Should environmental projects be subsidised? An empirical
Working Paper Series in Economics and Management
No. 06/03, December 2003
Department of Economics and Management
Norwegian College of Fishery Science
University of Tromsø
Should environmental projects be
subsidised? An empirical analysis.*
NORUT Social Science Research/University of Tromso
P.B. 6433, N-9294 Tromsoe/N-9291 Tromsoe
phone: +47 776 29400/fax: +47 776 29461
Imperfect markets, asymmetric information and transboundary pollution are all characteristics that
in most cases lead to inefficient “market” outcomes, and which thus are arguments for (public)
intervention in the market. On the other hand, these characteristics also imply strategic behaviour
by the economic agents, and then the effects of public intervention may be different from the
traditional results of e.g. subsidies.
The point of departure for this paper is the trading of an environmental project in a market with the
above mentioned characteristics and where the pollution is transboundary. The trade is promoted
by (foreign) authorities in that they offer a grant is trade takes place. We show that the effects of
the grant strongly depend on the interests of the authorities, and that the subsidisation does not
necessarily make the trading outcome more efficient.
Keywords: imperfect market, asymmetric information, transboundary pollution,
* I would like to thank Prof. Derek J. Clark for very useful comments and suggestions during the
work with the paper. I would also like to thank Prof. Manfred J. Holler, Dr. Claire W.Armstrong
and Heide Coenen for useful comments to earlier drafts of the paper. All errors in the paper are the
** Corresponding address: Institute of Socio Economics, University of Hamburg, von Melle Park
5, D-20146 Hamburg.
Should environmental projects be subsidised? An empirical analysis
There are several economically based reasons for subsidising environmental projects. One of
the consequences of the implementation of such projects is cleaner nature. This is a collective
good, and when people beyond the trading agents are positively affected by the
implementation of a project it can be said to have positive external effects. From basic
economic theory we know that both in the case of collective goods and when there exists
external effects, the market left alone provides a suboptimal quantity of the good.
Another argument is that when there is asymmetric information in a market, e.g. that the seller
of a project does not know the valuation of the buyer, one can show that under given
assumptions1 there is no efficient trading mechanism. In such a situation a subsidy may
promote the efficiency of the trading outcome.
Further, when the pollution caused by economic activity is transboundary, then it can be
shown2 that when each country only takes into consideration the damage made in their own
country when regulating the activity, the resulting polluting emissions will be too high
compared to what is socially optimal. When the countries affected by the pollution co-operate
the socially optimal emission level may be implemented, but in many cases this require a
1 The assumptions are that there is a probability for no gains from trade and that budget balance is required.
Myerson and Satterthwaite (1983) were the first to derive this result and it has been known as one of the
inefficiency theorems (Fudenberg and Tirole 1993, p.275).
2 Among these we can mention Mähler (1989), Kaitala et al (1991)
side-payment (which can be defined as a subsidy) to the country, which have to decrease its
emission the most3.
However, introducing a grant in a trading situation will have multiple and varying effects,
depending on the characteristics of the market and the ex ante information structure. Hence,
the aim of this paper is as follows:
I: To discuss the effects on market efficiency of a subsidy to an environmental project when
the market is monopolistic, there is asymmetric information and the trading agents behave
II: To derive the optimal (public) subsidy to offer in the above situation, and discuss how this
varies with the interests of the provider of the subsidy (e.g. authorities, Government).
As we see it, the market characteristics mentioned under I) are rather usual concerning the
trading of environmental projects. This will be further documented in section 2. Also, a
Government may have its own interests to follow in the environmental politics, and these
need not coincide with welfare maximisation. Hence, we give examples of different possible
interest of a Government offering subsidies to an environmental project.
The paper is organised as follows: In section two we present the empirical case on which the
analysis in this paper is based. In section 3 an analytical model is presented and solved by use
of a Bayesian game. In section 4 we concretise the role of the provider of the grant
3 For such a co-operative solution with side-payments see Kaitala et al (1991) who describes the situation
between Finland and parts of the former Soviet Union.
(Government), assuming three different types of this agent. In section 5 the results from the
model is discussed with respect to the empirical case, and in section 6 we conclude.
The border area between Finland, Norway and Russia is characterised by significant
environmental degradation due to pollution from an extensive industrial activity. One of the
main sources of this pollution is the nickel smelter Petsjenganikel in the town of Nikel. Since
1988 there have been negotiations between Finnish, Norwegian and Russian representatives
about modernising the plant and thus reducing emissions of sulphur dioxide by 90%. Both
Finnish and Norwegian companies possess the relevant technology, and they have developed
plans (projects) for how to transfer it to Petsjenganikel. The respective governments have
attached a grant to such a project if implemented. However, neither succeeded in reaching an
agreement with the Russian counterpart. Below is a short presentation of the negotiations.
