The Interaction of Global Warming Induced
Water-Cycle Changes and Industrial Production –
A Scenario Analysis for the
Upper Danube River Basin
(Working paper version February 28th 2009)
Jeßberger, C.*
M. Zimmer *
* Ifo-Institute for Economic Research
at the University of Munich
Poschingerstr. 5, 81679 Munich
[email protected], [email protected]
Abstract
The industry is one of the main users of water resources. Water is an essential
production factor and is used in almost every production process of a
manufactured good. Our main focus in this work is to analyse the effect of
differing climate or policy scenarios on the sustainable use of water-resources by
industrial producers in the Upper-Danube Catchment Area till 2025. We estimate
region-specific production functions including water as production factor to
calibrate and simulate interdisciplinary scenarios and its reactions effects to
climate change in the Upper-Danube river basin. Therefore we develop a flexible
theoretical framework to evaluate empirically the shadow values of industrial
water-use using a translog production-function. It includes quasi-fixed production
factors and proposes an extension to assess the shadow value of water which is
consistent with the previous research but does allow for efficient use of water
resources. Simulation results show wide differences in regional reactions of firms.
Also comparing scenarios with moderate or serious climate change illustrate the
severe future conditions for some regions.
Keywords: Environmental Decision Support System, Climate Change, WaterCycle, Translog Production Function, River Basin Management
JEL: D24, R30, Q01, Q25, Q52, Q53
Introduction
Our main focus in this work is to analyse the effect of differing climate or policy
scenarios on the sustainable use of water-resources by industrial producers in the
Upper-Danube Catchment Area. The natural and artificial water-cycles have
recently experienced a renaissance in political discussion. A prominent example is
the European Water Directive. But today the concerns have shifted away from the
focus on chemical pollution to climatic change issues like thermal pollution or
scarcity. To evaluate environmental regulations concerning the utilization of
water, it is essential to know how the resource users will react to the policy
measure and what the economic and environmental effects will be. We use
estimates of production functions in Germany which include water as production
factor to calibrate an interdisciplinary scenario simulation model. This model was
developed in an initiative by the German Ministry of Education and Research to
analyze the effects of climate change in the Upper-Danube river basin. In this so
called GCDSS-DANUBIA environmental decision support system we model
explicitly the industrial production and water usage, while we use the output of
the existing sub-models as input to our model and vice-versa. Thus we are able to
include the interaction and feed-back mechanisms of the industrial producers, and
for example social conditions like the labour market and migration or natural
conditions like aquifers or river-networks, which are all provided by the GCDSSDANUBIA.
Water resources obey natural borders and not administrative ones
In the philosophy of environmental decision support systems it is rather common
to observe a natural resource within its natural boundaries rather than to care for
administrative borders. This pays attention to the fact how natural phenomena
occur. Especially observing the water-cycle it is obvious that the watersheds
delimit the dispersion of pollution at least for surface waters. Computer-based
environmental decision support systems like GCDSS_DANUBIA account for this
and typically generate their results to be consistent with the natural borders.
Figure 1 shows the investigated Upper Danube River Basin which is the focus of
this analysis.
The large industrial water users that are responsible for the largest proportion of
total water usage in Germany are typically self-supplied. As such they perceive
water as an almost free production factor. Nevertheless, the extraction of water is
restricted by contingents. These sophisticated extraction permits are enacted by
the local environmental authorities. In industrial production processes water is
typically not consumed in the traditional sense. It is rather used for production
purposes and afterwards returned to the water cycle. The equivalent to
“consumption” is rather the reduction of the usable amount of the resource for
other natural or artificial utilisations. This might be due to the reduction of water
quality below a critical threshold. An upstream/downstream riparian conflict
might also be caused by a reason that is closer to the traditional interpretation of
consumption, if e.g. the water resources are evaporated in a cooling process.
2
Figure 1: The Upper Danube River Basin.
Source: GLOWA-Danube project
The GCDSS-DANUBIA
The theoretical foundations for the GCDSS-DANUBIA as well as the simulation
model itself have been developed and implemented in the GLOWA-Danube
project (www.glowa-danube.de) from 2001 to 2008. The target was to transfer the
theoretical fundamentals of integrated socio-economic and natural scientific
models of all project partners to a common scenario analysis and decision-support
system GCDSS-DANUBIA. The development of this tool has been completed
and the tool is able to simulate the interaction of the hydrological cycles and the
various socio-economic actors in the upper Danube river basin. It allows in
particular simulating climate scenarios and social shocks and their effects. The
objective of this work was a realistic simulation of the water-specific decision
processes of each for the water consumption relevant production sites and the
development and evaluation of climatic and social scenarios of interest. Figure 2
displays which values of interest are exchanged between the industrial model
described in this paper and the rest of the system.
3
Figure 2: interaction of the industry model with the GCDSS-DANUBIA
Industry model
Export
Import
Population/labour market
River water supply
Groundwater supply)
Public water supply
Water price
Capital market
Industrial production
River water demand
Ground water demand
Public water demand
Employment
Capital stock
Disciplinary sub-models
GCDSS-DANUBIA
Source: Ifo Institute
Model development
An excurse to the macroeconomics behind the decision support system
To evaluate the consequences of industrial production for local natural resources
it is necessary to simulate the complex production and decision process within a
production site on a microeconomic scale. Still, not all economic processes which
are needed to characterize economic development are necessarily or possibly
modelled on the level of an individual firm. The simulation of the industrial
production within the decision support system therefore consists of two separate
sub-models. Firstly a microeconomic agent-based model which is presented in
this work and simulates one representative industrial producer on each
industrialized square kilometre. Secondly a macroeconomic model is used to
endogenously generate the regional economic indicators with regard to spatial
spillovers. These two models do not work separately but form a simultaneous
system. The implementation of the macroeconomic model was the first step in
developing the industrial production module within the GCDSS-DANUBIA. This
model simulates supra-regional trends and the important economic
interdependencies that are not simulated in the agent-based model. A first basic
implementation of this model was based on the work of Langmantel and
Wackerbauer (2003) and has been developed further to the current state. The
model incorporates the influence of spatial structures like agglomerations on
economic developments. Thus, local economic developments are influenced by
spatial spillovers from surrounding regions and economic developments exhibit
multiplier effects through the interregional feedback mechanism. This accelerates
4
agglomeration effects, which are again limited by crowding externalities (e.g. in
using common infrastructure as roads) and local resources. As a result factor
prices rise. For modelling these interdependencies the macroeconomic model
employs the econometrically determined elasticities as well as the macroeconomic
functions for production, consumption, government spendings and prices for the
production factors. The elasticities were estimated in Langmantel and
Wackerbauer (2003) and were implemented in the current macroeconomic model.
Economic growth is modelled exogenously within the labour productivity with an
annual trend of 2%.
Modelling the industry
The industry is one of the main users of water resources. Water is an essential
production factor and is used in almost every production process of a
manufactured good. The mining industry and the industries that produce paper
products, chemicals and metals demand especially large quantities of water. Water
is used for cleaning, diluting, transporting a product, cooling, heating, generating
steam, sanitation and, of course, as a constituent in the final product. The
industries that employ large amounts of water are typically self-supplied. They
often circulate the water within the production process or use it multiple times in
consecutive processes. While multiple employment might follow from economic
considerations, cycle-use is rather a reaction to regulatory constraints1.
