and other open marine waters Danish Environ- mental Protection Agency, 1984. The frequency of
the recorded episodes of oxygen deficit was also found to have increased. Both the extent and
frequency of oxygen deficit have increased even more during the late 1980s, this being attributable
to the increase in nitrogen loading of the aquatic environment National Environmental Research
Institute, 1990 – 1997.
The increase in nitrogen loading is attributable to increased production in the agricultural sector
and growth in the amount of sewage effluent. The latter is the result of enhanced population growth
and increased industrial production. The major part of the increase is attributable to increased
agricultural production resulting from general in- tensification of production and to a lesser extent to
the incorporation of marginal land in production Christensen et al., 1994.
The negative environmental consequences of enhanced nutrient loading of the Danish aquatic
environment led Parliament to adopt the Action Plan on the Aquatic environment in 1987 Dub-
gaard, 1990; Rude and Frederiksen, 1994. Among other things, this stipulated that nitrogen discharge
from agricultural sources was to be reduced by
50 while discharges from other sources indus- try, sewage works were to be reduced by 60.
The present article quantifies the change in marine nitrogen loading due to intensification of
agricultural production, general economic growth and general structural shifts in the economy over
two decades from 1965 to 1989. The purpose of this paper is to analyse the background for the serious
situation at the end of the 1980s and place it in a historic perspective with a view to identifying the
main shifts in the agricultural sector and economy that have been responsible for this change in
loading. The intention is to identify the factors responsible for the development and quantify their
respective contributions. The analysis follows a top-down approach, focusing on the main tenden-
cies on both the environmental and economy sides.
The analysis is made at the national level because the political goals are set at this level. Nitrogen
loading is generally too high in the majority of the inner Danish marine waters and the adjoining
marine waters
2
Christensen et al., 1994, and the political goal of halving nitrogen loss applies uni-
formly to the whole country. An analysis at the national level is therefore meaningful.
The analysis was carried out by combining an input-output model with a nitrogen budget for
agriculture and emission factors for sewage effluent. This comprehensive integrated model sys-
tem enables behavioural and technological shifts in the household and production sectors to be related
to changes in nitrogen loading, thereby revealing the economic reasons for the development.
This study distinguishes itself by relating nitro- gen loading to the whole economy, i.e. to produc-
tion as well as consumption activities, and by applying input-output structural decomposition
analysis on a new area. Furthermore, it covers a very long period, analysing input-output tables
from 1966 to 1988, whereby it is possible to examine changes during three separate decades. A final
distinguishing feature is that the study includes an assessment of the importance to nitrogen loading
of foreign trade.
2. Model and data
The model consists of two parts, an economic input-output model and a nitrogen sub-model. The
latter describes nitrogen loading from the different economic sectors, while the former describes inter-
sectoral commodity flows within whole economy. The nitrogen flows from the nitrogen sub-model are
linked to the input-output model by estimating emission coefficients, i.e. nitrogen discharges per
unit of production per year from all sectors in the economy.
2
.
1
. The input-output model The model framework is the extended input-
output system, as introduced by Leontief and Ford 1972. The strength of the model is that it
2
While nitrogen is the primary factor causing eutrophica- tion of marine waters, phosphorus is the main cause of eu-
trophication in inland waters. Consequently, only marine nitrogen loading will be considered in this paper.
covers all sectors of the economy, operates on a very disaggregated level, and handles both direct
and indirect use of goods. Structural decomposition analysis has been
widely applied to changes in output, use of primary factors, etc. Fujimagari 1989, Forssell 1989,
Skolka 1989. With regard to environmental per- formance, energy consumption has been decom-
posed in Denmark by Pløger 1984, in the USA by Rose and Chen 1991, Boyd et al. 1987, in
Taiwan by Chen and Rose 1990, Li et al. 1990, in Singapore by Liu et al. 1992, and in five OECD
countries by Howarth et al. 1993. In addition, Lin and Polenske 1995 have decomposed energy-de-
mand changes in China, while Lin 1996 has examined the technological and energy-demand
implications. With regard to decompositional anal- ysis of emissions, CO
2
emissions have been decom- posed in Australia by Common and Salma 1992,
in nine OECD countries by Halvorsen et al. 1991 in Taiwan by Chang and Lin 1997, in China,
Taiwan and South Korea by Ang and Pandiyan 1997, in nine OECD countries by Torvanger
1991 and in Denmark by Wier 1998. The latter author also decomposed Danish SO
2
and NO
x
emissions. This paper further expands the environmental
use of decomposition analysis to nitrogen loading, presenting a decomposition of the changes in
nitrogen loading in Denmark over the period 1966 – 1988.
