriate
404
YEW-KWANG NG
public policy, we ignore the possible divergence between the government
objective function from that of the representative individual, while apologizing to the public choice school. We also ignore the indirect effects of a
change in income tax-rate t on pollution p and a change in abatementspending ratio a on leisure x and pollution p. (However, the more direct
effects of t on x and pollution charge f on p are taken fully into account.)
These indirect second-best effects are not only usually of smaller order of
magnitude than the ones we are interested here, they would also just add
terms of indeterminate signs without affecting the main point of the present
analysis. (Mathematically, they would just make the second-best adjustment
factors mentioned below even more complicated.) Assuming an interior
solution and the satisfaction of the second-order conditions, the first-order
conditions with respect to t, a, and f are respectively,
½ð1 aÞY ð1 aÞtwðdx=dtÞUG þ ½aY atwðdx=dtÞEA UE þ ðdx=dtÞUx
¼ ½Y þ ð1 tÞwðdx=dtÞcs Uc
ðA11Þ
E A UE ¼ UG
ðA12Þ
½p þ fðdp=dfÞUG þ ðdp=dfÞEp UE ¼ fcs ½p þ fðdp=dfÞ cp ðdp=dfÞgUc
ðA13Þ
Equation (A11) specifies that the optimal income tax-rate t is achieved when
its marginal benefit through a higher provision of G and A, including its
indirect effect, if any, in affecting leisure x, is just balanced by its marginal
cost of reducing consumption. Equation (A12) is the condition for the
optimal division of the tax revenue between G and abatement spending A. It
specifies that the optimal abatement-spending ratio a is achieved when the
marginal gain of a higher abatement through its effect on the environment is
just offset by the loss of forgone public goods. Equation (A13) specifies that
the optimal charge f is achieved when its marginal benefits through a higher
provision of G and a lower disruption are just balanced by its marginal costs
of reducing consumption through the reduction of financial consumption s
and the reduction of disruption p. (Note that, while the effect on disruption is
dp/df, the effects on government revenue and private financial consumption
are p + f(dp/df) or the marginal revenue of increasing f.)
Substitute (A5) and (A12) into (A11) and divide through by Y, yielding
ð1 bÞUG ¼ cs Uc
ðA14Þ
where b ” tw(dx/dt)/y is a second-best adjustment factor to account for a
possible effect of a change in income tax-rate t on work/leisure; the factor tw
is present because a marginal increase in leisure x decreases the tax revenue
by tw. If we ignore this second-best factor b, (A14) is our version of the
OPTIMAL ENVIRONMENTAL CHARGES
405
Samuelsonian condition for the optimal supply of public goods G, requiring
the equation of the (aggregate) MRS of G to consumption (or the marginal
valuation of G, using consumption as the numeraire) be equal to the MRT
(marginal cost). The second best adjustment factor b further says that, if the
higher income tax-rate t to finance for more G and A causes a decrease/
increase in work, the marginal cost is correspondingly higher/lower. Economists traditionally emphasise this factor b, taking it to be positive and
significant, suggesting that public goods like G and A have to yield marginal
direct benefits significantly larger than their marginal costs for them to be
worthwhile. However, Kaplow (1996) and Ng (2000a) argue that, even if a
higher tax itself causes disincentive effects, the higher public goods supplied
tend to cause offsetting effects, making the combined effects largely negligible
(without considering income effects). Moreover, if income, consumption, or
expenditure taxes actually just offset the distortion of the relative-income or
conspicuous consumption effects, they may be corrective rather than causing
any distortion to begin with. (See Frank 1999; Ng 2000b.)
From (A12) and (A14), we also have
ð1 bÞEA UE ¼ cs Uc
ðA15Þ
which is the optimality condition for the supply of abatement, requiring that
the marginal value of abatement through improving environmental quality
be equal to the marginal cost of reducing consumption, adjusted by the
second-best factor b if necessary. (Any one of Equations A12, A14 and A15
may of course be derived from the other two.)