A: The first phase (1990-1991)
The registration of long-transported pollution in Europe through the monitoring programmes
EMAP (European Monitoring Assessment Programme) in 1976-1978 showed severe
pollution streams from Russia to Finland and Norway. Later Russian-Norwegian expert
groups, under the Joint Norwegian-Russian Commission on Environmental Questions
explored the local environmental impacts of these pollution streams (AMAP 1997). One of
the main sources of the transboundary pollution streams between Russia, Finland and Norway
is the nickel-plant Petsjenganikel. In 1990 a general agreement was reached between Russian,
Finnish, Swedish and Norwegian authorities about reducing the emissions of SO2 from
Petsjenganikel. The Finnish company Outokumpu developed a technological solution,
involving a 90% reduction of the SO2 emissions, based on the emission level in 1989, with
costs equalling US$ 600 mill. Using a traditional cost-benefit analysis, the project was
recommended to the Russian government (Kola Science Centre et al, 1992). Both the Finnish
and Norwegian government attached a grant to the project. The Norwegian grant amounted to
US$ 42 mill, whereas the Finnish grant consisted of deliveries of equipment and services to
B: The second phase (1992-1995)
At a Nordic-Russian meeting between foreign Ministers in Kirkenes, Norway in 1992, the
Russian Government, led by Jeltsin rejected the Outokumpu-project. The reason given was
the very high costs of the project. In 1993 Russian authorities arranged an international
competition for the modernisation-project, encompassing emission abatement and
technological upgrading of Petsjenganikel. Five proposals were delivered, among them one
from Outokumpu and one from a Norwegian-Swedish consortium.
The Norwegian-Swedish proposal was chosen as the winner. It amounted to US$ 297 mill,
and involved a reduction of SO2 emissions by 90%. As a consequence of further negotiations,
the costs were reduced to US$ 257 mill. The costs of the winning project were to be divided
between Norway, in the form of a grant, (1/6), the Russian government (1/6), and Norilsk
Nickel, the mothercompany of Petsjenganikel, (2/3) (Kvaerner Technology 1996). When
Outokumpu lost the competition the Finnish government withdrew all support to a
modernisation project for Petsjenganikel.
In 1995, the Russian private investment bank Oneximbank bought a majority of the shares in
Norilsk Nickel, the mother company of Petsjenganikel. Consequently, the responsibility for
implementing the project was transferred from Russian authorities to the private owners.
Before accepting such an implementation, the leaders of the mother-company demanded that
Russian authorities still share some of the costs, specifically through tax relief. In March 1996
President Jeltsin gave a general relief from export-tax. This, however, was insufficient for the
company to implement the modernisation project at Petsjenganikel.
C: The third phase (1996-2001)
In December 1996 an agreement about technical details of the modernisation project was
reached between the leaders of the consortium and Petsjenganikel, but the project could not be
implemented before the leaders of Norilsk Nickel signed the agreement. The negotiations with
the Russian government about further public subsidies to the project were continued. In 1997
the Nordic Investment Bank was employed to replace the Norwegian Government in the
negotiations. The Bank worked together with representatives from Norilsk Nickel to find a
feasible solution. In June 2001 an agreement on a Russian modernisation project for
Petsjenganikel was reached, and in December 2001 The Norwegian Ministry of Environment
signed an agreement with the Nordic Investment Bank about financial support to this project,
which is assumed to cost US$ 91,5 mill. and reduce the SO2 emissions with 90% compared to
the emission level in 1999 (Norwegian Ministry of Environment, 2001).
It has been argued that the governmental grants mainly serve to support national exporting
industries, and scientists both from Russia and Finland question its environmental effects
(Kotov and Nikitina 2000, Hiltunen 1995). Representatives of the Regional Environmental
Committee in Murmansk County also doubt the effects of the grant due to the high price of
the projects offered, and of which the grant only constitutes a minor part. These are interesting
viewpoints in the light of our model results, and will be discussed further in the next sections.
Concerning the theoretical basis for this paper, there is a vast literature discussing the effects
of an exogenous subsidy under imperfect competition such as oligopoly, and there are also
analyses in which the subsidy is treated endogenously, but then the information is symmetric.
The literature relevant for this paper can roughly be divided into two categories;
i) micro-economic models with an exogenous subsidy. These are either traditional
neo-classical models or models within industrial organisation, characterised by
imperfect markets and/or information
ii) industrial organisation models of symmetric information and an endogenous
subsidy decided by a social planner (welfare maximisation)
From the first group this paper especially has exploited the inefficiency theorem derived by
Myerson and Satterthwaite (1983), and in general the theories in mechanism design presented
in Fudenberg and Tirole (1993). To the degree that these introduce a subsidy, this is done
exogenously. From the second group can be mentioned analyses by Laplante (1990),
Herander (1995), and White (1996). These mainly assume symmetric information, and
discuss the optimal subsidy or tax in order to maximise social welfare. Theoretically, this
paper combines aspects from the two above mentioned categories, as it models an imperfect
market structure with asymmetric information and introduces an endogenously decided
subsidy. Finally, this paper is strongly inspired by and indebted to the literature on
transboundary environmental problems, especially the works of Kaitala et al (1991) and
Huber and Wirl (1995). Empirical work belonging to this literature is Peszko and Zylicz
(1998), who has been used as an example of the relevance of the analysis. They discuss the
subsidisation of environmental projects in Eastern European countries.
A trade model with subsidies
There are three agents in the model; buyer, seller and the provider of the grant, the latter
hereafter called Government. The seller belongs to the same country as the Government,
whereas the buyer is a foreign agent.
v P X
U (W (P c) G)X
be von Neuman-Morgenstern utility functions for buyer, seller and Government respectively.
