J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 429
Ž .
tables documented in Statistics Denmark, 1986 . The tables encompass 117 production sectors and nine categories of final demand. One of the latter is
private consumption, which is divided into 66 components, five of which are direct energy consumption by households.
v
Energy-flow matrices for the years 1966]1992 from Statistics Denmark contain-
ing energy consumption for the 117 production sectors as well as for five Ž
categories of household consumption matrices documented in Statistics Den- .
mark, 1983 . Energy demand is reported for 25 types of energy in both monetary, physical and calorific terms. The latter is used in the present study
as emission factors are given relative to calorific terms. Ž
.
v
Supplementary statistics of renewable energy wood, straw, refuse and others Ž
. from Ministry of Environment and Energy Danish Energy Agency, 1997 . The
energy flows from Statistics Denmark lack a complete inventory of renewable energy, which underestimates energy consumption by households and power
plants. For present purposes the energy flow matrices from Statistics Denmark have been extended with data on consumption of renewable energy, increasing
Ž .
the number of energy types from 25 to 30 cf. Munksgaard et al., 1998 .
v
CO emission factors
for the 22 primary fuels are part of the European
2
Ž .
CORINAIR database Fenhann et al., 1997 . The factors are calculated on the basis of the carbon content of the fuels. Emission factors for the converted
Ž .
energy types electricity, district heating and gas have previously been calcu- lated from the primary emission factors and the energy inputs to the energy
Ž .
production sector Munksgaard et al., 1998 . Finally, CO emission factors for
2
renewable energy types are considered to be zero, as it is assumed that CO
2
emissions from, e.g. straw and wood are absorbed in new biomass production. While the data concerning input]output tables and CO emission factors are
2
quite robust, there is some uncertainty associated with some of the energy flows, especially during the period 1966]1975. Moreover, the description of joint produc-
tion of heat and electricity in power plants suffers from some weaknesses concern- ing estimation of input demand.
In the Danish input]output tables, no specific knowledge on foreign technology is incorporated. Thus, with regard to imported goods, an important assumption is
made: Foreign technology is assumed to be identical to Danish technology. This assumption entails considerable uncertainty, as 20 of the total CO emissions
2
Ž .
3
direct and indirect from households take place abroad.
4. Results
Direct and indirect household CO emissions increased by 7 from 38.6 million
2
tonnes in 1966 to 41.4 million tonnes in 1992. The indirect emissions accounted for Ž
. a major part of this growth } 2.5 million tonnes 15 , while the direct emissions
Ž .
accounted for only 0.1 million tonnes 1 . CO emissions from direct consump-
2
tion have exceeded emissions from indirect consumption during the whole period
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 430
although the difference has been diminishing. In 1992, direct emissions thus accounted for 21 million tonnes whereas indirect emissions accounted for 20
million tonnes. Main emission changes and the driving forces are highlighted in the household
CO emission account in Table 1. As is apparent from the table, the stable direct
2
CO emissions hide two major trends: a sizeable growth in energy consumption
2
and changes in household fuel mix which have partly offset this growth. The increase in indirect CO emissions is primarily due to the general growth in private
2
consumption, whereas substantial energy conservation having reduced energy in- tensity in most production sectors.
For reasons of brevity, the results of the decomposition analysis of direct and indirect CO emissions are reported in Sections 4.1 and 4.2 below at a highly
2
aggregated level. However, information from decomposition performed at a more Ž
. disaggregated level Munksgaard et al., 1998 is referred whenever needed and
some detailed results based on commodity groups are examined in Section 4.3. 4.1. Direct CO emissions
2
As is apparent from Table 2, direct CO emissions remained rather stable over
2
the period 1966]1992 although there was a substantial increase from 1966 to 1976, mainly due to increasing household energy consumption. During the 1970s, how-
ever, substantial energy conservation and shifts towards less polluting fuels modi- fied this effect.