Substitute (A14) into (A13), yielding
EP UE ¼ cp Uc ½b=ð1 bÞð1 þ 1=gÞcs Uc
ðA16Þ
where g ” (dp/df)f/p is the elasticity (proportionate response) of pollution
with respect to the charge. Equation (A16) is the optimality condition for the
amount of pollution, specifying the equality of the marginal damage of
pollution with the marginal benefit (to the consumption/production sector;
note that cp is negative) plus a second-best adjustment factor if b is not zero.
Substituting cp from (A6) into (A16), we have
f ¼ EP UE =fcs Uc ½1 þ ð1 þ 1=gÞb=ð1 bÞg
ðA17Þ
as the condition for the optimal fine on pollution. However, even apart from
the second-best adjustment factor in the big brackets, this traditional condition depends not only on how pollution affects environmental quality (EP)
but also on the subjective marginal valuation of environmental quality (UE)
which is more difficult to estimate, especially with high uncertainty and longlasting effects involving the future generations. Substitute UE from (A15) into
(A17), yielding
406
YEW-KWANG NG
f ¼ EP =fEA ð1 þ b=gÞg
ðA18Þ
which are very simple and is in terms of the marginal rate of technical substitution rather than subjective substitution and hence easier to estimate.
Ignoring the second-best adjustment factor, (A18) means that the optimal
pollution charge equals the (absolute) marginal rate of technical substitution
between pollution and abatement spending in maintaining a given level of
environmental quality. This means that a unit of pollution (or any other form
of environmental disruption) should be charged at how much it costs at the
margin to clear up pollution (disruption) by one unit. The rationale of this
follows from the (usually compelling) assumption that some abatement is
desirable (the assumption of an interior solution here). If it costs $x at the
margin to to improve environmental quality by one unit, it is desirable to
discourage disruption by imposing a per-unit charge of $x. This means that
the tricky problem of how to estimate the optimal charge on environmental
disruption can be relatively easily solved. Instead of undertaking the very
difficult estimation of the marginal damages of disruption (the marginal effect
of disruption on environmental quality times the social marginal valuation of
environmental quality) as required by (A17), the optimal charge may simply
be determined at the marginal costs of abatement. Any activity that causes a
decrease in environmental quality by a marginal unit (however measured)
should be charged at the marginal costs of improving the environmental
quality by the same one (same in measurement, but not necessarily the
identical) unit. Moreover, this principle may be made more specific. If it is
desirable to spend to reduce CO2, any activity that increases CO2 by one unit
should be charged at the marginal costs of reducing CO2 by one unit; similarly for other forms of disruption.
Since g is negative, the second-best adjustment factor b in (A18), if positive
(i.e. if income taxation causes disincentive effects) as traditionally believed,
suggests that the optimal charge on disruption should be even higher than
envisaged in the above first-best analysis, if (but not only if) the values of EP/
EA and g are not significantly affected by b or if, as are likely, the secondary
effects of a higher b are not more than offsetting to its primary effects. This is
our version of the double dividend. (The size and even the sign of this double
dividend may change depending on whether we focus on employment or
welfare, on short or long runs, on the definitions of second-best optimal tax,
with a closed or open economy, with separable or non-separable utility
functions, with the environment as a factor of production or not, etc., see,
e.g. Felder and Schleiniger 1995; Schob 1997; Bovenberg 1998; Kahn and
Farmer 1999; Bosquet 2000; de Mooij 2000; Goodstein 2002; Parry and
Bento 2000; Parry and Oates 2000; Schwartz and Repetto 2000; Bosello et al.
2001.)
OPTIMAL ENVIRONMENTAL CHARGES
407
Acknowledgement
I am grateful to an anonymous referee for this journal for helpful comments.
Note
1. Parts of our results may be deduced from other analyses of similar problems (e.g. Bovenberg
& Ploeg 1994) but neither of the two important points have been noted. Our argument also
differs from that of Bimonte (1999) who argues for the use of the marginal benefit to the
polluters as an estimate of the optimal charge and proposes using iteration to reach the
optimal value. In contrast, our proposal here is using the prevailing marginal cost of
abatement (i.e. reducing pollution/disruption) to the society as the lower bound for the
disruption charge.