Buyer’s valuation, v, is private information, but it is common knowledge that it is distributed
on [ V,⎯V], with distribution function F(v), and density function F'(v) = f(v). Throughout the
analysis we will assume the uniform distribution, such that f(v) ≡(⎯V-V)-1. P is the price the
seller offers the buyer for implementing the project, whereas c represents the costs of offering
the project, and this we assume to be exogenously fixed. X is the probability that trade takes
The Government represents domestic consumer and producer interests. W denotes consumers’
surplus of the project as assessed by the Government. This can be interpreted as the positive
external effects of implementation of the project as valued by the Government. α is a fixed
parameter, equal to or larger than zero, and defining the share of the seller’s profit, which
affects Government’s utility. G is the nominal value of the grant. W and α are exogenously
fixed, and these are the variables to be varied for different types of Government.
The endogenous variables to be decided within the model are the price offered by the seller,
P, the grant offered by Government, G, and the probability of trade, X. Further, we introduce
an endogenous variable denoted (equilibrium) efficiency loss, defined as possible gains from
trade, which are not realised in equilibrium.
We now define a 2-stage game between the 3 above mentioned agents. At stage 1 the
Government and the seller announces a grant and a price for the project simultaneously, and
forward an offer, equalling seller's price minus grant, to the buyer. The buyer does nothing. At
stage 2 the buyer either accepts or rejects the offer, and Government and seller do nothing.
When the offer is accepted, the project is implemented, which results in pay-off to the players
given by the utility functions (1)-(3). If the offer is rejected, all players get a pay-off equal to
zero. We assume that the 3 agents all maximise utility and act strategically. The latter implies
that each player takes into account the (expected) action of the other players when choosing
its own behaviour. A grant is only offered, if Government’s utility, when taking into
consideration the expected actions of the trading agents, is higher when offering a grant
compared to not offering a grant.
As a benchmark we derive the market solution in the absence of an interfering Government.
This market outcome is given by (4) and (5)4.
4 This is the same as the outcome of the monopoly-pricing model when demand is linear.
which is accepted by the buyer iff v≥P*. The probability of trade in market equilibrium is
X( P* ) =
) ( 2
In the special case when c=V, X=1/2. When c>V the probability of trade is below 1/2.
The above solution implies inefficiencies due to the market imperfections when v ∈ (c, P*).
Further, the existence of externalities, represented by W, contributes in another way to the
inefficiency of the market solution. In total, the inefficiency, hereafter called efficiency loss
and denoted Z, is given by
() ( )
ZW v c dF v
which in market equilibrium equals
) ( 8
When there are gains from trade Z* is positive, which means that the market equilibrium is
always ex ante inefficient.
5 Conditions for this result to be valid are that ⎯V>P*>V, and P≥0.
The equilibrium inefficiency is higher the larger the difference between upper valuation limit
and actual project costs. The reason is that the higher⎯V, the higher P*, and thus the higher
F(P*). For given v this implies a higher probability of no trade. Also, it is larger the larger the
difference between costs and lower valuation limit, and the argumentation is the same as
above. The higher c the higher the price offered by the seller, and, for given v, the lower is the
probability of trade. That the efficiency loss increases in W is quite intuitive. On the other
hand, the larger the uncertainty about the buyer’s valuation, measured as the difference
between⎯V and V, the smaller the expected inefficiency. The reason is that upper and lower
distribution limits do not affect Z* through v symmetrically. The higher⎯V the higher v may
be, and the higher the probability of trade for given c. This reduces Z*. On the other hand, a
lower V implies lower potential values on v. But only for v higher than c no trade cause
inefficiency, such that only a part of the expected reduction in v caused by reduced V will lead
to an increase in Z*.
Letting the Government interfere in the trade by offering a grant implies that for every price
offered by the seller, the buyer now has to pay
π = P-G
Consequently, when deriving X, P must be substituted by π. Seller and Government maximise
(2) and (3) simultaneously, which provides the following reaction functions;
W (1 )P (1
Solving for the reactions functions simultaneously we get
(4 ) W
( )(c V)
The price offered to buyer in equilibrium is then
) (c W)(2)
6 For π* to be positive it must be the case that W<c+K⎯V, where K=(4-4α+3α2/4)(2-α/2)-1. When α=1, this
equal W<c+⎯V/2, whereas when α=0, it equals W<c+2⎯V.
P*G and G* are best responses to the action of the other player, and thus constitute a Nash-
equilibrium in the stage-1 game between seller and Government. (10) shows that when there
is a grant the seller will increase the price compared to when there is no grant, and he will
“confiscate” half of the grant. This result crucially depends on the distribution of the unknown
variable. Assuming the uniform distribution, which indicates linear demand, the seller will
always confiscate half of the grant. For the normal and right-biased distributions the
confiscated share depends on the equilibrium price without a grant. The higher this price the
lower the probability that the seller will confiscate parts of the grant, or, the lower the
expected share she will confiscate when a grant is introduced. For market equilibrium prices
below the expected value of buyer’s valuation the seller will always confiscate a part of the
grant. Generally, the higher the point probability for the price offered, i e the more likely that
this price is forwarded, the lower is the confiscated share of the grant. Only exceptionally the
whole grant is confiscated. These results are derived more formally in the appendix.
The grant is obviously increasing in W, and increasing W causes larger effects on the grant
than corresponding increases in the other exogenous variables. The effects of c and⎯V on the
grant depend on the size of α. (11) shows that the critical value of α is 2/3. For α below this c
has a positive and⎯V has a negative effect on the grant, and for α above 2/3 the opposite is the
case. The reason for this change in the effect of c and⎯V on the grant is that for low α, the
utility of trade to Government mainly consists of W, and the larger c the larger the grant must
be in order to secure trade such that W can be received. On the other hand, the higher⎯V the
lower the grant needs to be in order to secure trade. The higher α the more important the
seller’s profit is to Government’s utility, and the profit is higher the lower c and the higher⎯V.