In the simulation model the industry sub-model mimics the decisions of the
relevant industrial production sites as close as possible to reality and as abstract as
necessary and sensible. The decision process is focused on the questions of the
optimal production output and of how to produce this output with minimal costs
given regulative and resource constraints. In coherence with the dominant
research question in this work we spotlight the use of water resources in the
production process. A further special interest lies in the technologies used to
increase the efficiency of water usage. These technologies, such as cycle usage or
the use of the resource in multiple successive processes determines the utilization
factor of water. To put it simply: How often is each litre of water that enters the
production process employed in the process before it leaves the facility again? We
assume the agent to behave rationally given that her information is limited by her
perceptive abilities and her imperfect expectations. Therefore the cognitive
process of decision-making is due to a dynamic generation of decision rules as an
adaptive response to the perceived changes in the environment.
The conditions influencing the agent’s decisions can be categorized into three
groups: Factors which the agent perceives as exogenous and thus not
influenceable by his actions, factors that he perceives as being influenced by his
decisions, and factors which he can directly determine by choice. In our modelling
approach examples for exogenous factors are technological progress and the
condition of the water resources. Rather than incorporating the signal about the
sustainability of water-usage directly into the decision process, it is used to
1
See Egerer and Zimmer (2006).
5
determine the amount of regulation imposed on the production site by the local
environmental agencies. This approach has been identified as preferential since
the industrial producers themselves cannot observe the sustainability of their
resource usage. While counterintuitive at first glance, this is the consequence of
simple information asymmetry. It is indeed true that in reality the production site
cannot observe the consequences of its water consumption and that the monitoring
of the environmental effects is done by the local environmental authorities2. These
also have the means to regulate the water usage by extraction or effluent charges
or by limiting the amount of water extraction. Among the factors the industrial
agent perceives as influenceable are the water-related expenditures (including
eventual charges). These are indirectly determined through factors of his direct
choice, namely his investments in technologies that reduce pollution discharge in
the effluents or increase the utilisation factor in the production process. Other
important factors of direct choice are the labour employed and the production
output.
Inside the industrial producer
Depending on the available resources there are different approaches to model the
industrial agent. Typically, the final implementation is based on anecdotal
evidence, theoretical considerations or econometric estimates. To construct the
industrial agent we explored all three of these options. In the theoretical approach
the production function of the firm is modelled and then the optimal factor
demands used in the simulation are derived from this model. To mimic the
production process it is necessary to gather as much information about it as
possible. To achieve this we conducted a questionnaire campaign, did field- and
telephone interviews and visited actual production sites. This participatory process
involving the industrial water users was described in depth in Egerer and Zimmer
(2006). The final step is to examine the available data that allows conclusions on
the production technology to be drawn.
As a result we designed the industrial production sites as profit-maximizing
entities in a competitive market environment. It was an essential requirement in
the construction to consider the effects of climate change and environmental
pollution. As discussed earlier it is reasonable to model the resulting
consequences for the firm as regulatory constraints. The model should not only be
able to simulate the effects on water demand and water-saving technology, but
should also capture the impact on the job market and local GDP. Due to the
integrative nature of GCDSS-DANUBIA these characteristics influence the
macroeconomic and the discipline-specific models, which will in turn create a
feedback on the industrial agent. The model results are calculated on the scale of a
single representative industrial production site on each industrialized square
2
This might not be the case for regions that are less restrictive than Germany and Austria
concerning the regulation of environmental pollution. In the observed area production facilities are
typically restricted before the environmental effects are obvious to the producer.
6
kilometre within the observed catchment area3. The characteristics of the agent are
determined by the local natural environment, the economic conditions and by the
econometric estimates of the production technology.
En = ∑i pn ,i X n ,i + ∑ k pn , k X n , k
(I)
Part of the profit-maximizing behaviour of a company is to minimize the
production cost for a given production output. With pn being the prices of the
production factors employed the total expenditures En of a production site include
the aggregated costs for the variable production factors Xn and quasi-fixed factors
of production X n 4.
Yn = f [X n , X n , T ]
(II)
The output Yn of the industrial facility is a function of the vectors of variable and
quasi-fixed production factors employed and of the technology level at time T .
This black-box, converting multiple inputs in the production output, mirrors the
technical production process in a production site.
Max profitn = pn ,Y Yn − En
xn
(III)
∏
The utility of the company can be measured by the profit n that it generates with
the production output5. The managerial effort is aimed at maximizing the
difference between revenues pn,YYn and expenditures En 6.
3
For the Upper-Danube Catchment Area this corresponds to a total of 1354 representative agents
as identified by analyzing the remote sensing data.
4
In the analysis of industrial water-use the term of quasi-fixed production factors commonly refers
to factors that are exposed to legal regulations and thus the amount employed in the production
process cannot be chosen arbitrarily.
5
In a competitive market real profits will be very limited, but the effort to maximize them enables
the company to stay in the market.
6
In an interdisciplinary project like GLOWA-Danube it is important to communicate the
disciplinary economic logic to the other disciplines within the network. It should therefore
explicitly be mentioned that this approach is well in line with the contemporary view of decision
processes in other disciplines. We want to highlight this exemplarily for the concept of satisficing,
which is a popular approach in modelling the decision process of an agent within the
psychological discipline. The idea of satisficing was motivated by the economic Nobel laureate
Herbert A. Simon. Sidney G. Winter applied the concept of satisficing to the behaviour of the firm
(Satisficing, Selection and the Innovating Remnant. (Quarterly Journal of Economics, 1971). He
showed that the traditional mathematical methodology of profit maximization is a consistent
representation of the decision process of a firm that is aiming to satisfy a simple rule of thumb
(e.g. at least zero profits as deficits result in bankruptcy in the long run) and that underlies a
random evolutionary process that generates new production technologies which the firm is free to
employ. Given a competitive market a simple Markov-Chain analysis reveals that in the long run
only the companies behaving optimally will survive. Thus the concept of satisficing and profitmaximization are unified to a common result by the existence of a competitive market. Since other
7
Due to the availability of the data, the production technology has to be estimated
on an aggregated scale that is larger than the individual production site. Thus it is
sensible to assume a production technology which is additive separable and
homogeneous of degree one. The obvious choice to start with is a Cobb-Douglas
production technology:
( )
Yn = e β 0 t ∏ i xiβ i
(IV)
The solution to the maximization problem is:
MRS =
β i xk pi
=
β k xi pk
(V)
xi∗ =
⎛⎛ p
y βi
⎜⎜ k
∏
k ⎜
t
⎜ ⎝ βk
β 0e pi
⎝
⎞
⎟⎟
⎠
βk
⎞
⎟
⎟
⎠
(VI)
Assuming output maximization at a given cost we get:
(
C = ∑i
⎛
⎛ ⎛ p ⎞ βk ⎞ ⎞
y βi
⎜
∏ ⎜⎜ k ⎟ ⎟⎟
p x = ∑i pi
⎜ β 0et pi k ⎜ ⎜⎝ β k ⎟⎠ ⎟ ⎟
⎝
⎠⎠
⎝
∗
i i
Y∗ =
)
C
⎛
⎛
⎜ β ∏ ⎜ ⎛⎜ pk
∑
β 0 i⎜ i k ⎜ ⎜⎝ β k
⎝
⎝
e −t
⎞
⎟⎟
⎠
βk
⎞⎞
⎟⎟
⎟⎟
⎠⎠
(VII)
And the price elasticity of the factor demand for constant returns to scale will be:
⎛⎛ p
⎜⎜ k
∏
k
βk
⎜ ⎜⎝ β k
⎛ ⎛ p ⎞ ⎞ β0
⎝
k
⎜
⎟
∏ k ⎜⎜ ⎟⎟
⎜ ⎝ βk ⎠ ⎟
β0
⎝
⎠
∂y ∗ p j
=
∂p j Y ∗ e −t
βj
e −t
⎞
⎟⎟
⎠
βk
⎞
⎟=β
j
⎟
⎠
(VIII)
individual decision processes are typically not penalized by a market if they are not in accordance
to optimal behaviour, this conclusion will not necessarily hold for other disciplines.