A further decomposition analysis shows that especially demand for electricity and gasoline has increased. Electricity demand has increased continuously through-
out the period due to the introduction of household electrical appliances. From mid 1970s, major energy conservation measures were introduced in the
household sector as a consequence of the rising oil and gasoline prices in 1973 and 1978. This gave house owners an incentive to insulate walls and roofs and led the
construction sector to develop low-energy residential buildings. From the mid-1970s, energy efficient district heating and natural gas were introduced in many single
family homes instead of oil fired heating systems. Moreover, gasoline consumption decreased by 22 between 1978 and 1983. From 1983 to 1992, however, gasoline
consumption increased by 37. Thus in the 1990s, increasing demand for electric-
3
Ž .
In Munksgaard et al. 1998 we explore the importance of this assumption and conclude that the import assumption may have significant influence regarding some of the minor effects in the decomposi-
tion analysis. Nevertheless, all major conclusions are robust. Furthermore, CO emissions from
2
imported goods have constituted an almost constant share of total household CO emissions, thus
2
implying that analyses of changes over time will be less sensitive to the import technology assumption. However, it is important to encompass emissions from import. Hence, when imported goods are
included in the analysis, neither the input mix effect nor the effect from changes in composition of consumption commodity aggregates, have any significant influence on CO changes. However, when
2
excluding imported goods, CO emissions are continuously reduced during the period 1966]1992 due to
2
changes in the two effects. This indicates that demand for CO intensive goods change towards
2
increasing demand for foreign products.
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 431
Table 1
a
Changes in CO emissions from household consumption from 1966 to 1992
2
CO emissions in 1966 38.6
2
Change in direct CO emissions q
0.1
2
Energy consumption level q
3.9 Energy mix
y 2.9
Emission factors y
0.9 Change in indirect CO emissions
q 2.6
2
Consumer beha
¨
iour: Consumption level
q 8.3
Mix between aggregated commodities y
0.7 Mix within aggregated commodities
q 0.1
Total consumer behaviour q
7.7 Beha
¨
iour of the firm: Input mix
0.0 Emission factors
y 0.1
Energy mix q
0.5 Energy intensity
y 5.5
Total behaviour of the firm y
5.1 CO emissions in 1992
41.4
2 a
Note: The discrepancy in the sum is due to rounding off of decimal places in the individual figures.
ity and gasoline may threaten future attempts to control CO emissions from the
2
household sector. Changes in household fuels had little impact on CO emissions, except from
2
1966 to 1977. During this period, most single family homes installed oil fired central heating systems instead of coal and coke ovens. The overall effect was a
13.7 decrease in emissions. Changes in fuel composition in the energy production
Table 2 Decomposition of direct CO emissions
2
Period Energy
Energy mix Emission
Total consumption
factors level
1966]1971 20.2
y 6.2
y 2.1
11.9 1971]1976
8.9 y
0.2 0.7
9.4 1976]1981
y 9.3
y 1.7
2.5 y
8.5 1981]1986
0.8 y
0.4 y
1.9 y
1.5 1986]1992
y 5.2
0.3 y
3.8 y
8.8 1966]1992
18.5 y
13.7 y
4.3 0.5
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 432
Ž .
sector also reduced CO emissions, but only to a small extent 4.3 . Emissions
2
decreased until the early 1970s, whereafter they increased until the mid-1980s. Since then CO emissions have decreased continuously. The explanation is that
2
until the first oil crisis in 1973, power plants were substituting oil for coal. As coal is more CO intensive, this was environmentally beneficial. After the oil crisis,
2
however, the power sector substituted from oil to coal, consequently enhancing CO emissions. Since the mid-1980s, a higher share of natural gas, straw and refuse
2
in energy production has ensured a total decline in emissions. Increased efficiency in the energy sector from increased co-production of heat and power has also
contributed to the reduction in CO emissions.
2
4.2. Indirect CO emissions
2
The indirect CO emissions from household consumption increased by 15
2
from 1966 to 1992, as shown in Table 3. Most of the growth can be explained by the increasing level of private consump-
tion, especially during the periods 1966]1976 and 1981]1986. Over the whole period 1966]1992, growth in the level of consumption generated a 47 increase in
CO emissions. Further decomposition revealed that this is mainly due to an
2
increase in the demand for transportation
4
and consumption associated with recreation and entertainment.