Thus the grant decreases in c and increases in⎯V.
Seller's price is strictly increasing in all the explanatory variables, and the size of these
interdependencies varies with the size of α. Again, α=2/3 is the critical value. For this value
all the explanatory variables have the same influence on the price. For α different from 2/3 we
have two situations. When α is above 2/3, denoted the “high price case”, generating a high
profit has positive effects as this generates a high grant, which again is an incentive for the
seller to offer a high price. In this case W and⎯V are the explanatory variables with the largest
effects on the offered price. For given c we have that the higher⎯V and W the higher is the
seller’s profit. When α<2/3, denoted the “low price case”, generating a high profit does not, to
the same degree, have this effect, and thus the seller has less incentives to increase the profit.
In this case c is the explanatory variable with the largest effect on the price. For given W
and⎯V, the higher c the lower the profit.
The price the buyer has to pay increases in⎯V and c, and decreases in W, which seem quite
intuitive. It can be shown that for α>2/3, π* is always lower than P*, and for α<2/3, π* is
lower than P* when the condition for a positive grant is fulfilled. Thus, in the grant-
equilibrium the seller will never confiscate the whole grant.
Both P and G increase when α increases. Knowing that Government’s willingness to offer a
grant increases in α, the seller will also increase her price when α increases. For a sufficiently
low W and for given values on the explanatory variables, low values on α will imply that no
grant will be offered, but as it rises the Government will find it profitable to offer a grant. On
the other hand, it is not clear how α affects the price offered to the buyer. This depends on the
size of the other explanatory variables. The higher⎯V and c, and the lower W, the more likely
that π will decrease in α, when α is large. The reason is that a large α implies that the seller’s
profit is important for the grant. The higher⎯V the higher the profit, ceteris paribus, and this
compensates for the negative effects of a high c on the profit. To secure trade in this situation
the grant must be high such that the price offered to buyer is sufficiently low. It can be shown
that the change in the grant due to changes in the explanatory variables decreases as α
increases. This means that when α is large, changes in W, c and⎯V will cause smaller changes
in the grant compared to when α is large. Thus, α stabilises the grant with regard to changes
in the explanatory variables.
The criterion for a grant is
From (14) it is easy to see that α≥2/3 is a sufficient criterion for a grant. An α lower than this
requires either a high W and c, and/or a low⎯V, for there to be a grant.
These results concerning the effects of α on the grant are interesting in the light of the
“accusations” from scientists and environmentalists in the three countries, that the grant serve
mainly industrial interests and to a lesser degree environmental interests. Our results show
that industrial and environmental interests may be compatible, as the larger the industrial
interests in this case, the higher the probability that there will be a grant, and the more stable
the grant for changes in other explanatory variables.
The probability of trade in equilibrium is
(2 )(VW c)
It is straightforward to see that X*G increases in W,⎯V and α, and decreases in c, which is
also intuitive. Given the optimal strategy for the buyer (accept iff v≥π*), it follows from
above that X*G > X*. In other words, that a grant always increases the probability of trade.
In the 3-player game an efficient solution would imply that trade takes place when v+W+α(P-
c) ≥ c, whereas in equilibrium trade takes place when v ≥ π*. Using (13) it can be shown that
π*> c-W-? (P-c) when(?V+W)>c, which is the weakest condition for there to be a positive
probability for gains from trade. Thus, given that there may be gains from trade, the
equilibrium will always imply inefficiencies. An expression for this efficiency loss when there
is a grant is
() ( )(()) ( )
Z v c dF v
W P c
G dF v
When discussing this ex ante efficiency loss there are two effects, which must be taken into
consideration, represented by the two terms in the expression:
the grant, which may be introduced by the Government, implies a lower price
offered to the buyer, and this reduces the efficiency loss
when a grant is introduced, and this, due to strategic behaviour, does not entirely
internalise the utility of the Government from trade, there is an efficiency loss to
this player, which has to be added and thus cause the total efficiency loss to
When the grant fully internalises the utility of the Government from trade ii) equals zero, and
the efficiency loss to Government disappears. In this case, given that G>0 and that i) implies a
lower price offered to buyer, the total efficiency loss is reduced as a consequence of the grant
compared to the market solution. In all other cases, the efficiency gains, which are achieved
by i), must be compared to the losses, due to the strategic behaviour of the Government, given
by ii). For the uniform distribution of buyer’s valuation it can be shown that in equilibrium the
grant is always lower than Government’s gross utility of trade, leaving ii) positive.
In the grant-equilibrium derived above the efficiency loss equals
) ( 2
)))( ( 2
Inserting for the endogenous variables in this expression we get7
( )(4 4) (2) (
(3)(2) (3 )(2)
7 The former term on the right hand side is positive when there is a positive probability for gains from trade,
whereas the latter term is positive when (⎯V-c)>W(4-α2/4)(8-9α+7α2/4)-1., which, when α=0, reduces to
W<2(⎯V-c), and when α=1 reduces to W<(⎯V-c)/5.