8
Estimating the Value of Water
To evaluate environmental regulations concerning the utilization of water, it is
indispensable to know the value of this resource for its consumer. Unfortunately,
the information currently available on the valuation of water by industry is rare
and limited to few locations. In this paper we close part of this knowledge gap by
providing estimates for the shadow prices of water and the most common effluent
contaminates. Within a translog-framework we analyze sector- and regionspecific production functions in Germany. We abstract from restrictive
frameworks commonly used. Thus, instead of imposing linear homogeneity or
separability a priory, we extend the more flexible framework proposed by Kim
(1992). We allow for variable returns to scale and non-homothetic production
technologies and generalize the framework by including quasi-fixed factors of
production. This extension is essential to estimate the shadow costs of regulations
that limit water extraction or pollution. The issue of industrial water use has been
largely absent in the journals of the economic profession. Widening the search
reveals only a few more contributions to this topic. Noteworthy are two recent
research papers by Dupont and Renzetti (2003) and by Dachraoui and Harchaoui
(2004). Both analyze the valuation of water for the Canadian industrial sector
using a translog-cost framework but do not address the issue of water pollution.
A general approach for the translog production function estimation
framework
For the estimation of the macroeconomic production functions we will later on
employ a simple Cobb-Douglas specification, but we will at this stage generalize
the theoretical estimation approach in order to have a flexible framework for the
estimation of the microeconomic results. Within this translog-framework we are
able to analyze sector- and region-specific production functions in Germany. We
further abstract from restrictive frameworks commonly used. Thus, instead of
imposing linear homogeneity or separability a priory, we extend the more flexible
framework proposed by Kim (1992). We allow for variable returns to scale and
non-homothetic production technologies and generalize the framework by
including quasi-fixed factors of production. This extension is essential to estimate
the shadow costs of regulations that limit water extraction or pollution. In the
following section we want to outline the particular features of our specification of
the input inverse demand framework proposed by Kim (1992). While his
framework already allows for a non-homothetic production technology and
variable returns to scale we additionally focus on the implications of including
quasi-fixed production factors. Our special interest lies in determining their
marginal shadow values to the producer. The generalized production function has
the following form:
Yn = f [X n , X n , T ]
(I)
9
Yn is the output level of firm n with X n being the vector of variable production
factors, X n the vector of quasi-fixed factors of production and T the time index
used to measure technological change. Writing this production function in its
translog specification yields the following expression:
ln Yn = α n , 0 + ∑i α n ,i ln X n ,i + ∑ k α n , k ln X n , k +
1
∑ ∑ β n,ij ln X n,i ln X n, j
2 i j
1
1
β ln X n ,i ln X n , k + ∑ k ∑l β n , kl ln X n , k ln X n ,l
∑
i ∑ k n , ik
2
2
1
+ ∑i δ n ,iT ln X n ,iT + ∑ k δ n , kT ln X n , kT + δ n ,T T + δ n ,TT ln T 2
2
+
(II)
The indices i and j identify the different variable inputs and k and l the quasi-fixed
factors. For the estimation we impose symmetry on βn,ij = βn,ji , βn,ik = βn,ki and
βn,kl = βn,lk . We know that for the general model in (I) the first-order conditions
from the output-maximization under a given expenditure-constraint are:
∂Yn
= λn pn ,i
∂X n,i
note that for competitive firms:
∂En ∂Cn 1
=
=
∂Yn ∂Yn λn
(III)
Where pn,i is the price of the ith input and the Lagrange multiplier is the reciprocal
of the marginal expenditure necessary to increase the output in the short run. With
En being the total expenditure and Cn being the variable costs within the
expenditure it is trivial that the marginal cost increase is equal to the marginal
increase in the variable costs. For the maximization problem the expenditure
constraint as well reduces to its variable expenditures counterpart:
En = ∑i pn ,i X n ,i + ∑ k pn , k X n , k
⇒ C n = ∑ i pn , i X n , i
(IV)
Solving equation (III) for pn,i , then substituting it in equation (IV) and finally
solving for λn yields
λn =
1
∂Yn
X n,i
∑
i
∂X n ,i
Cn
(V)
The inverse input demand follows from substituting (V) back into (III) and
solving for pn,i
10
pn , i
∂Yn
∂X n,i
=
Cn
∂Yn
X
∑ j ∂X n, j
n, j
With ∂Yn = Yn∂lnYn and ∂Xn,i = Xn,i∂lnXn,i the share of variable costs Sn,i
input i is then
∂ ln Yn
X p
∂ ln X n ,i
S n,i ≡ n,i n,i =
∂ ln Y
Cn
∑ j ∂ ln X n
n, j
(VI)
spent for
(VII)
The derivation of the marginal products from equation (II) is obvious. Substituting
these back into the share equation yields the appropriate term for the estimation
S n,i =
∑α
j
α n,i + ∑ j β n ,ij ln X n , j + ∑ k β n,ik ln X n, k + δ n,iT T
n, j
+ ∑i ∑ j β n,ij ln X n , j + ∑i ∑ k β n ,ik ln X n, k + ∑ j δ n, jT T
(VIII)
Compared to the formulation without quasi-fixed productions factors this
expression gains the additional terms containing the sums over the logs of the
quasi-fixed inputs. For efficient estimates, common practice suggests to
simultaneously estimate the nonlinear multivariate equation system composed of
(II) and the available cost shares (VIII). Kim (1992) points out that cost share can
only move in the unit interval, which violates the normality assumption for the
error terms. Since the cost shares have a logistic-normal distribution it might be
preferable to estimate the following equation if more than one cost share is
available7:
⎡ α n ,i + ∑ j β n ,ij ln X n , j + ∑k β n ,ik ln X n , k + δ n ,iT T ⎤
⎡ S n ,i ⎤
⎡ p n ,i X n ,i ⎤
⎥
ln ⎢
⎥ = ln ⎢
⎥ = ln ⎢
⎢⎣α n , h + ∑ j β n , hj ln X n , j + ∑k β n ,hk ln X n , k + δ n , hT T ⎥⎦
⎢⎣ S n , h ⎥⎦
⎢⎣ p n , h X n ,h ⎥⎦
The estimated system does not loose efficiency since the otherwise excluded cost
share equation can now be used as denominator. Compared to (VIII) this equation
is essentially reduced in complexity. For specific applications it has the additional
advantage that the total variable costs do not need to be known as they cancel
down in the equation. The shadow value zn of being able to vary one of the quasi-
7
The full system with all cost shares cannot be estimated because its variance-covariance matrix is
singular and non-diagonal. This problem is due to the fact that the variable cost shares sum to
unity. It is solved by simultaneously estimating the production function and all but one cost share
equations. Christensen and Greene (1976) showed that the estimation results are independent of
which cost share is excluded for maximum likelihood estimators.
11
fixed factors of production is equal to the marginal reduction in total expenditure
resulting from a marginal variation:
− z n,k =
∂En
∂X n , k
(IX)
Usually this shadow value is derived from the translog cost-function directly. This
is not always desirable, since input prices might not be available but factor
quantities are known. In our analysis of industrial water-use no market exists for
this natural resource and thus this production factor is typically not priced at all.
Water is assigned to the industrial producers by the public authorities.