The analytical framework is the static activity x activity input-output model with endogenous im-
port. At a given point in time, nitrogen loading is given by,
N
t
= w
t
I − A
t −
1
D
t
d
t
, 1
where N
t
, is a scalar
3
of total nitrogen loading, w
t
, is a 1 × 117 vector of nitrogen discharges from all
sectors i.e. the output of the nitrogen sub-model per unit of production i.e. w
t
, is a vector of emission factors, I − A
− 7
t
is the 117 × 117 Leon- tief inverse matrix, D
t
is a 117 × 9 matrix for the composition of final demand final demand appor-
tioned by economic activity and d
t
, is a 9 × 1 vector for the absolute level of final demand for all
categories. The nine final demand categories are private consumption, public consumption, exports,
stock building, agricultural breeding stock build- ing, imputed bank service charges and investments
gross fixed capital formation in machinery, in transport equipment and in construction.
The structural decomposition analysis is em- ployed to clarify how the different factors in Eq. 1
have affected nitrogen loading. Total change in loading is decomposed into the effects due to the
four components; loading per unit produced emis- sion factor effect, input mix in production sectors,
commodity mix of final demand and level of final demand.
The analysis was performed by changing the components one by one in order to quantify the
contribution of each effect to the total change in loading. This contribution may be weighted using
either the base year values for the other three components or by the current year values. The
interaction effect is quite large, and will therefore cause considerable bias. The effect of a change in
each component has therefore been determined using an average of the two approaches, as pro-
posed by Fujimagari 1989 and Sawyer 1992.
The change in emissions from sectors given by Eq. 1 from time t − 1 until time t is
N
t
− N
t − 1
= Dw+DI−A
− 1
+ DD+Dd
where the emission factor effect is Dw=[w
t
− w
t − 1
I − A
t − 1 −
1
D
t − 1
d
t − 1
] +
[w
t
− w
t − 1
I − A
t −
1
D
t
d
t
]2 The input mix effect is
DI−A
− 1
= [w
t
I − A
t −
1
− I − A
t − 1 −
1
D
t − 1
d
t − 1
] +
[w
t − 1
I − A
t −
1
I − A
t −
1
D
t
d
t
]2 the composition of final demand effect is
DD=[w
t
I − A
t −
1
D
t
− D
t − 1
d
t − 1
] +
[w
t − 1
I − A
t − 1 −
1
D
t
− D
t − 1
d
t
]2 and the level of final demand effect is
3
Applying element-wise multiplication to the total loading rather than matrix multiplication yields an activity x final
demand matrix. This has been done to provide full informa- tion on changes in all sectors and demands, and will be
referred to, but not presented in this paper.
Dd=[w
t
I − A
t −
1
D
t
d
t
− d
t − 1
] +
[w
t − 1
I − A
t − 1 −
1
D
t − 1
d
t
− d
t − 1
]2 The economic data used for the study are the
Danish input-output tables Statistics Denmark from 1966 to 1988 expressed in constant 1980
prices. The tables encompass 117 sectors and nine categories of final demand. Data on nitrogen in-
puts and outputs are from the Ministry of Agri- culture, Statistics Denmark, the Ministry of
Environment and Energy, and the Danish Envi- ronmental Protection Agency. The data used in
the nitrogen budget is documented in Section 2.2.2.
2
.
2
. The nitrogen sub-model
2
.
2
.
1
. The nitrogen budget at the close of the
1980
s The nitrogen budget for Denmark in the late
1980s is summarized in Fig. 1. The main tenden- cies are described here. This is the first time such
a detailed national scale budget has been drawn up. The greatest uncertainty is related to the
agricultural data, the loading data for sewage and atmospheric deposition being quite robust. The
flows should be interpreted as an average for the period 1986 – 1989. In reality, large climate-depen-
dent variation will occur every year.
The sectors of the economy mainly responsible for nitrogen loading are agriculture, industry and
households, with the former accounting for the major part. Nitrogen balances for the agricultural
sector can be calculated as farm-gate balances or soil-surface balances
4
Schleef and Kleinhanss, 1997. The soil-surface approach was selected for
the estimations in the present study because this is the most appropriate balance when calculating the
losses relating to nitrate leaching to the aquatic environment.
As is apparent from Fig. 1, the main nitrogen inputs are animal and commercial fertilizers.
Other sources of nitrogen include nitrogen fixa- tion, sewage sludge and atmospheric deposition.