Comparative statics show that the ex ante efficiency loss is negatively related to W, such that
the higher W the lower ZG*. The explanation to this is that the higher W the larger the grant,
which means that the more of the external valuation of the project is internalised. Further, the
higher the difference (⎯V-c) the higher ZG*. This intuitively makes sense as the higher this
difference the higher the expected monopoly profit. On the other hand, the grant is less
affected by changes in (⎯V-c) than is the price offered by the seller, such that the price offered
to buyer increases. Consequently, the upper integral limit increases with (⎯V-c), and for
given⎯V c must obviously decrease if (⎯V-c) increase, such that the integral in total increases.
The effects of α on ZG* are ambiguous. Numerical computations show that for positive W
(W=0,15),⎯V=1 and c=0,8, ZG* increases in α when α>1,263, whereas for numbers lower
than this it decreases. In other words, up to a given value the efficiency loss decreases when
Government put more weight on seller’s profit in its utility function. Above this value, the
opposite is the case, which means that there is a unique value on α that minimises the
efficiency loss. This value varies with varying values on the other explanatory variables. For
example, if we assume no externalities, i e W=0, ceteris paribus, the size on α that minimises
efficiency loss equal 1,5. This shows a certain trade of between W and α concerning ZG*, and
this we will go deeper into in the next section.
Three specifications of the model
As shown in section 3 the model results strongly depend on the assumptions made about the
interests of the Government. In this section we divide between three different types of
Government, each characterised by specific values on W and α. Both α and W may equal
zero, which would imply that Government does not attach any valuation to the project.
If we assume that the Government is a social planner aiming at maximise social welfare, W
would typically represent the consumer surplus and α the share of the producer surplus taken
into account, and this parameter would be set equal to 1. Alternatively, it is possible to
assume that the Government is a kind of political agent with its own interests. We will assume
two different kinds of the political agent Government. The first we call an industrial promoter,
and set α=1,5 whereas W=08. The latter implies that the government put no weight on the
consumer surplus (the externalities) in its utility function, or it assumes the externalities to be
insignificant. The second type of political agent we call a populist. To this kind of government
vote maximisation is the goal. We now have to redefine W, and not to bring about any
confusion we denote it WP. This is now the gross utility to Government if the project is
implemented. It may be measured as the value of being in government multiplied with the
change in probability for staying in this position if they can take the honour of having reached
an agreement about the project.
Table I presents the equilibrium solutions for the endogenous variables for the three types of
8 In section 3 we showed that for given numbers on the other explanatory variables, α=1,5 minimise the
expected efficiency loss when W=0. Though this is not an explicit interest of the Government we assume that it
is of no interest to it to increase α beyond this value.
Under the industrial promoter the grant and the price offered by the seller increase
significantly in⎯V, and the former decreases equally significant in c. Thus, the price offered to
the buyer increases in c and decreases in⎯V. Under the populist government the grant
increases in W and c and decreases in⎯V. This means that c and⎯V influence the grant in
opposite directions in the two cases. This result also has an intuitive explanation. The higher
the difference (⎯V-c) the higher is the expected profit to the seller. This contributes positively
to the utility of the industrial promoter but not to the populist. Thus, the former will have a
larger incentive to increase the grant in order to increase the probability of trade than has the
latter when either⎯V is high, c is low, or both. Under the social planner type of government
the grant is increasing in⎯V and decreasing in c, as was the case for the industrial promoter,
but the changes are more moderate. In addition it also increases in W. The price offered by the
seller increases in all the explanatory variables, but contrary to the populist case, it increases
more strongly in⎯V and W, and more weakly in c.
It has already been shown that when α>2/3 there will always be a grant. Thus, a social
planner and an industrial promoter will with certainty offer a grant. Under the assumption of
linear demand the seller will never confiscate the whole grant, and consequently the price
offered to buyer will be lower, which implies a higher probability of trade in these two cases.
Similarly, comparing ZG* from table I and (7) it can be shown that the efficiency loss is
always lower when there is a grant when the Government is either an industrial promoter or a
social planner. In other words, with these types of Government a grant will always be offered,
and it will lead to a higher probability of trade and a lower efficiency loss in equilibrium.
A condition for there to be a grant when the government is a populist is WP>(⎯V-c)/2). When
this is fulfilled the priced offered to buyer will decrease and the probability of trade will
increase. However, in contrast to the above example one cannot draw the conclusion that a
grant always implies a lower efficiency loss in this case. The reason is that WP may differ
from W. For low numbers on WP, it may be the case that the efficiency loss is higher when
this type of government interferes in the trade. The reason is that when WP is very low, G will
also be low, whereas the price offered by the seller increases. The latter will then cause ZG* to
rise more than the former cause it to fall. For the efficiency loss to be lower when a grant is
introduced it must be the case that either W and/or WP are of a certain significance. For given
W, the higher WP and the lower (⎯V-c) is, the higher is the probability for reduced efficiency
loss when a grant is introduced. When W=WP a grant will always reduce the efficiency loss.
Comparing the effects of a grant between the different types of Government it is intuitive that
the lower W and WP, the more likely that the probability of trade is highest under an industrial
promoter. And, the higher W is compared to WP, the more likely that the probability of trade
is higher under a social planner compared to under a populist. Similarly, the lower W and WP
the more likely that efficiency loss is lower under an industrial promoter compared to the two
other, and the higher W is compared to WP, the more likely that the efficiency loss is lower
under a social planner compared to under a populist.