Nevertheless, it is a common misinterpretation that because of the assignment
process water can be considered a quasi-fixed production factor. Since the water
is almost never consumed in the production process, the total amount of water
employed in the industrial facility can be increased by employing the same unit of
water several times within the production process. Observing production
processes on-site and interviewing various production engineers uniformly
revealed evidence that this allows them to exceed the limitations of the
contingent. The difference being that the cycle- or multiple-utilization is more
costly than the one-time usage. To specify our production framework accordingly
we define the first variable production factor as the water used in the production
process:
ln X n ,1 ≡ ln (ρWn + Wn )
(X)
In this specification Wn is the amount of water that the company is allowed to
extract according to its water contingent and ρWn is the re-use equivalent of this
extracted water. This term is included if and only if the contingent is binding8. Wn
is then either – if the contingent is binding – the additional amount of water
employed in the production process or – if the contingent is not binding – the total
amount of water that is used in the production process9. The conversion of the
fixed water contingent into the re-use equivalent ρWn accounts for the fact that
the primary extracted water has different evaluation for the production process
than the re-used water. Additional purification techniques have to be applied in
order to recycle the once employed water to its original extraction quality. Thus,
8
If the contingent is binding can easily be checked by comparing the known amount of water
employed in the production process to the also known amount of water extraction. If the relation is
larger than one we define the contingent as binding.
9
It is obvious that within the production process these two sources of water are perfect substitutes
once the already employed water had been purified for further usage. Nevertheless our
specification differs from the approaches used to estimate the shadow value of water, as the so far
used translog-cost functions do not follow out of duality to our specification.
12
we would expect the estimate for ρ to be lager than unity10. The separation of
variable and quasi-fixed production factors is not only necessary to evaluate the
shadow value of the water intake but also to assess the value of polluting the
effluents. From the firm’s perspective the pollutants are a factor of production. If
the discharge is free the firm will emit until no additional pollution results from
the production process any longer and would have to be intentionally (and costly)
produced. An obvious method to reduce the pollution discharge would be to
sanction it by imposing a fee. However, for effluents it is also common to directly
regulate the pollution level11. Since industrial producers have no incentive to
further reduce their pollution discharges12 and legal limits are low enough to be
binding, we can consider effluent pollution as quasi-fixed. Returning to the
shadow value of pollution in equation (IX) we need to address the fact that we
estimate the production function and not the translog-cost function. By simple
manipulation and using the conditions derived in equations (III) and (V) we can
express the shadow value as:
− z n,k =
∂E n
∂X n , k
∂Yn
∂X n , k ∂Yn ∂X n , k
C
=
=
= n
∂Yn
λ
X n,k
∂En
∂ ln Yn
∂ ln X n , k
Yn ∂ ln Yn
X n ,i
X n ,i ∂ ln X n ,i
Yn
∑
i
(XI)
Deriving the first-order conditions from the translog-production function in (II)
and substituting them in the above equation we can formulate the shadow value of
a quasi-fixed production factor as a function of the estimated parameters, the
factor inputs and the variable costs:
10
Regarding the interpretation it might be more intuitive to choose the notation as
ln X n,1 ≡ ln(Wn + ρWn ) . This wouldn’t change the results but would require further deviation from
the standard notation used for translog-production functions.
11
If the companies discharge their effluents into public sewage treatment systems they have to pay
waste-water fees. For direct-discharge (usually into a surface water body), e.g. of cooling water,
they are usually within their contingent-restricted pollution discharge. Further details can be found
in chapter 1.1 of this work. Typically the local public environmental office will set the individual
limits for the company by balancing legal regulations, local economic conditions, demands of
other economic entities wanting to pollute the resource or preferring it unpolluted, specific
characteristics of the polluted resource and political orientation of local authorities (reflecting the
local orientation of the population). Close monitoring hardly allows deviation to higher levels than
the absolute pollution discharge intended by the environmental office.
12
Companies often claim to be environmentally friendly. They validate that by various more or
less official labels on their products that proclaim the environment-friendliness. This might lead to
the conclusion that pressure through the consumers can result in a reduction of pollution. For the
investigations preceding this work we had a special focus on this point. The surprisingly honest
message we got from all the company representatives that we interviewed can be concluded in two
essential facts:
1st: The causality is the other way around. Since the companies are so strictly regulated they fulfil
the requirements of the environmental labels anyway.
2nd: The consumers do not reward environmental friendliness to an extent which would justify
further efforts in the reduction of pollution.
13
− zn , k =
Cn
X n, k
α n , k + ∑i β n ,ik ln X n ,i + ∑l β n , kl ln X n ,l + δ n , kT T
∑ j α n, j + ∑i ∑ j β n,ij ln X n, j + ∑i ∑k β n,ik ln X n,k + ∑ j δ n, jT T
(XII)
To determine the variable costs it is appropriate to assume the pollution discharge
to be free-of-charge ∑ k pn , k X n , k = 0 . Thus, it follows from equation (IV) that
En = Cn. Likewise the shadow value of being able to increase the water contingent
is13:
− z n,Wn = ρ
Cn
X n ,1
∑α
j
α n,1 + ∑i β n ,1i ln X n ,i + ∑l β n ,1k ln X n,k + δ n ,1T T
n, j
+ ∑i ∑ j β n,ij ln X n, j + ∑i ∑k β n ,ik ln X n ,k + ∑ j δ n , jT T
From equations (VI) and (X) it is obvious that − zn, X n ,1
(XIII)
= pn, X n ,1 . Hence, the
shadow value of increasing the contingent is higher than the water costs if ρ is
lager than unity.14.
Macro data estimation results
The data for the analysis is provided by the Federal Statistical Office of Germany
and covers the essential macroeconomic indicators plus the water usage by the
industry for the sixteen German states from 1990 until 2007. Table 1 shows the
estimation results using various specifications for the econometric model and
assuming a Cobb Douglas production function. The results differ only marginally
over the different specifications and thus we want to focus our interpretation on
the estimates of the nonlinear model as specified by equation (II) and (X). Due to
the log-log specification the coefficients can be directly interpreted as elasticities.
These elasticities will be used in our simulation model later on. Due to the simple
form of the production function that we have chosen for these estimates it is easy
to cross-check the plausibility of the results by deriving the prices for the
production factors which follow from the primal production function. These
prices are listed in the last column of table 1. The costs of capital and costs per
employee are well within a plausible range and close to the actual values observed
in Germany but our main interest lies in the water costs. Jessberger and Zimmer
(2009) list German average effluent charges in 2005 with about 2.28€ for each
cubic meter of water and the 2007 average costs for public water with 1.85€
(which serve as a good indicator of extraction and supply costs for the public
∂ ln Yn
∂ ln Yn
with
in equation (IX).
∂Wn
X n , k ∂ ln X n , k
13
This is easy to prove by substituting
14
Note that water contingent is only binding if Wn < X n ,1 . Since the water costs for multiple- or
cycle-use are likely to be higher, or at least as high as the extraction costs for freshwater, this
theoretical result predicts that the production sites that are bound by the contingent constraint have
a higher shadow value for a marginal increase in the contingent.
14
water suppliers) which sums to more than four Euro of total costs for each cubic
meter of water. You would expect that self supplying water would only make
sense if it reduces costs compared to being supplied by a public water supplier and
indeed the estimated costs of 2.85€ per cubic meter lie well below the four Euro
of the public water supply. The estimated shadow value of increasing the
extraction contingent by one cubic meter is as expected above the average water
costs. This figure gives an impression on how much the water is indeed worth in
the production process. Since due to the legal constraints the industrial producer
have to substitute water in the production process by other production factors,
they would be willing to pay an additional 8.53€ for an increase of the contingent
by one cubic meter of water.