Approximately one-third of the deposition takes place as NO
x
nitrogen oxides and two-thirds as NH
x
ammonia and ammonium. The major sources of the NO
x
are the transport sector and power plants, while the NH
x
is mainly derived from volatilization from agriculture. Asman,
1990; Asman and Runge, 1991. According to the estimated nitrogen budget see
Section 2.3.2, 46 of the nitrogen input to the fields is removed with harvested crops. This up-
take is in good agreement with other studies of Danish agriculture based on the soil-surface ap-
proach, where estimates of uptake range from 41 – 44 Nielsen, 1990 to 45 Ministry of Agri-
culture, 1991 to 48 Schleef and Kleinhanss, 1997. Using the farm-gate approach, Andersen
1996 estimated a utilization rate of 49 nitro- gen removed in the sold products expressed as a
percentage of nitrogen input. Nitrogen balances have also been determined for the EU member
states EU 12 by Brouwer et al. 1995 using the farm-gate approach. From that study it can be
estimated that 49 of the nitrogen input was removed in the sold products in Denmark in
19901991. The estimated utilization rate for the Netherlands was 31, the difference between
the rates in the two countries mainly being expli- cable by differences in livestock production and
density. The average utilization rate for the EU member states was 50. The nitrogen utilization
rate in Danish agricultural production is therefore average in a European context, and the estimated
value for Danish agriculture reported by Brouwer et al. is in good agreement with the utilization rate
estimated in the present study.
4
With a farm-gate balance, the interface for the balance is the farm as a whole, all nitrogen inputs to the farm being
calculated, i.e. fodder, livestock and commercial fertilizer, as well as nitrogen fixed from the atmosphere, atmospheric depo-
sition of nitrogen and nitrogen applied in sewage sludge. The outputs are calculated as the nitrogen in all sold products
crops as well as animals and the loss to the environment. When using the farm-gate approach, the internal nitrogen
streams in the farm are not explicitly calculated, i.e. the nitrogen in animal fertilizer and fodder crops. With the soil-
surface approach, in contrast, the interface in the balance is the soil surface. In this case, the nitrogen inputs are commer-
cial fertilizer and animal fertilizer, as well as nitrogen fixed
from the atmosphere, atmospheric deposition of nitrogen and nitrogen applied in sewage sludge. The outputs are calculated
as harvested crops for both internal and external use, and loss to the environment.
M .
Wier ,
B .
Hasler Ecological
Economics
30 1999
317 –
331
321 Fig. 1. Nitrogen budget for Denmark, late 1980s metric tonnes N. It is assumed that nitrogen in animal fertiliser is equal to nitrogen in fodder consumed less the
nitrogen in livestock products. Includes a background load i.e. loading that cannot be attributed to a specific source of 23 000 tonnes N. Approximately 175 000 tonnes are retained in inland waters. Sources: Asman 1990, Asman and Runge 1991, Runge et al. 1991, Ministry of Agriculture 1991, Danish Environmental
Protection Agency 1991a,b, 1992a,b, Nielsen 1990.
As is apparent from Fig. 1, losses to the atmo- sphere take place through denitrification in soil
water and water bodies, soil erosion and ammonia volatilization from livestock housing and the stor-
age and spreading of animal fertilizer. Loss to the aquatic environment takes place as leaching from
the root zone, whereafter the nitrogen is trans- ported to lakes and watercourses and further on
to the sea. As shown in Fig. 1, this amounted to 250 000 tonnes in the late 1980s. However, much
of the nitrogen is retained during transport and, on average, only 75 000 tonnes of the nitrogen
leaching from rural land actually reaches the sea annually, 70, of the total being retained in in-
land waters Danish Environmental Protection Agency, 1991a. Total input to the sea amounts to
110 000 tonnes Danish Environmental Protec- tion Agency, 1991a, including loading from agri-
culture, industry, household sewage, deposition to inland waters, etc.
Atmospheric deposition and inflow from for- eign seas are sizeable, while direct nitrogen load-
ing of marine waters by sewage effluent is minor. For inland water bodies, however, sewage effluent
from households, industry and fish farming may constitute an important source of nitrogen input.
2
.
2
.
2
. Historical de6elopment in the agricultural nitrogen budget
In this section, leaching from the root zone of agricultural land every fifth year from the mid
1960s until the present is estimated according to the nitrogen mass balance principle as the residual
of estimates of all nitrogen inputs and outputs. These estimated levels of leaching are not the
actual leaching in the different years, but an ap- proximation of the expected leaching level given
the actual harvest size, atmospheric deposition, input of animal and commercial fertilizers, etc.,
and assuming that climatic conditions in the en- tire period from 1960 – 1989 were the same as the
average conditions during the late 1980s.