Table II presents a numerical example.
The results in table II strongly depend on the numbers attached to the exogenous variables.
We have used⎯V as a numeraire, setting it equal to 1. Setting c equal to 0,8 and V equal to 0,1
implies that the costs of the project are higher than the buyer’s expected valuation. Further, it
is assumed that the utility to a populist Government of an implementation of the project, WP,
is higher than the consumer surplus of the project, W. This assumption will be argued for and
explained more in depth in section 5. One should be aware that small changes in these
variables may cause significant changes in the endogenous variables. Therefore, one should
read table 2 only as one among many possible scenarios. Also, the results in both this and the
previous section strongly depend on the game form. In the appendix we have analysed the
situation when the grant and seller’s price is decided in a Stackelberg game.
The grant is highest when we assume an industrial promoter government. In this case also the
price offered by the seller is highest, but the high grant manages to compensate for the high
price such that price offered to the buyer is lower compared to the two other cases.
Consequently, the probability of trade is the highest in this case. Also, the efficiency loss is
the lowest. The price offered by the seller is lowest under the populist government. On the
other hand, so is also the grant, and as a result the price offered to the buyer is higher in this
case compared to both other cases, which include a grant. Consequently, the probability of
trade is lowest under this type of government, and the efficiency loss is the highest.
The probability of trade follows automatically from the price offered to the buyer, and is
strictly decreasing when this price increases. The effects on this variable of varying the
explanatory variables are symmetric for all three types of government, differing only in
intensity. The expression for the efficiency loss varies between the types of government, and
it strongly depends on the relationship between W and α. In general there is a trade off
between W (WP) and α in Government’s utility function, such that different combinations
provide the same efficiency loss. The fact that the efficiency loss is lower under the industrial
promoter, for which W=0, indicate that W and WP has not been high enough in order to
compensate for lower (zero) α. Given the α’s in the three cases, necessary conditions for the
efficiency loss under a social planner and a populist to equal that under an industrial promoter
is; W= 0,15 and WP = 0,36. On the other hand, a lower α would increase the efficiency loss
under an industrial promoter. Finally, in the numerical example, the efficiency loss is lower
due to the grant independent of the type of government.
Discussing the results with respect to the empirical case
The Nordic modernisation projects developed for, offered to Petsjenganikel and negotiated
about during the years 1992-1996 were all rejected. As an alternative the mother company of
Petsjenganikel launched a new, internal plan for modernisation of the plant, which would lead
to a corresponding reduction in harmful emissions. This plan presupposed internal
responsibility for the project, only using technical assistance and necessary input from
external supporters. In spite of this change in the role of agents the Norwegian government
maintained the grant originally offered. According to the model presented in section 3 this
implies that α in the case of the Norwegian Government was low. In contrast, the Finnish
government withdrew their grant when a modernisation project developed and offered by a
Finnish company was rejected by the Russian buyer. According to the model this implies that
α has been high under the Finnish government.
Sticking to the Finnish case, it is claimed that the Finnish co-operation with Murmansk
region, in which the Petsjenganikel modernisation is far the most important, mainly has been
stimulated by the high level of Finnish environmental technology (Kuokkonen 1993, p.19). At
the same time, the Russian participants in this co-operation criticise the Finns for their strong
promotion of own technology and lack of financial backing (Kuokkonen 1993, p.35). The
former may be interpreted as if α, in the case of the Finnish Government, is high. The latter,
however, may indicate that this does not lead to a corresponding high grant. This combination
of results are inconsistent with the model results. The Finnish reply to the Russian critics was
to point out the difficult economic situation in Finland at that time. Taking budget constraints
into account the model may be modified by introducing an upper boundary to the grant, such
that we get; G=[max GUG, Γ], where Γ is the upper limit of the grant.
W, defined as the consumer surplus of the project, is the valuation of the implementation of
the project by those who live in the affected area. It is true that these people are significantly
affected by the sulphurdioxide emissions from Petsjenganikel. However, as the population
only count a couple of thousand, the total valuation, derived e.g. by the contingent valuation
method, will be limited. On the other hand, the Petsjenganikel case received much national
publicity, especially in Norway, in the mid 1990’s. This raised the awareness for the case, and
it became a demand from people beyond those living in the area that the Government must get
involved in order to find a solution to the pollution problem. Further, in 1992 the Barents
Region was founded through a Norwegian initiative. One of the targeted issues for this
Nordic-Russian cooperation was the environment. From this it is easy to defend the
assumption made in section 4 that WP is larger than W.
Thus, the empirical facts tend in the direction of the following claims:
α for the Norwegian government was low
the consumer surplus W, as derived by e.g. the contingent valuation method, was
the valuation to the government of supporting and promoting the project, WP, was
of a certain significance
Based on the above claims, what can be said about the role of the Norwegian government in
the Petsjenganikel case? Obviously, they draw in the direction of a populist type of
Government. The main reasons for this is 1) the fact that the grant was maintained though the
accepted project contained no Norwegian industrial interests, and thus indicate an
insignificant α, and 2) the rather high value of the grant relatively to the number of affected
people and compared to Norwegian support to other transboundary environmental projects.