CobbDouglas
Model
(1)
(2)
(3)
OLS
OLS
OLS
employees
0.549**
0.548**
0.533**
capital
0.406**
0.407**
0.315**
used water
-
0.056**
-
extracted water
0.056**
-
rho
(shadow value
parameter)
-
year
region dummies
constant
(4)
(5)
(6)
Derived
prices from
OLS NL-model NL-model model (6)
27,542€
per year
0.526** 0.549** 0.530**
and
employee
6.0%
0.313** 0.406** 0.315**
interest
rate
2.85€
0.047**
per
cubic meter
0.054**
-
0.056**
0.052**
-
-
-
2.812**
2.990**
0.010**
0.010**
0.012**
0.012**
0.010**
0.012**
no
no
yes
yes
no
yes
8.53€
per
cubic meter
-13.751* -13.964* -16.012** -15.499** -13.915* -15.883**
R2
0.9850
0.9850
0.9996
0.9996
0.9854
0.9996
observations
160
160
160
160
160
160
Table 1: estimation results; 1% (**) and 5% (*).
It should be noted that our estimates for Germany are well above those of Dupont
and Renzentti (2003) for the Canadian manufacturing. Their estimates range
between 0.34 cent/m3 and 6.17 cent/m3 with an average shadow value of
15
0.62 cent/m3 (1991 Canadian $). The estimates of Dachraoui and Harchaoui for
the shadow value of water utilization including recirculation range between
−0.37 $/m3 and 1.02 $/m3 with an average of 0.57 $/m3 for the water-intensive
industries (Canadian $).
Validation of the model
For calibrating our model we use a 5-year-periode between 2001 and 2006. Using
the aforementioned estimated coefficients of the macroeconomic model our
industrial model simulates Gross Regional Products (GRPs) of each German and
Austrian county within the scope of a reasonable 95 percent quantile of maximal
+3.24 and -11.30 percent (with -3.62 percent mean) of deviation compared to the
statistical numbers15 in 2006. All deviations of each county are sorted and plotted
in figure 3.
Figure 3: Sorted deviations of simulated GRPs in 2006 of a 5 year simulation period
25%
15%
deviations
5%
-5%
-15%
-25%
-35%
-45%
counties sorted by deviations
Source: Ifo Institute
The extreme but isolated deviations can be explained by artificial regional specific
production shocks of a single county in that period which cannot be reproduced
efficiently in our simulation model. But with regard to computation time
efficiency and computing power and as the majority of all deviations are very
close to the mean we assume that our model can simulate future GRPs with a
tolerable precision.
15
Federal Statistical Office of Germany, 2008
16
Economic and social scenarios in frames of climate change
Two contrasting scenarios – liberalization of trade, following a globalizing world;
and sustainability, as a growing environmentalism in the society - and a baseline
scenario have been invented by the whole project team. The industrial model
directly acts and responses to these scenarios.
The economy in the baseline scenario
In the baseline scenario the industrial model simply reacts as it is compiled and
validated with current economic behaviour and development. Here we employed
our regression as mentioned above. All five adjusting screws are in mid-position
to obtain a benchmark for the other scenarios. These screws are “investment costs
for re-using water”, “costs for extracting water”, “subsidies for environmental
protection”, “cost of capital”, and “labour costs”. In the liberalization-scenario
and Sustainability-scenario the screws are changed according to the following
table.
Table 1: List of screws for the industrial model
Screw declaration
Liberalization
Screw
ChangeCostOfWaterReuse
investment costs for re-using
Sustainability
scenario
scenario
constant
decreasing
water
ChangeCostOfExtraction
costs for extracting water
constant
increasing
ChangeSubsidies
subsidies for environmental
constant
increasing
protection
ChangeCostOfCapital
cost of capital
decreasing
decreasing
ChangeWages
labour costs
constant
increasing
Source: GLOWA Danube scenarios, GLOWA Danube project
Industrial model in the liberalization-scenario
The configuration of the screws for the liberalization-scenario is based on the
following facts.
Germany is rich in water resources. A long term mean of 188 billion cubic meter
water are available per year whereas the total water consumption only adds up to
35.6 billion cubic meters. Only 0.8 of the 5.4 billion cubic meters public water
supply are consumed by the industrial production sector. In other words as about
81 percent of water supply is not consumed German water resource conditions for
industrial consumption are convenient today and in the future. Due to that stable
and continual growth is assumed in this scenario and the costs for extracting water
stay at a moderate rate. Also very little investments in water re-usage technology
are expected. Investment costs for re-using water stay at a high level because only
few subsidies of environmental protection are assumed. To get feeling of the
17
development of future costs for extracting water we use public drinking water
prices as a benchmark:
Table 2: sewage charge prices conform to the fresh water benchmark weighted by habitants
€/m3 2002
€/m3 2005
Change
p.a.
Old West German states
2,05
2,16
5,4%
1,8%
Newly-formed
2,47
2,87
16,2%
5,1%
2,11
2,28
8,1%
2,6%
German states
Germany
Source: BDEW
Table 3: mean water prices in Germany in 2007
€/m3 2001
€/m3 2007
Change
p.a.
Old West German states 1,64
1,79
9,1%
1,5%
Newly-formed
2,05
2,15
4,9%
0,8%
1,70
1,85
8,8%
1,4%
German states
Germany
Source: BDEW
As the public water supply operates cost-covering we assume that the costs for
industrial water consumption will be similar. Likewise we assume moderate
development of labour costs because basic conditions are not really changing
compared to those today. The only screw which is changed is cost of capital as a
reaction to the global cost of capital because of the globalization.
Industrial model in the sustainability-scenario
Here governmental general conditions change in terms of increasing subsidies for
environmental protection. Due to that cost of capital for environmental protection
projects decrease. But those subsidies are partly financed by higher labour costs.
Moreover statutory requirements of water usage will become more strictly and the
costs of water consumption increase. For example an increase of 5 Cent could be
possible assuming that in Bavaria there will be established the same additional
price for water extraction as in Baden-Wuerttemberg (called “water-cent”);
according to the following table. It is assumed that the funds of the „water-cent“
will not be used to balance fiscal expenditures but to be invested in earmarked
projects in water intense production plants.
18
Table 4: „water-cent“ per each m3 sponsored amount of drinking water in German states
State Water‐cent Explanations Yearly payments Application of funds Baden‐Württemberg 5,1 since 1988 No earmark Bayern – Berlin 31 ca. 55 Mio. € Protection of ground water Brandenburg 10,2 With two times of ca. 20,2 Mio. € Realization increase since 1994 of WRRL, maintenance of dikes , etc. Bremen Hamburg 5 7 bzw. 8 since 1993, confirmed ca. 0,7 Mio. € of in 4 / 04 WVU For about. 12 years, 3,0 Mio. € vom WVU increased in 12 / 05 Hessen – in 1 / 03 abgeschafft Mecklenburg‐Western 1,8 Updating the water‐
ca. 1,7 Mio. € For ground water Pomerania Niedersachsen 5,1 pfennig of the DDR, sparing confirmed in 1/ 03 arrangements Confirmed in 12 / 04 ca. 20 Mio. € of the For ground water public water supply sparing arrangements Nordrhein‐Westfalen 4,5 Since 1.2.2004 72 Mio. € for drinking water and Realization of WRRL process 2) water (2005) Rheinland‐Pfalz – Schleswig‐Holstein 5 respectively since 1.1.2004 ca. 24,5 Mio. € Earmark reduced to Saarland (6 Proposed (up to 3 Mio. €) (partly earmarked) respectively. introduce by state‐
7) government in 2007 Sachsen 1,5 ca. 3,4 Mio. € earmarked Sachsen‐Anhalt – Thüringen – 11 1) 50 % To 1) 5 Cent for business enterprises as end-consumer if it consumes more than 1.500 m3 of water
in time period, 11 Cent for all other end-consumers
2) Posible to apply against expenditures within the farming cooperation
Source: BDEW
19
Simulation results
As an introduction to the simulation results figure 4 displays the development of
the population as generated by the Demography model.