The elements of the nitrogen budget are shown in Table 1 below, inputs being listed in the upper
half and outputs in the lower half. Note that the balance is based on the soil-surface approach.
A considerable number of statistics were needed to calculate the figures, and some assumptions
have been made in order to estimate the develop- ment in input and output flows. Biological fixa-
tion at the end of the 1980s is estimated at 30 000 tonnes Nyear Ministry of Agriculture, 1991.
The amount of nitrogen fixed depends on crop composition, in particular the area planted with
pulses and clover grass, and is calculated on the basis of a constant fixation coefficient per hectare
for each crop type multiplied by the total area of pulses and clover grass. The fixation coefficient is
estimated on the basis of data from the end of the 1980s Ministry of Agriculture, 1991. In other
words, it is assumed that fixation by each crop type is constant per hectare over time, which
implies that the percentage of clover in grass fields is also assumed to be constant. It is difficult to
determine whether the percentage of clover in grass has changed during the early part of the
study period, however. Thus based on the findings of Kyllingsbæk 1995, the percentage of clover in
grass is assumed to have remained constant dur- ing the period 1965 – 1980. In the period from
1980 to the end of the 1980s, the clover content increased Kyllingsbæk, 1995, but insufficient
data is available to enable the increase to be quantified. Kyllingsbæk’s findings indicate that
nitrogen fixation in the 1980s is underestimated in our study.
The figure for nitrogen input in sewage sludge is derived from Kyllingsbæk 1995, who calcu-
lates this to be 3 million kg Nyear in 1970, increasing to 5 million kg Nyear in 1989. Based
on these figures, we assume a constant input of 3 million kg Nyear from the mid 1960s to the
beginning of the 1980s, 4 million kg Nyear in the mid 1980s, and 5 million kg Nyear at the end of
the 1980s. The estimates are subject to a consider- able degree of uncertainty, albeit that this is of
little importance for the budget as a whole due to the small magnitude of these inputs.
Input from commercial fertilizers was taken directly from the agriculture statistics Statistics
Denmark, various years. Nitrogen input from animal fertilizers has been estimated to be 292 000
tonnes Nyear in 19851986 Laursen, 1989, 330 – 340 000 tonnes Nyear in the late 1980s Sibbesen,
1990, 330 000 tonnes Nyear in the late 1980s Ministry of Agriculture, 1991, and 337 000 ton-
M .
Wier ,
B .
Hasler Ecological
Economics
30 1999
317 –
331
323 Table 1
Nitrogen budget for Danish agriculture 1000 metric tonnes Nyear
a
NH
x
-deposition Commercial fertilizers
Animal fertilizer Biological fixation
Input flows Sewage sludge
NO
x
-deposition 30
206 270
19651966 36
3 13
285 34
290 14
19701971 3
34 34
330 285
19751976 31
3 15
315 3
375 19801981
20 28
38 380
30 320
4 22
39 19851986
39 390
320 30
19881989 5
20 Volatization from
Dentrification Leaching from the
Output flows Removed with harvested
Volatilization from comm. Fertil- root zone
crops manure
izers 40
70 19651966
84 13
350 50
160 19701971
80 19
345 50
175 365
19751976 90
21 60
220 19801981
24 98
380 19851986
240 24
60 100
375 19881989
25 100
370 60
250
a
Sources: See Fig. 1.
nes Nyear in 19871988 Nielsen, 1990. As a compromise, input from this source is assumed to
be 320 000 tonnes Nyear in the present study. For the period 1965 – 1985, production of nitrogen
in animal fertilizer was determined from the trend in consumption of animal fertilizer as estimated in
Jensen and Reenberg 1986. It should be men- tioned that these estimates are subject to some
degree of uncertainty. To validate the results, we have also calculated the input of nitrogen in ani-
mal fertilizer based on the composition of the livestock and the current norms for the nitrogen
content of manure per animal Laursen, 1987, which are available for 1975 onwards. The find-
ings with this method only deviate slightly 1 – 7 from the estimates of Laursen 1987 except for
19651966, where the norm-based method yields an estimate that differs from ours by 14.
However, the use of the norm-based estimation method in 19651966 poses considerable prob-
lems. The protein content of the fodder is of decisive importance for the nitrogen content of
the animal fertilizer. Since the protein content increased from the 1960s to the 1970s, figures for
the nitrogen content of animal fertilizer in 1965 1966 based on 1975 norms will be overestimated.