In other words, the combination of significant publicity about the Petsjenganikel case and a
government not possessing the omnipotence to act as a social planner and without strong
industrial interests in this case makes it likely to assume that the Norwegian government has
played the role as a populist in this case. What conclusions can then be drawn with regard to
the endogenous variables? In section 4 we showed that under this type of government the
valuation of the effects of the project must be of a certain size for there to be a grant. A grant
would always increase the probability of trade, but the effects on the efficiency loss where not
clear. As the Norwegian government did offer a grant, clearly WP must have been of a certain
size (i e larger than (⎯V-c)/2). Consequently, due to the grant the probability of trade has been
increased (though not enough to secure trade). Whether the efficiency loss has been reduced
or not can only be discussed conditionally. The larger W and the smaller (⎯V-c), the larger the
probability that the grant has reduced the efficiency loss in equilibrium. We have already
argued that W is probably not very high. On the other hand, (⎯V-c) may be very small, as the
Russian willingness to pay for the project is obviously quite limited, and the costs of the
Nordic projects were relatively high. Thus, it is not unlikely that the Norwegian grant offered
if a Nordic modernisation project was implemented at Petsjenganikel, has implied reduced
equilibrium inefficiency in the trading situation.
Introducing an endogenously decided grant in a trading situation characterised by bilateral
monopoly and asymmetric information have effects on the probability of trade and efficiency
loss in equilibrium, and these effects depend heavily on the interests of the provider of the
grant. Assuming that the provider of the grant behave strategically it can be shown that the
more weight it puts on industrial interests, defined as the producer surplus and measured by α,
the higher the probability that it will provide a grant. Further, the higher the grant and the
more stable the grant will be with regard to changes in other explanatory variables. A public
grant may also be offered of pure popular reasons. In this case the valuation of the effects of
the grant by the provider must be rather high in order to secure the same effects on efficiency
loss in equilibrium as when there were strong industrial interests. The traditional role for a
provider of a grant is that of a social planner, who takes both consumer and producer surplus
into consideration when fixing the grant.
In the first and last case, when α>2/3, a grant will always be offered. This leads to a higher
probability of trade and a lower efficiency loss in equilibrium. Assuming a populist a grant
will be offered if its valuation of the project exceeds a certain part of the gains from trade.
Given a grant in this case the probability of trade will always be higher, whereas it is
ambiguous whether the efficiency loss will be lower.
In situations with transboundary pollution, efforts from downstream countries, characterised
by subsidies to national industry in order to offer abatement technology to the polluting plants
in the upstream country, sometimes are accused of serving industrial interests rather than
environmental interests. Our analysis shows that the more a government let national industrial
interests matter in their utility function, the more likely it is that they will offer a grant, and
the larger the grant will be. It also shows that this grant is not fully confiscated by the seller of
the abatement technology, such that the polluting country will also be better off as a
consequence of the grant.
A: A general derivation of seller’s confiscation of the grant
A general expression for seller’s utility is
U (P c)(1 F(P))
which gives the following equilibrium price when there is no grant
when there is a grant the corresponding expressions are
U (P c)(1 F(P G))
1 F(P G)
P*G > P* if 1 F(P G)
The left hand side of the latter inequality is always larger than 1. Hence, those distributions in
which f(P) is always decreasing in P, may not fulfil the above condition. In the uniform
distribution the right hand side is always equal to one, such that the inequality is always
fulfilled, and it will always be optimal for the seller to confiscate only a part of the grant in
equilibrium. For the normal and right-biased distributions, such as the ?2 and Pareto
distribution, we have the following property:
If P>E(v)+G then f(P-G) >f(P) and vice versa
Thus, we can conclude; the larger P without a grant, the smaller the probability that the
inequality is fulfilled, which means that the smaller is the probability that the seller will
confiscate any share of the grant. From the same argumentation we have that the probability
for total confiscation is then even smaller. For small P and all P lower than E(v) the seller will
always confiscate a share of the grant when this is introduced, and in general we have that the
lower the price offered the larger the confiscated share will be.
Similarly; in the case of uniform distribution, the seller will never confiscate the whole grant,
but will always take a part of it. Assuming other distributions, both the extreme situations may
be the case. If the original price (the price without a grant) is very low, the seller may
confiscate the whole grant, and for high original prices, he may not confiscate any share.
B: Analysis of the game when the grant and seller’s price is decided in a Stackelberg game
Alternatively, the negotiations about grant and price offered by seller can be formulated as a
Stackelberg game. This, however, provides more limited results than in the Bertrand setting.
Assuming that the seller is the leader, the model breaks down when α→1. In this case the
seller knows that the Government will fix the grant in order to keep any price offered within
the limits for buyer’s valuation. Knowing this a profit maximising seller will increase the
price infinitely, and thus the grant also has to be infinitely large in order to secure a positive
probability of trade.
When α=0 solving the model provides the following results for seller’s price and grant:
Assuming that the Government is the leader, when α=0, we get the following results for
seller’s price and grant:
In both cases price offered to seller, probability of trade and efficiency loss equals
) ( 4
) ( 8
)) ( 5)(
) ( 3 (
The difference between the two formulations of the Stackelberg game is that when seller is
the leader both price offered by seller and grant is higher compared to when Government is
the leader. It can be shown that when α=0, the Bertrand game provides a lower price offered
to buyer, and thus a higher probability of trade compared to the Stackelberg game. Also, it
provides a lower efficiency loss.