5
Table 3 Decomposition of indirect CO emissions
2
Period Consumer behavior
Total Mix within
Mix between Consumption
aggregated aggregated
level commodities
commodities 1966]1971
y 0.5
y 1.9
14.7 16.8
1971]1976 y
0.3 y
3.9 14.6
y 1.8
1976]1981 1.1
0.7 y
2.9 y
2.9 1981]1986
0.6 y
2.4 18.6
9.5 1986]1992
0.5 1.7
0.1 y
5.7 1966]1992
0.7 y
4.1 47.2
15.0 Behaviour of the firm
Input mix Emission
Energy mix Energy
factors intensity
1966]1971 9.4
y 2.9
0.6 y
2.6 1971]1976
y 10.1
1.6 1.2
y 4.8
1976]1981 3.3
2.9 2.9
y 10.9
1981]1986 y
3.1 y
0.4 0.0
y 3.7
1986]1992 0.8
y 1.1
y 0.4
y 7.4
1966]1992 y
0.2 y
0.4 3.0
y 31.1
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 433
The second most important factor is reduced energy intensity, which generated a 31 decrease in emissions over the period 1966]1992. Many energy conservation
projects were implemented during this period, especially in the second half of the 1970s after the oil crisis
6
. During the periods 1976]1981 and 1986]1992, growth in the overall level of
consumption was rather stable, partly due to reduced consumption of clothing and cars, and CO emissions consequently decreased. This was also the case for the
2
period 1971]1976, as changes in input mix and energy conservation in production sectors together with changes in commodity mix in households resulted in an
overall total decline in CO emissions } despite considerable growth in private
2
consumption. Generally speaking, however, structural effects, i.e. changes in input mix, energy mix, and commodity mix have had little impact.
4.3. Intensity of commodities for household consumption Changes in household commodity mix reduced CO emissions by 4.1. While
2
this effect is minor, analysis at the detailed commodity level shows that the CO
2
effects from changes in commodity mix was substantial. CO emissions from
2
private consumption during the period 1966]1992 are shown for eight aggregated commodity groups in Table 4. Detailed CO intensities for different commodities
2
are shown in Appendix B. CO
emissions associated with the consumption of clothing and household
2
Ž .
appliances including operation decreased by 30 and 10, respectively, between 1966 and 1992. Emissions associated with consumption of the remaining commodi-
ties increased in some cases significantly. Thus, emissions from the consumption of Ž
services mail and telecommunications, law and financial services, private teaching .
and day-care increased by 234, while emissions from consumption of recreation Ž
increased by 59 and emissions from transportation vehicles and purchased .
transport services increased by 54. Emissions from consumption of a given commodity may change as a conse-
quence of changes in consumption of the commodity andror changes in CO
2
intensity, i.e. the direct and indirect CO emissions from consuming and producing
2
the commodity. Consumption and CO intensity are shown for the commodity
2
groups in Table 5. In 1966, food, beverages and tobacco, and clothing were thus the most CO
2
Ž .
intensive commodity groups Table 5 . In 1992, transport took the lead together
4
Transport covers consumption of vehicles, public transport, taxis, and freight purchased by house- holds. Note that consumption of gasoline is not included } it is encompassed by direct household
energy consumption.
5
Entertainment covers manufacturing of recreational goods, entertainment, cultural services, media consumption, etc.
6
Further decomposition shows that this was carried out throughout the economy, but principally in the manufacture of glass and plastic products, distilling plus extraction of raw materials. However, a few
sectors were moving in the wrong direction: manufacture of food products, railway and bus transport, water works and supply, and fishing and fish farming, all of which exhibited increasing energy intensity.
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 434
Table 4 Ž
. CO emissions in 1966 and 1992 by commodity groups million tonnes CO
2 2
Commodity group 1966
1992 Change in emissions
Foods 5.5
5.5 1
Beverages and tobacco 0.8
1.0 25
Clothing 2.3
1.6 y
30 Household appliances
3.2 2.8
y 10
incl. operation Health
0.6 0.7
18 Recreation and enter-
2.4 3.9
59 tainment
Services 0.3
1.1 234
a
Transport 2.0
3.0 54
a
Includes vehicles and public transport services.
with foods and beverages and tobacco. Comparison of changes in CO intensity
2
and consumption reveals that the sizeable growth in emissions from services, recreation and entertainment, and transport is not due to changes in emission
intensity since the emission intensity of transport has only increased by 5, while that for services and recreation and entertainment has decreased by no less than
8 and 25, respectively. Emission growth is due to increasing consumption of all of these commodity groups, and is partly offset by reduced CO intensity.
