ham et al., 1995. The cost-effectiveness analysis evades the problem of monetarising ecological
effects, measuring the emission reduction in one physical unit e.g. tons of nitrogen per year. As a
consequence, the cost-effectiveness analysis allows a relative rating of the measures examined only.
Furthermore, since different pollutants cannot sensibly be aggregated in physical terms the dam-
age of a ton of CO
2
-emission is not equal to the damage resulting from the emission of a ton of
nitrate, the traditional cost-effectiveness analysis is restricted to one — hopefully — major pollutant
and ignores all the other ecological effects. Studies in this field have been carried out for a variety of
pollutants in different regions, for example, Ma¨ler and Olsson 1990 for sulphur in Europe, Krup-
nick and Walls 1992 for ozone in American cities, Johnsen 1993 for phosphorus and Koop-
man 1995 for CO
2
-emissions in Europe. Gren 1993, Zylicz 1995 additionally consider the im-
pact that different locations of emission sources have on environmental quality. This study is fo-
cused on another issue. In trying to overcome the drawbacks of valuing the invaluable, or of being
limited to considering the emissions of one pollu- tant only, a comprehensive cost-effectiveness anal-
ysis is proposed. In this analysis, effectiveness is measured in terms of reduction of different pollu-
tants aggregated to a damage index, while all the other effects apart from pollutant reduction are
monetarised and expressed as cost. Similar ap- proaches are already applied in health economics,
where effectiveness is measured as a health index, for example, as quality-adjusted life-years Tor-
rance, 1986.
2
With this method, it is intended to possibly include all economic and ecological consequences
in order to derive concrete results to evaluate environmental policies. The goal of the analysis is
to provide the government with information on the comprehensive cost-effectiveness of measures
to reduce nitrogen emissions. This allows the deci- sion maker to choose measures such that a given
environmental standard can be reached at mini- mal cost.
The next section discusses methodical questions and explains the aggregation of various pollutants
to an index. In particular, the problems of assess- ing greenhouse gas emissions and weighing pollu-
tants with different geographical ranges are addressed. In Section 3, the cost-effectiveness
analysis is applied to a range of pollution reduc- tion measures which are at present under political
discussion in Switzerland. These measures are de- scribed briefly and the results are presented in two
stages. While the first stage ignores the interde- pendencies of different measures, the second stage
considers the sequence of introducing the mea- sures according to their cost-effectiveness. Section
4 concludes.
2. Methodical considerations
Particularly in Europe, the aggregation of dif- ferent ecological effects to an index has developed
into a specific branch of research. Not unexpect- edly, this research has not resulted in a single
method that is accepted ubiquitously.
3
It is not the purpose of this paper to discuss the various
propositions and their relative merits at length. Instead, one method in particular will be consid-
ered, one that has been developed in Switzerland Ahbe et al., 1990 and is currently also employed
in Scandinavia, the Netherlands, Austria and Japan. The so called UBP-method, from the Ger-
man ‘Umwelt-Belastungs-Punkte’ Environmental Impact Points, is adapted to the specific question
of evaluating measures of reducing nitrogen emis- sions. These adaptations refer to the choice of
critical flows, the inclusion of different greenhouse gas emissions and the distinction between nation-
ally and internationally harmful emissions.
The determination of cost, on the other hand, is methodically straightforward and based on the
opportunity cost concept. Since all effects apart from the pollutant emission reduction must be
expressed as cost, other external cost savings, for example, noise reduction of traffic measures, need
to be monetarised.
2
In this context, Torrance uses the terminology of cost-util- ity analysis.
3
For a survey on various valuation methods, see Braun- schweig et al. 1996, Lindeijer 1996.