Figure 4: population development 2012-2025
(GLOWA catchment area framed in red, numbers in percent)
Source: Ifo Institute
The general trend of a declining population in Germany is mainly caused by low
fertility rates and declining immigration. The regional differences of -1% to -9%
are a result of domestic migration which is driven by regional differences in
economic attractiveness and other amenities, like recreational value. For Austria
the situation is less drastic since some regions still benefit from high fertility rates
or stronger immigration.
20
Based upon the Baseline climate trend we calculated three scenarios: a BaselineScenario (Social Megatrend 1 (GMT1)), a Liberalisation-Scenario (GMT2), and a
Sustainability-Scenario. The results of the Liberalisation-Scenario and the
Sustainability-Scenario were derived from local simulation runs of the interacted
Demography- and Economy-Model combined with Dummyfiles providing the
input from the remaining models. The Baseline-Scenario on the contrary describes
results of a joint run combined with demography, groundwater, groundwaterflow,
groundwatertransport, houshold, landsurface, tourism and watersupply. The
results are shown and described as differences of the gross domestic product and
the groundwater-demand, over a period of 2012 to 2025. The Baseline-Scenario
describes an economic development, which results from continuing todays trends.
Therefore it serves as reference for evaluating the differences in the developments
compared to the other scenarios (GMT2 and GMT3).
Development of industrial groundwater-demand
At first we take a look at the development of the industrial groundwater-demand
from 2012 to 2025, as shown in the figure 5. It doesn’t show the difference in the
development between two scenarios but displays the absolute difference between
2012 and 2025 as simulated in the Basline-Model. The map shows the
percentaged change of industrial groundwater-demand on every square kilometre.
Comparing the years 2012 and 2025 the differences in the industrial groundwaterdemand rang from a minimum of -50% to a maximum of +15%. This corresponds
to annual average growth-rates from -1.35% to +1.19%. It is easy to discover
broad local differences in the changes in the industrial groundwater-demand.
Because of the climate induced decline in groundwater-supply the largest
reduction of groundwater usage is observed in the Pre-Alps. In the northern part
of the river basin the situation remains relaxed. A climatic decline of the industrial
groundwater-demand can not be noticed in that area. Looking at the cities
Augsburg and Salzburg, which are located in the German respectively the
Austrian Pre-Alps, the different reactions of industrial groundwater-demand to the
climatic changes, are obvious. Both regions are significantly affected by the
differing groundwater-supply.
21
Figure 5: industrial groundwater-demand 2011-2025 GMT1
(Values in percent)
Source: Ifo Institute
For a closer look on the regional discrepancies in climate conditions and the
region’s reaction we expose a German and Austrian city (Augsburg and Salzburg)
as a benchmark. In Augsburg temporary the groundwater-supply is relaxing
between the years 2017 and 2018. But until 2025 ground water conditions finally
become so precarious that the regional industry reduces its ground water demand
considerably by more than 25% compared to 2012. Primarily this reaction is
based on the very close movement of groundwater demand and lagged
groundwater conditions (cf. figure 6).
22
Figure 6: Change in industrial ground water demand in comparison to 2012 and 1 year lagged
ground water conditions in Augsburg.
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
0
-5%
1
-10%
2
-15%
-20%
3
-25%
ground water conditions
Change in industrial ground water
demand in comparison to 2012
0%
4
-30%
industrial ground water demand
ground water conditions (1 year lagged)
Source: Ifo Institute
In contrast the reactions to the industrial ground water demand in Salzburg is
another but as there are precarious ground water conditions in 2025 (similar to
Augsburg) the industry reduces it ground water demand nearly as much as the
industry in Augsburg. But comparing the years between 2017 and 2023 the
ground water conditions in Salzburg are notably better and consequently the
industrial ground water demand can stay at a same high level as in 2012.
Figure 7: Change in industrial ground water demand in comparison to 2012 and 1 year lagged
ground water conditions in Salzburg.
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
0
-5%
1
-10%
-15%
2
-20%
-25%
3
-30%
industrial ground water demand
ground water conditions (1 year lagged)
Source: Ifo Institute
23
ground water conditions
Change in industrial ground water
demand in comparison to 2012
0%
This shows up the regional discrepancies in in German and Austrian ground water
conditions, and respectively industrial ground water demand, influenced by
different climate change effects in the alpine upland.
Comparison of the scenarios for industrial groundwater-demand
Figure 8 shows the difference in the relative change between the social scenarios
Liberalisation (GMT2) and Sustainability (GMT3) for the industrial groundwaterdemand for each industrial populated square kilometre from 2012 to 2025. This
means to subtract the absolute percentage change over the whole period in the one
scenario from the one in the other scenario. The such generated figure is the
difference between the two scenarios. It can be interpreted as an indicator
measuring e.g. the effect of a difference in the assumptions about climate
development. Thus, positive values describe a higher water demand in the
Sustainability-Scenario (GMT3), while negative percentages indicate a lower
industrial groundwater-demand in the Sustainability-Scenario (GMT3). On a
percentage basis the differences in the GLOWA river basin range from -3.1% to
0.0%.
This translates to an annual difference range which lies approximately between 0.24% and 0.0%. The GLOWA river basin is dominated by negative effects.
Because of the interaction with ground water conditions, which are available for
the simulation only for the river basin, a real variety of different reactions is
present just inside the red framed GLOWA Danube river basin. If water
conditions of an industrial used square kilometre are more severe than those of
surrounding areas of industrial production, the production will shift to those areas.
Here the level of reaction discrepancies is amplified by the subventions for water
saving technology, which is higher in the Sustainability-Scenario (GMT3).
24
Figure 8: industrial groundwater-demand 2011-2025 , comparison of the scenarios
(GMT2 vs. GMT3, values in percent)
Source: Ifo Institute
A decline in the industrial groundwater demand is less common but therefore
tends to be stronger. Because of a stonger reaction to the deteriorated
groundwater-supply, the industrial groundwater-demand in the Pre-Alps-region
declines more in the Sustainability-Scenario (GMT3) than it does in the
Liberalisation-Scenario (GMT2). In the following figure these differences of
changes are drawn:
25
Difference of changes between the
liberilasation and the sustainability scenario
Figure 9: Differences in changes in industrial ground water demand in comparison to 2012
between the liberalisation and the sustainability scenario
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
0.0%
-0.5%
-1.0%
-1.5%
-2.0%
German scenario differences of changes in industrial ground water demand
Austrian scenario differences of changes in industrial ground water demand
Source: Ifo Institute
Development of the gross regional product
On the basis of the GMT1-results we show the dimensions in which industrial
players adjust their production to the changing environment. According to this,
the map shows the percental difference of the gross regional product on every
square kilometre.
26
Figure 10: Gross regional product 2011-2025 GMT1
(Values in percent)
Source: Ifo Institute
Comparing the years 2012 and 2025 we find a difference of the gross regional
product between -0% and -6%. This would correspond to an annual average
changing rate of -0.00% to -1.15%. The real decline of the gross regional product
of up to 6% is forced by both the decline in population and the resulting decline in
employable population. The domestic migration towards Southern Germany has
just a calming effect as well and is unable to compensate the declining populationtrend. The lager decline of the gross regional product in the Pre-Alps is caused by
poor groundwater-conditions and becomes again apparent, even though the
decline is not very strong. The coherency between the population and the gross
regional product becomes obvious by looking at the German alpine upland (here
27
using the example of Augsburg). The declining ground water demand plays an
insignificant role here.