2
The highest growth rates are observed in the consumption of services, transport, and recreation and entertainment. This indicates changing consumption patterns
towards a life style with more telecommunications, more travelling and more leisure activities. From an environmental point of view this development is mainly
Table 5 Consumption and CO intensity for different commodity groups
2
Ž Commodity group
Consumption thousand CO intensity
2
. Ž
. million 1980-DKK
kgrDKK 1966
1992 Change
1966 1992
Change Foods
26.6 34.4
28 0.20
0.16 y
22 Beverages and
4.2 6.6
46 0.20
0.16 y
21 tobacco
Clothing 13.4
13.7 2
0.17 0.12
y 31
Household appl- 33.9
55.5 54
0.09 0.05
y 45
Ž iances incl. opera-
. tion
Health 4.9
7.4 50
0.12 0.10
y 22
Recreation and 15.5
32.8 94
0.16 0.12
y 25
entertainment Services
4.6 16.5
196 0.07
0.07 y
8
a
Transport 11.7
17.1 39
0.17 0.18
5
a
Includes vehicles and public transport services.
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 435
beneficial since services and recreation and entertainment are characterised by low CO intensity. However, the increasing demand for transport services may consti-
2
tute a severe problem in the future and considerable efforts should therefore be made to control CO emissions from this activity.
2
As shown in Table 4, emissions from the consumption of foods accounted for 5]6 million tonnes CO per year during the period 1966]1992. Consumption of
2
foods is responsible for more than 10 of total Danish CO emissions from the
2
household sector. Comparison of indirect and direct household emissions reveals Ž
that only emissions associated with electricity consumption 8 million tonnes CO
2
. per year are greater than those associated with the consumption of foods. In
comparison, gasoline consumption only accounts for 5 million tonnes CO . One
2
should, therefore, bear in mind that CO emissions can be influenced in many
2
ways. Controlling private transport is just one possibility, while creating incentives for energy savings in food production is another. Trying to influence life style and
consumption habits is a third approach. In fact, policies directed towards household consumption of commodities other
than energy offer considerable potential for reducing CO emissions. As the
2
differences in the CO intensity of the various commodities is sizeable, changes in
2
commodity mix towards less CO intensive goods could be of significant impor-
2
tance. Table 6 illustrates the large variation in CO intensity, listing the five
2
commodities with the highest and lowest CO intensity in 1966 and 1992.
2
In 1992, the most CO intensive commodity was transport which is very energy
2
intensive. Second came public transport, followed by various food products. The five commodities with the lowest CO intensity are various types of services. This
2
implies that greater demand for services together with reduced consumption of commodities such as transport, foods and beverages will be accompanied by major
decreases in CO emissions. The reduction potential of altering the commodity mix
2
suggests policies should be directed towards this end. One possibility is labelling, another is a green levy programme in which the highest levies are imposed on the
most polluting commodities. Such policies would encourage the consumers to
Table 6 Ž
. Commodities with the highest and lowest CO intensity kgrDKK
2
Highestr 1966
1992 lowest
Top 5 Fruit and vegetables
0.38 Transport
0.30 Sport and camping equipment
0.33 Margarine etc.
0.22 Sugar
0.30 Other foods
0.22 Wine and liquor
0.30 Fruit and vegetables
0.21 Margarine etc.
0.28 Sugar
0.20 Bottom 5
Life insurance etc. 0.05
Education 0.06
Health and accident insurances 0.05
Medical care 0.05
Housing 0.04
Housing 0.03
Private organisations 0.03
Private organisations 0.03
Domestic servants 0.00
Domestic servants 0.02
J. Munksgaard, K.A. Pedersen r Energy Economics 22 2000 423]440 436
spend more of their budget on services and other commodities with low CO
2
intensity. However, the impact of a green levy programme will depend on how price elastic the demand is.
5. Concluding remarks