Table 1 Annual actual and critical flows of pollutants in year 2002
a
Pollutant Critical flow 1000 t
Actual flow 1000 t Marginal damage index points per 1000
year
− 1
t year
− 1
11 Nitrogen oxide
b
0.34 41
Ammonia 54
25 0.086
Nitrogen into lakes and 1000
500 0.14
rivers
c
16 Nitrogen into ground water
0.11 29
Volatile organic compounds 172
65 0.041
25 Sulphur dioxide
0.050 32
1 3
2.9 Phosphorus
115 Chemical oxygen demand
120 0.0080
13 500 000 Carbon dioxide equivalent
0.00011 27 000 000
d a
According to the considerations on the affected areas, the flows of carbon dioxide equivalent refer to the whole world, the flows of nitrogen into lakes and rivers to the North Sea and all the other pollutant flows to Switzerland. If not indicated otherwise, all
flows are from the Swiss Federal Office for Environment, Forests and Landscape, 1996.
b
Measured as tons of nitrogen.
c
Oslo and Paris Commissions 1993.
d
Forecasting precisely the actual flow of global carbon dioxide equivalents in the year 2002 depends decisively on the global measures undertaken to curb greenhouse gas emissions until then. In this study, we depart from forecasts of the Swiss flow and
assume a share of 0.2 of world-wide emissions. This is a rather crude assumption but the results do not react sensitively to it. Apart from that, specific measures of other countries to reduce carbon dioxide emissions — and nitrogen emissions into lake and
rivers — are not taken into account.
2
.
1
. Effecti6eness
:
the UBP-method The basic idea of the UBP-method is to nor-
malise quantities of different pollutants with their respective critical flows, assigning to pollutants
with lower critical flows a higher weight see the first term of the r.h.s. of Eq. 1. This weighting
is motivated by the close connection between the environmental impact and the critical flow of a
pollutant. Low critical flows express high environ- mental damage per quantity and vice versa.
Compare, for example, the critical flows of nitro- gen into lakes and rivers and phosphorus in Table
1: the higher environmental damage per quantity phosphorus is reflected in a lower critical flow.
Since the critical flows serve as an indicator for environmental impact, the particular choice of the
flows is decisive when applying the method.
In the original UBP-method, critical flows are derived from quality standards that are politically
set, e.g. the maximum allowed concentration of nitrogen oxides in the air. This approach has the
major drawback that politically set environmental standards often include cost considerations and
do not express ecological minimum standards. When an index based on these political standards
is used to calculate the cost-effectiveness, the eco- nomic cost aspect is counted twice, appearing in
the nominator cost as well as in the denominator effectiveness of the result. Therefore, in this
paper critical flows derived from ecological con- siderations are used, i.e. maximum flows of pollu-
tants
that — with the
current knowledge
available — will not disturb the ecological equi- librium. These critical flows have been determined
by experts of the Swiss Federal Office for Envi- ronment, Forests and Landscape, and are pub-
lished in Swiss Federal Office for Environment, Forests and Landscapes 1996. The critical flows
take into account damages from nitrogen oxides, ground layer ozone, acid soils and waters as well
as eutrophication. It must be noted that these flows refer to a nationwide average and do not
take into consideration local ecological condi- tions. For a spatially differentiated environmental
policy, such a local determination of critical flows would be helpful.
Fig. 1. Marginal damage function of nitrogen oxide.
A second step, constructing the pollution index, is based on the assumption that the marginal
damage of a pollutant is linear in the actual flow see the second term of the r.h.s. of Eq. 1.
While the idea of increasing marginal damage is familiar to economists, the linearity assumption is
admittedly arbitrary. But as long as there is no clear evidence for other functional forms, it seems
sensible to choose the simple function.