Figure 11: Change of Gross Regional Product and Population in comparison to the basis year 2012
in Augsburg
2012 2013 2014
2015 2016 2017 2018
2019 2020 2021 2022
2023 2024 2025
1%
-1%
-2%
-3%
-4%
Change to 2012
0%
-5%
-6%
Gross Regional Product (GRP)
Population
Source: Ifo Institute
But the deteriorating ground water conditions and therefore the declining
industrial ground water demand have a negative impact on the gross regional
product in Austria (respectively Salzburg), even though there is a constant
increase in population.
Figure 12: Change of Gross Regional Product and Population in comparison to the basis year 2012
in Salzburg
2015 2016 2017 2018
2019 2020 2021 2022
2023 2024 2025
8%
6%
4%
2%
0%
-2%
-4%
-6%
Gross Regional Product (GRP)
Source: Ifo Institute
28
Population
Change to 2012
2012 2013 2014
According to this fundamentally different effects are responsible for the decline of
the gross regional product, but lead to the same result eventually.
Comparison of the senarios for the gross regional product
Figure 13 illustrates the difference of the growth rate of the gross regional product
on every square kilometre from 2012 to 2025, in comparison with the two social
scenarios Liberalisation (GMT2) and Sustainability (GMT3). Therefore positive
values represent a stronger growth of the gross regional product (GMT3) in the
Sustainability-Scenario, while negative values in the same scenario indicate a
slower growth. The main difference in the growth rates during the entire period of
13 years ranges between -1.6% and 2.6%, which equals an annual difference of 0.12% and +0.20%, respectively.
29
Figure 13: gross regional product 2011-2025, comparison of the scenarios
(GMT2 vs. GMT3, values in percent)
Source: Ifo Institute
Although there aren’t a lot of industrial populated square kilometres in the Alpsregion, a stronger negative growth of the gross regional product predominates in
the Sustainability-Scenario (GMT3) in comparison with the LiberalisationScenario (GMT2). At this point the stronger reaction to the worsened
groundwater-conditions is again responsible, because the production shifts to
those square kilometres, where the ground water resources take a better
development over the years. That is to say, areas with sufficient water benefit (or
suffer less) from a climatic change of the ground water availability, because
industrial production will be indeed shifted to those areas.
30
Development of water saving technologies
Figure 14: Industrial water technology 2011-2025 in baseline scenario
(Values in percent)
Source: Ifo Institute
This picture displays the change in the water usage factor of a firm per square
kilometre. This factor indicates the relative level of re-used water of total water in
the production process. Only some dots in the picture inside the red framed
GLOWA Danube river basin signal a negative growth rate of this factor (red: -0.6
to 0 percent growth). The most firms can improve their water usage in terms of
saving fresh water by substituting it with recycled or multiple employed water
which is passed back into the production line. However just the dark green square
kilometres indicate a firm which shows a growth higher than 0.2 percent
31
Conclusions
We developed a methodology to econometrically investigate the value of water
resources for the industrial production. The framework incorporates regulated
production factors and thus is especially suited to analyse the shadow value of the
water contingents in Germany. We solve the conflict of previous theoretical
models by combining fixed water contingents with the possibility to use the
resource multiple times or in a cycle. Thus the company is free to efficiently
choose the amount of water employed while the value of the water contingents
can still be determined. For the German industry the costs of water are estimated
to be around three Euros and the shadow value of the water contingents to be
around eight and a half Euro per cubic meter of water. Therefore the estimate for
the costs of the self-supplying industrial sector lie below the known supply and
recycling costs of the public water suppliers which range above four Euros on
average. We use the estimated coefficients to calibrate our industrial model (as a
part of the regional decision support system GCDSS-DANUBIA) and to simulate
the effects of differing climate and policy scenarios until the year 2025. The
results show a general decline in the water usage accompanied by a worsening of
the conditions of the natural water-cycles while large regional disparities in the
analyzed Upper Danube river basin can be observed. The results allow the
identification of regional hot spots and to quantify the effects of various policy
measures. This analysis only states very few exemplary results to give any
potential stakeholders an idea of the capability of the decision support system.
32
Appendix I: estimation results used for the calibration of the simulation
Table A.1: coefficients used for the calibration of the simulation
(1)
(2)
Derived
Model
prices from
OLS
OLS
model (2)
year
0.010**
employees
0.549**
capital
0.406**
extracted water
0.056**
region dummies
no
constant
0.012**
27,698€
per year
0.533**
and
employee
6.0%
0.315**
interest
rate
3.07€
0.054**
per
cubic meter
yes
-13.751* -16.012**
R2
0.9850
0.9996
observations
160
160
Source: Ifo Institute
The coefficients in table A.1 for the calibration of the model were estimated in
Jessberger, Zimmer (2009). Due to the log-log specification the coefficients can
be directly interpreted as elasticities. The implicit prices for the production factors
which follow from the Cobb-Douglas specification of the production function are
listed in the last column of table 1. As seen in table 3 in the scenario chapter the
German average effluent charges in 2005 were about 2.28€ for each cubic meter
of water. Since public water suppliers in Germany charge their water on a nonprofit base according to their extraction and supply costs, we can use their prices
as an indicator for the extraction cost of the self supplied industrial producers. The
2007 average costs for public water were 1.85€ as indicated in table 4. As
expected the estimated costs of 3.07€ per cubic meter lie well below the roughly
four Euro of the public water supply.
33
Appendix II: Implementation of the Agent-Based Industry Model in GCDSSDANUBIA
To implement the industrial agents in the DANUBIA environmental decision
support system the economic model had to be constructed according to the
requirements of the DeepActor-Framework (see: Mauser, Janisch, … 2007, …).
The DeepActor-Framework is the core of all agent-based socio-economic models
in GCDSS-DANUBIA. It is itself part of the framework that links and coordinates
all the discipline-specific sub-models. The following UML-diagram illustrates the
coding of the industrial agent. The industrial agent classes (DAI) extend the
superordinate abstract classes of the DeepActor-Framework. They add to the
Abstract-ActorModel, AbstractActor, AbstractPlan and AbstractAction the
attributes and methods specific to the industry model16. The DAI_Model class
contains the initialization data and stores the updated values in each simulation
period.
16
UML refers to unified modelling language, a notation convention common to computer science
applications. Since the general intuition of the illustration is accessible without deeper knowledge
of the terminology we will abstract from a detailed introduction into object-oriented programming.
In the nineties, due to the increasing complexity of computer programs, object-oriented
programming replaced traditional approaches based on sub-routines. In general it is sufficient to
know that in contrast to traditional programming, object-oriented programs simply consist of a
collection of objects. Typical types of objects are classes which are represented by the coloured
squares. These classes have a name at the top of the square and contain several methods listed at
the bottom. They can for example extend (here being a specialisation / arrows with the hollowed
triangle as a spike) a more general class provided by the framework or another class specific to the
demography module (arrow spiked arrows). They can draw on collections of module specific
standard routines (arrow spiked dashed arrows) which for example enable the interchange of data
between the discipline specific sub-modules via the framework-interfaces or simply read (and
write) in a data-files object (arrow spiked dashed arrows with <<use>>). The idea of object
oriented programming gets clear if you imagine you and your boss in your institute. Let’s assume
there exists the task to produce a printed essay. Then, the institute with everything in it would be
the program to solve this task. To do so it can make use of all the more or less helpful objects
inside the institute. These objects include you, your boss, a computer, computer programs, a
printer, a paper bin and some flowers. Now, the capability of your boss to write an essay is
extended to the capability to write a nicely formatted essay in print letters by his usage of the
computer and the computer programs (methods) within it. You however extend the capabilities of
your boss to also do all this annoying data preparation (standard routines) which then will be
included in the essay as part of your boss’s achievement. To perform those standard routines your
capabilities are as well extended by the usage of the computer and the computer programs and
those programs again can use the existing data-files for your data-preparation (read the data-files).