Keeping to the case when Government is the leader it can be shown that when α=1 the grant
and the price offered by seller is higher in the Bertrand setting compared to the Stackelberg
setting. On the other hand, the price offered to buyer is lower in the Bertrand setting, such that
the probability of trade is higher when grant and seller’s price is decided in a Bertrand game
compared to a Stackelberg game where Government is the leader. It can also be shown that
the efficiency loss is lower in the former case.
AMAP (1997): Forurensning i Arktis. Tilstandsrapport om det arktiske miljøet. Oslo
Fudenberg, D. and Tirole, J. (1993): Game Theory. MIT Press
Herander, M-G. (1995): A welfare theoretic argument for regional subsidization of industry in
the presence of inferior technology. Open economies review ( 6), 255-263. Kluwer Academic
Hiltunen, H. (1995): Finnish strategies to reduce transboundary pollution from the Kola
peninsula. Lessons from implementing joint Implementation. YSSP 1995, IEC Project
Huber, C. and Wirl, F. (1995): Optimal incentives to reduce transboundary emissions: Theory
and empirical illustration to sulphur emissions in Austria and (former) Czechoslovakia.
Internal Discussion Paper, Technical University of Vienna.
Kaitala, V., Pohjola, M. and Tahvonen, O. (1991): An analysis of SO2 negotiations between
Finland and the Soviet Union. Finnish Economic Papers 4 ( 2), 104-112.
Kola Science Centre, Russia, Ekono Environmental Technology, Finland, 3E Economics,
Norway (1992): Evaluation of key environmental impacts of Pechenga-Nikel modernisation
project. Volume 1. Final report.
Kuokkanen, M. (1993): Preconditions for environmental protection in the neighbouring areas
of Finland. The Finnish Institute of International Affairs, Helsinki
Kvaerner Technology (1996): Reconstruction Project for Petsjenganickel Metallurgical
Laplante, B. (1990): Producer surplus and subsidization of pollution control device: A non-
monotonic relationship. The Journal of Industrial Economics 39, 15-23.
Mäler, K.G. (1989): The acid rain game. In Folmer, H. and von Ireland, E. (eds): Valuation
Methods and Policy Making in Environmental Economics. Elsevier Science Publishers B.V.
Myerson, R. and Satterthwaite, A. (1983): Efficient mechanisms for bilateral trading. Journal
of Economic Theory 28, 265-281.
Peszko, G. and Zylicz, T.(1998): Environmental Financing in European Economies in
Transition. Environmental and Resource Economics 11 ( 3-4), 521-538.
The Norwegian Ministry of Environment (2001):
1) Avtale mellom Det Norske Miljöverndepartementet og den nordiske investeringsbanken om
forvaltning av tilskudd til modernisering av Kola gruve og metallurgiske kombinat.
2) Avtale mellom Kongeriket Norges regjering og Den Russiske Föderasjonens regjering
vedrörende modernisering av Kola gruve og metallurgiske selskap AS
White, M-D. (1996): Mixed oligopoly, privatization and subsidization. Economics letters
(53), 189-195. Elsevier Science
The endogenous variables given the three different types of Government
) ( 5
W c V
) ( 2
) ( 2
V c W
) ( 3
) ( 4
) ( 9
Type of Government
Price offered by seller
Price buyer has to pay
Probability of trade
⎯V=1, V=0,1, W=0,05, WP=0,25, c=0,8
Calculation of the endogenous variables for different types of Government
No grant Social planner
An environmental project is a project, which e.g. aims at reducing polluting emissions. In
many cases such projects are traded in the market and implemented in a private company.
Due to the character of such projects the market will often be monopolistic. Further, the seller
will normally not know the buyer’s valuation of the project. On the other hand, implementing
an environmental project will normally have implications for people beyond the traders, so-
called external effects, in the form of cleaner nature, which is a collective good. This makes
the implementation of the project more valuable to the society as a whole than to the buyer
alone. The market characterisations imply that the trading situation is inefficient in that trade
does not necessarily take place though it is profitable to both trading agents (and thus even
more profitable to the society). Consequently, we have a situation where trade is higher
valued by the society as a whole than in the market, but where trade takes place with lower
probability than what is “optimal”. This obviously is inefficient.
In this paper we analyse how the introduction of a (public) subsidy, in order to promote trade,
effects a trading situation as the one characterised above, and we ask; i) does the grant
increase the probability of trade? and ii) does it make the trading outcome more efficient? By
the latter is meant whether trade to a larger degree takes place in the market when it is
profitable to the society compared to when there were no subsidy.
Our results show that the probability of trade increases due to the subsidy. The results with
regard to market efficiency are more ambiguous, and depend on the interests of the provider
of the subsidy (authorities). An interesting result is that the more weight the authorities put on
the producer surplus in the welfare function, the higher is the optimal subsidy it offers, and
the less is the market inefficiency.
The empirical case, which has inspired this paper, is the rebuilding of the Petsjenganikel
factory in Northwest Russia, close to the Norwegian border. The very large and polluting
emissions of sulphur-dioxide from this factory caused Norwegian authorities to offer a
subsidy if the factory implemented a rebuilding (environmental project), which would result
in 90% reduction in the emissions. The rebuilding should be implemented by a Norwegian
supplier. This latter fact caused severe critics from environmentalists on both side of the
border, accusing the Norwegian authorities for trying to profit economically on a “popular”
environmental issue. However, as mentioned above, our analysis show that taking industrial
considerations into account may affect the environmental effects positively, because a
positive link between industrial and environmental interests will increase the optimal subsidy
Word count: 453