4
With these two valuation steps, the relative marginal damage of a pollutant MD
i
can be formulated:
MD
i
= 1
Fc
i
F
i
Fc
i
1 where Fc
i
is the critical flow of pollutant i and F
i
is the actual flow of pollutant i. The benefit of reducing emissions of a pollutant
B
i
by DF
i
can then be calculated as an integral: B
i
= −
F
i
− DF
i
F
i
1 Fc
i
F
i
Fc
i
dF
i
. 2
To illustrate the pollution index, the marginal damage function for nitrogen oxide is represented
in Fig. 1. With an actual yearly flow of 41 000 t and a critical flow of 11 000 t, the marginal dam-
age per thousand tons of nitrogen oxide is 0.34 see also Table 1. According to formula 2, an
emission reduction of DF yields a benefit the size of the shaded area in Fig. 1.
Fig. 1 also shows that the marginal damage below the critical flow is small but not zero. Since
the precise determination of the critical flows is very difficult and depends on the ecological
knowledge currently available, it seems sensible to assign emission reductions below the critical val-
ues a small but non zero benefit
5
and to choose an intercept of zero.
Note that the slope of the marginal damage function does not influence the results, since any
linear transformation of the function leads to the same index ranking. Choosing the same slope for
all pollutants on the other hand implies an equal marginal damage of a normalised quantity at the
critical flow.
Finally, to calculate the environmental benefit of all pollutant reductions E, the pollutant spe-
cific benefits B
i
are summed up over all pollu- tants. This yields the effectiveness of a measure to
5
A sensitivity analysis with zero marginal damage below the critical flow only changes the results marginally. In particular,
the relative cost-effectiveness of the evaluated measures does not change. Note also, that only the actual flow of chemical
oxygen demand is below its critical flow.
4
An empirical discussion of the linearity assumption is given by Mu¨ller-Wenk, examining the shape of damage func-
tions for some impact categories, in Braunschweig et al. 1996.
be used when assessing the comprehensive cost- effectiveness.
E =
i
B
i
= −
i F
i
− DF
i
F
i
1 Fc
i
F
i
Fc
i
dF
i
. 3
The summing up as presented in Eq. 3 allows for a trade-off between different pollutants caus-
ing damage but takes no account of possible synergies or antagonies. Thus, the damage of one
pollutant is assessed independently of the emis- sion level of other pollutants. With more knowl-
edge
about such
interdependencies, more
complicated indexes could be applied on the same methodical basis.
6
2
.
2
. Greenhouse gas emissions and their effecti6e range
The political measures analysed in the present study
reduce — besides other
pollutants — the emissions of three greenhouse gases: carbon diox-
ide CO
2
, methane CH
4
and laughing gas N
2
O. Since these three substances all contribute to the same ecological problem of increasing the
greenhouse gas effect, it makes no sense to deter- mine critical flows for each substance. Instead, the
substances are aggregated to carbon dioxide equivalents according to their global warming po-
tential in a time scale of 100 years, i.e. CO
2
= 1,
CH
4
= 11 and N
2
O = 270 IPCC, cited in Swiss Confederation, 1994, p. 15. Then a critical flow
of carbon dioxide equivalents is chosen. Since no definite critical flow can yet be quantified, the
minimum proposition of the IPCC cited in Pro- Clim, 1996, p. 12 is adopted to half the present
global amount of annual greenhouse gases release. Additionally, a sensitivity analysis with lower crit-
ical flows is undertaken.
On evaluating the relative damage of pollutant emissions, the further problem arises that differ-
ent pollutants affect a different range of the popu- lation. While, for example, nitrate emissions into
ground water in Switzerland do not cross national boundaries, nitrate emissions into lakes and rivers
pollute the North Sea and hence affect the popu- lation of the neighboring states too.