While looking at the flower object might distract you from your frustration and thus speed up the
process it is preferable if you don’t have to use the paper bin object for the final storage of the
essay. Rather you just want to use the printer method to store the essay (write it into a data-base
file) and thus complete the task of the institute.
34
Figure A.2: UML illustration of the industrial agent (own illustration).
Source: Ifo Institute
The DAI_ActionEnv interface contains the methods the agent (DAI_Actor) needs
to perceive his environment and communicate his planned factor employments
and production to the DAI_Action (representing the production facility). The
EconomyToActorController interface is used to export data to the other disciplinespecific
sub-models
in
the
framework.
Correspondingly,
the
ActorControllerToEconomy allows data import. The macroeconomic model class
DAI_District is embedded in the industrial agent. To enable a user-friendly
adjustment of different scenarios we developed a graphical user interface (GUI)
which allows the user the definition of simple scenarios (Figure A.3). In
particular, this editor allows the definition of the essential elasticities and trends.
35
Figure A.3: Scenario editor for the Economy sub-model17
(GCDSS-DANUBIA economy model).
Source: Ifo Institute
17
Trends are entered in the form of 1.02 for a positive and 0.98 for a negative trend of 2%.
36
References
Bundesverband der deutschen Gas- und Wasserwirtschaft (1994). 105. Wasserstatistik
Bundesrepublik Deutschland – Berichtsjahr 1993. Bonn. Germany.
Bundesverband der deutschen Gas- und Wasserwirtschaft (1997). Wassertarife 1997. Bonn.
Germany.
Bundesverband der deutschen Gas- und Wasserwirtschaft (2001a). 112. BGW-Wasserstatistik
Bundesrepublik Deutschland – Berichtsjahr 2000. Bonn. Germany.
Bundesverband der deutschen Gas- und Wasserwirtschaft (2001b). Tarife 2001. Bonn. Germany.
Chung J.W. (1994) Utility and Production Functions. Theory and Applications. Blackwell.
Cambridge. UK.
Dachraoui, K. and T.M. Harchaoui (2004). Water Use, Shadow Price and the Canadian Business
Sector Productivity Performance. Economic Analysis Research Paper No. 26/2004. Statistics
Canada.
Dupont, D.P. and S. Renzetti (2001). The Role of Water in Manufacturing. Environmental and
Resource Economics, Volume 18, pp. 411-432.
Dupont, D.P. and S. Renzetti (2003). The Value of Water in Manufacturing. CSERGE Working
Paper ECM 03-03.
Egerer, Matthias (2005). Global Change and the Effects on Industrial Water Use in the Catchment
Area of the Upper Danube, Conference paper: 7th Nordic Environmental Social Science Research
Conference. Göteborg University.
Egerer Matthias and Markus Zimmer (2006). Does Global Change Matter? – The case of
Industries in the Upper Danube Catchment Area, Transactions on Ecology and the Environment
98, 75-88.
Egerer, M. and M. Zimmer (2006). Weiterentwicklung des Regionalmodells RIWU zu einem
„tiefen“ Akteursmodell. In: Mauser, W. and U. Stasser (eds). GLOWA-Danubue - Integrative
Techniken, Szenarien und Strategien zur Zukunft des Wassers im Einzugsgebiet der Oberen
Donau. www.glowa-danube.de .
European Environment Agency (Ed.) (2000). Sustainable Use of Europes Water? – State,
Prospects and Issues. Copenhagen. Denmark.
Federal Ministry for Agriculture, Forestry, Environment and Water Management (Ed.) (2003).
Umsetzung der Wasserrahmenrichtlinie in Österreich - Wirtschaftliche Analyse der
Wassernutzung für den Produktions- und Dienstleistungssektor. Vienna. Austria.
Federal Statistical Office Germany (Ed.) (2003). Umwelt – Wasserversorgung und
Abwasserbeseitigung in der Industrie, Fachserie 19, Reihe 2.2, Wiesbaden. Germany.
Federal Statistical Office Germany (Ed.) (2005). Umweltnutzung und Wirtschaft – Tabellen (Band
1) zu den Umweltökonomischen Gesamtrechnungen 2005. Wiesbaden. Germany.
Gispert, C. (2004). The Economic Analysis of Industrial Water Demand: a review. Environment an
Planning C: Government and Policy, Volume 22, pp. 15-30.
37
Griffin, R.C. (2006). Water Resource Economics: The Analysis of Scarcity, Policies and Projects.
The MIT Press: Cambridge. London. UK.
Jessberger, C. and M. Zimmer (2009) Estimating the Value of Water for Industrial Production in
Germany. Working paper
Loucks, D.P. and J.S. Gladwell (1999). Sustainability Criteria for Water Resource Systems.
UNESCO International Hydrology Series. Cambridge. UK.
Lübbe, Eiko (2002) Jahreswasserbericht der Wasserwirtschaft : gemeinsamer Bericht der mit der
Wasserwirtschaft befassten Bundesministerien - Haushaltsjahr 2001, Wasser & Boden 54 (7/8),
2002, pp. 5-13.
Österreichische Vereinigung des Gas- und Wasserfaches (2002) Daten – Wasser; Betriebsergebnisse der Wasserwerke Österreichs, Wien 2002.
Renzetti, S. and E. Elgar (2002). The Economics of Industrial Water Use. Northampton,
Massachusetts. USA.
Reynaud, A. (2003). An Econometric Estimation of Industrial Water Demand in France.
Environmental and Resource Economics, Volume 25, pp. 213-232.
Rothenberger, Dieter (2003) Report zur Entwicklung des Versorgungssektors Wasser, Sektorreport Siedlungswasserwirtschaft für das Projekt “Integrierte Mikrosysteme der Versor-gung”
im Auftrag des Bundesministeriums für Bildung und Forschung, Centre for Innovation Research in
the Utility Sector, Kastanienbaum (Schweiz) 2003.
Statistik Austria (2002) Regionale Wasser-, Abwasser, und Abfalldaten 1980-1999, Statisti-sche
Nachrichten 5, 2002, pp. 374-380.
Statistik Austria (2005a) Volkszählung 2001 – Haushalte und Familien, Wien 2005.
Statistik Austria (2005b) Zukünftige Bevölkerungsentwicklung Österreichs 2005 – 2050 (2075),
Statistische Nachrichten 11/2005.
Statistisches Bundesamt (2003) Bevölkerung Deutschlands bis 2050 - 10. koordinierte Bevölkerungsvorausberechnung, Wiesbaden 2003.
Statistisches Bundesamt (2005) Umweltnutzung und Wirtschaft – Tabellen (Band 1) zu den
Umweltökonomischen Gesamtrechnungen 2005, Wiesbaden 2005.
Youn Kim, H. (1992). The Translog Production Function and Variable Returns to Scale The
Review of Economics and Statistics, Volume 74, pp. 546-552.
Vörösmarty, C.J., Green, P., Salisbury, J. & Lammers, R.B. (2000) Global Water Resources:
Vulnerability from Climate Change and Population Growth. In: Science 14 July 2000, Vol. 289.
no. 5477, pp. 284 – 288
38