We have therefore chosen not to correct our estimates. It can be concluded, though, that the
uncertainty in the estimates of the nitrogen con- tent of animal fertilizer is greatest in the first part
of the period studied.
Atmospheric deposition of NO
x
and NH
x
in the late 1980s was estimated from emission data for
agriculture, transport and power stations using the TREND model for atmospheric transport and
deposition Asman and Runge, 1991; Asman and van Jaarsveld, 1992. Atmospheric deposition of
NO
x
back through time was estimated on the basis of information on NO
x
emissions CORI- NAIR database, National Environmental Re-
search Institute, deposition being assumed to be proportional to emissions. While this approach
does not take into account regional shifts in the geographic location of emission sources at home
and abroad, only a small part of NO
x
emissions are deposited locally Asman and Runge, 1991
and this
weakness is
therefore of
minor significance.
Atmospheric deposition of ammonia and am- monium back through time was estimated by
assuming that these were proportional to the N content of animal manure and the N content of
commercial fertilizer. The data used was from the end of the 1980s Asman and Runge, 1991; Statis-
tics Denmark various years; Danish Environ- mental Protection Agency, 1991a.
Denitrification was estimated as a constant fraction of total fertilizer consumption 8. Am-
monia volatilization was estimated as constant fractions of animal and commercial fertilizer con-
sumption 31 and 6, respectively. The estimates are based on information from the end of the
1980s Ministry of Agriculture, 1991; Danish En- vironmental Protection Agency, 1991a; Asman
and Runge, 1991.
Note that changes in the handling of animal fertilizer are not taken into account. Depending
on storage conditions and the time and method of application, between 3 and 40 of the nitrogen in
animal fertilizer volatilizes in the form of ammo- nia Danish Agricultural Advisory Centre, various
years. Until the end of the 1980s, there was no decisive change in the utilization of animal fertil-
izer, and this problem is therefore of limited sig- nificance. It should be mentioned, though, that
there has been a general shift from solid to liquid manure which has not been taken into account.
As a consequence, ammonia volatilization at the beginning of the study period has probably been
overestimated to some extent.
Nitrogen removal in crops was assumed to be constant over time per kilogram of different
crops. Information on nitrogen uptake by the various crops derives from Danish Agricultural
Advisory Centre various years, while the infor- mation on harvested crops derives from Statistics
Denmark various years. The uptake coefficients are from 1970, and are assumed to apply to the
whole study period.
An uncertainty factor in the calculation of ni- trogen removal is that there has been a general
increase in nitrogen uptake per kilogram crop Kyllingsbæk, 1995. There is insufficient data to
quantify this increase, however. Therefore, it can be assumed that our figures for nitrogen removal
in the crops are underestimated in 1980 since the
Fig. 2. Marine nitrogen loading from industry and agriculture apportioned by demand 1988; metric tonnes N.
uptake coefficients used are from 1970. It should be mentioned, though, that our estimates are in
full agreement with those of the Ministry of Agriculture 1991.
It should also be noted that nitrogen removal has not increased much during the study period
6, which is in poor agreement with the far greater simultaneous increase in crop yield. This
is attributable to changes in crop composition, primarily due to the decrease in areas planted
with clover grass since the latter takes up three times as much nitrogen per hectare as cereals.
Thus, clover grass fields accounted for 29 of the total area of arable land in 1965 but only
7.5 in 1988.
Finally, leaching was calculated as the resid- ual rounded up. As such, leaching also in-
cluded the non-quantifiable loss in connection with silage and changes in the soil nitrogen
pool.
As there is a considerable time lag in the transport of nitrogen in soil and water bodies
due to chemical and biological processes, 70 of the nitrogen leaching from the root zone was
assumed to be retained in inland waters, with only 30 eventually reaching the sea Danish
Environmental Protection Agency, 1991a.
As is apparent from Table 1, leaching from the root zone has increased considerably since
the mid 1960s especially in the first half of the period. This is due to the increase in nitrogen
input from animal and commercial fertilizers, to- gether with almost unchanged nitrogen removal
in crops. The nitrogen input surplus has more than tripled during the period.
2
.
2
.
3
. Nitrogen loading from sewage effluents The sewage part of the nitrogen sub-model is
quite simple, concerning only loading from indus- trial sectors. Nitrogen loading from sewage works
is measured in a nationwide monitoring pro- gramme, which was initiated in 1988 as part of
the Action Plan on the Aquatic Environment. The estimates back through time were made by cor-
recting 19881989 discharges with data on the development in sewage treatment technology
Tonni
Christensen, Danish
EPA, personal
communication.
3. Nitrogen loading and the economy — results and discussion