7
Greenhouse gas emissions, on the other hand, have global
consequences regardless of the origin of emission. In taking into account these different effective
ranges it must be remembered that the considered environmental effects are all of the public bad
type — meaning that any person is affected by pollution irrespective of the number of others
being affected. To be precise, the marginal dam- age MD
i
as expressed in Eq. 1 is the marginal damage per person affected by pollution of sub-
stance i, and the total marginal damage of sub- stance i could be calculated by multiplying the
marginal damage MD
i
by the number of people affected. Now, if all pollutants affected the same
number of people, such a multiplication would be unnecessary since the index is only a relative
measurement and has no meaning in absolute terms. For our purposes though, it is necessary to
adjust the valuation of nitrate emissions into lakes and rivers as well as greenhouse gas emissions in
order to take into account the larger geographical effect of these two substances in comparison with
the other pollutants. Hence, the marginal damage in Eq. 1 must be extended with a factor that
expresses the size of the affected population rela- tive to a reference population see Eq. 1. To
illustrate the extended Eq. 1, consider two pol- lutants with the same actual and critical flows but
with a different effective range. Without taking into account the different number of people af-
fected, the two pollutants were assigned the same marginal damage. But since the marginal damage
per person affected is equal only, an adjustment with a population factor is needed. Therefore, the
pollutant with a larger effective range is assigned a higher total marginal damage. By the same
token, the benefit and the total effectiveness of Eqs. 2 and 3 must be adjusted accordingly:
MD
i
= 1
Fc
i
F
i
Fc
i
P
i
P
CH
1
7
The nitrogen run-off into the Mediterranean Sea is not considered an ecological problem by the Swiss Federal Office
for Environment, Forests and Landscape because the coastal waters are deeper than in the North Sea. Therefore, this effect
is ignored.
6
For a theoretical discussion on this topic and an applica- tion to the production of ozone, see Von Ungern-Sternberg
1987.
B
i
= −
F
i
− DF
i
F
i
1 Fc
i
F
i
Fc
i
P
i
P
CH
dF
i
2 E =
i
B
i
= −
i F
i
− DF
i
F
i
1 Fc
i
F
i
Fc
i
P
i
P
CH
dF
i
3 where Fc
i
is the critical flow of pollutant i in affected area; F
i
is the actual flow of pollutant i in affected area; P
i
is the population in area affected by pollutant i CO
2
-equivalents: 5.25 billion; ni- trogen into lakes and rivers: 250 million, all other
pollutants: 7 million = P
CH 8
; and P
CH
is the pop- ulation in Switzerland 7 million.
For a cost-effectiveness analysis that only con- sidered the national effects of emission reduction,
no adjustment for international public goods would be necessary. As a consequence, the calcu-
lated effectiveness of greenhouse gas and nitrogen reductions would be biased downwardly, since the
international benefits of a national emission re- duction would be ignored. In this study, it has
been decided to include the international effects of national policies as well.
9
2
.
3
. The relati6e marginal damage of pollutant emissions
In Table 1 the pollutants and the corresponding actual and critical flows per year are listed. Since
it is assumed that until the year 2002 the agricul- tural measures will realise their full effect, we
chose 2002 as the reference point in time. Hence, all actual flows and all emission reductions refer
to the year 2002. The critical flows, on the other hand, do not depend on the choice of a reference
year. The last column of Table 1 shows the mar- ginal damage per thousand tons of pollutant
emission — calculated according to Eq. 1. The last column in Table 1 shows a high mar-
ginal damage due to phosphorus emissions. This is mainly due to the relatively low critical phos-
phorus flow giving a unit of emission a relatively large weight. At the other end of the scale, the
global critical flow of greenhouse gas emissions is very high. This explains why the corresponding
marginal damage per unit of emission is low even though it is considered that the total world popu-
lation is affected by greenhouse gas emissions. The other factor explaining the relative marginal
damage of an emission unit is the ratio of actual and critical flow, which is highest for nitrogen
oxide and lowest for chemical oxygen demand.
10
Table 1 also shows that although the ratio of actual and critical flow for nitrogen into lakes and
rivers and for carbon dioxide equivalent is the same, the marginal damage differs substantially.
Again, this is due to the different critical flows of the two pollutants, giving a thousand tons of
nitrogen emissions a much larger weight than the same amount of greenhouse gas emissions.
3. An application to nitrogen reduction measures