References
Batabyal, A. A., ed. (2000), The Economics of International Environmental Agreements. Aldershot: Ashgate.
Bimonte, S. (1999), ‘An algorithm for optimal Pigouvian taxes without benefit data’, Environmental and Resource Economics 13, 1–11.
Bosello, F., C. Carraro and M. Galeotti, (2001), ‘The double dividend issue: Modeling
strategies and empirical findings’, Environment & Development Economics 6(1), 9–45.
Bosquet, B. (2000), ‘Environmental tax reform: Does it work? A survey of the empirical
evidence’, Ecological Economics 34(1), 19–32.
Bovenberg, A. L. (1998), ‘Green tax reforms: Implications for welfare and distribution’, Swiss
Journal of Economics and Statistics 134(3), 271–95.
Bovenberg, A. L. and Ploeg, F. van der (1994), ‘Environmental policy, public finance and the
labor market in a second-best world’, Journal of Public Economics 55, 349–390.
De Mooij, R. A. (2000), Environmental Taxation and the Double Dividend. Amsterdam: NorthHolland.
Felder, S. and R. Schleiniger, (1995), ‘Domestic environmental policy and international factor
mobility: A general equilibrium analysis’, Swiss Journal of Economics and Statistics 131(3),
547–558.
Frank, R. H. (1999), Luxury Fever: Why Money Fails to Satisfy in an Era of Excess. New
York: The Free Press.
Goodstein, Eban (2002), ‘Labor supply and the double-dividend’, Ecological Economicsm
42(1–2), 101–106.
Kahn, J. R. & A. Farmer (1999), ‘The double dividend, second-best worlds, and real-world
environmental policy’, Ecological Economics 30(3), 433–439.
Kaplow, L. (1996), ‘The optimal supply of public goods and the distortionary cost of taxation’, National Tax Journal 49(4), 513–533.
Lee, S. Ho and I. KIM (2000), ‘Self-selection and optimal nonlinear effluent charges’, Environmental & Resource Economics 16, 1–14.
Ng, Y. -K. (1979/1983), Welfare Economics: Introduction and Development of Basic Concepts.
Basingstoke, Hampshire, U.K.: Macmillan.
Ng, Y. -K. (1984), ‘Quasi-Pareto social improvements’, American Economic Review, December,
74(5), 1033–1050.
408
YEW-KWANG NG
Ng, Y. -K. (2000a), ‘The optimal size of public spending and the distortionary costs of
taxation’, National Tax Journal 52(2), 253–272.
Ng, Y. -K. (2000b), Efficiency, Equality, and public Policy: With a case for Higher Public
Spending, Basingstoke, Hampshire: Macmillan.
Ng, Y. -K. (2003), ‘From preference to happiness: Towards a more complete welfare economics’, Social Choice and Welfare 20, 307–350.
Ng, Y. -K. (2004), Welfare Economics: Towards a More Complete Analysis. London: Macmillan/Palgrave.
Ng, Y. -K. and B. D. Liu (2004), ‘Global environmental protection: Solving the international
public goods problem by empowering the United Nations through cooperation with
WTO’, International Journal of Global Environmental Issues 3(4), 409–417.
Parry, I. W. H. and A. M. Bento (2000), ‘Tax deductions, environmental policy, and the
‘‘double dividend’’ hypothesis’, Journal of Environmental Economics & Management 39(1),
67–96.
Parry, I. W. H. & W. E. Oates (2000), ‘Policy Analysis in the Presence of Distorting Taxes’,
Journal of Policy Analysis and Management 19(4), 603–613.
Schob, R. (1997), ‘Environmental taxes and pre-existing distortions: The normalization trap’,
International Tax and Public finance 4(2), 167–176.
Schwartz, Je. & R. Repetto (2000), ‘Nonseparable utility and the double dividend debate’,
Environmental & Resource Economics 15(2), 149–157.