Food Policy 55 (2015) 41–53

Contents lists available at ScienceDirect

Food Policy
journal homepage: www.elsevier.com/locate/foodpol

Effects of an environmental tax on meat and dairy consumption in

Sweden
Sarah Säll ⇑,1, Ing-Marie Gren
Department of Economics, Swedish University of Agricultural Sciences, Box 7013, 750 07 Uppsala, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: This study evaluated the environmental impacts of introducing an environmental tax on meat and dairy
Received 12 May 2014 consumption in Sweden. Three meat products (beef, pork and chicken), four dairy products (milk, fer-
Received in revised form 22 May 2015 mented products, cream and cheese) and four pollutants generating environmental damage (greenhouse
Accepted 23 May 2015
gases (GHG), nitrogen, ammonia and phosphorus) were included in the analysis. The unit tax applied cor-
responded to between 8.9% and 33.3% of the respective price per kg product in 2009. Consumer response
to the tax was calculated by econometric estimates of the almost ideal demand system (AIDS) for meat
Keywords:
and dairy products, using per capita consumption data and prices. The results indicated relatively inelas-
Environmental meat and dairy tax
Environmental impacts
tic own price elasticities and high income elasticities for all meat products and slightly lower for dairy
Sweden products. Simultaneous introduction of a tax on all seven products decreased emissions of GHG, nitrogen,
Meat and dairy demand elasticities ammonia and phosphorus from the livestock sector by up to 12%.
Ó 2015 Elsevier Ltd. All rights reserved.

Introduction (2014) concluded that it will be difficult to reach worldwide cli-

Global meat production increased by 245% and dairy produc-
tion increased by 70% between 1961 and 2001, and is likely to con-
tinue to increase in the near future, driven by economic
development in many countries (Steinfeld and Gerber, 2010).
Several studies have reported detrimental environmental effects
of meat and food production, such as climate change from green-
house gas (GHG) emissions and impaired water quality from nutri-
ent leaching (e.g. FAO, 2006; Galloway et al., 2008; Gerber et al.,
2013). FAO (2006) found livestock to be responsible for 18% of total
GHG (or carbon equivalents, CO2/e) emissions worldwide, although
this figure was revised downwards to 14.5% by Gerber et al. (2013).
However, McMichael et al. (2007) concluded that 22% of total
anthropogenic GHG emissions come from agriculture, with 80%
of those from livestock. In addition to GHG emissions, Galloway
et al. (2008) point out the important role of the agricultural sector
in the increasing threat of reactive nitrogen on global scale. The
need to reduce consumption of meat and dairy products for envi-
ronmental reasons has been emphasised by several researchers
(e.g. UNEP, 2009; Röös and Tjärnemo, 2011; Cederberg et al.,
2013; Bajželj et al., 2014; Hedenus et al., 2014.). Hedenus et al.
⇑ Corresponding author. Tel.: +46 70 942 35 66.
E-mail address: sarah.sall@slu.se (S. Säll).
1 2
With special thanks to Professor Yves Surry, Department of Economics, Swedish
University of Agricultural Sciences, for help with the model and valuable comments.
mate targets without dietary changes.
The negative environmental externalities of meat and dairy pro-
duction could, in principle, be mitigated by means of a Pigovian tax
(Pigou, 1957). This would involve an increase in the consumer
prices corresponding to the marginal damage costs, so that meat
and dairy commodities carry their associated social costs. Such a
tax would provide incentives to reduce environmentally damaging
practices in animal food production, and at the same time promot-
ing the dietary changes needed.
The aim of this study was to evaluate the effects of a Pigovian
tax on meat and dairy consumption in Sweden. Total consumption
of beef, pork and chicken in Sweden increased from 460 000 to
725 000 tons2 between 1990 and 2009, which is an increase of
58% (Swedish Board of Agriculture, statistical database). In addition
to GHG, this study included nitrogen, ammonia and phosphorus
emissions, which have been causing eutrophication in inland and
coastal waters, particularly in the Baltic Sea, since the early 1970s,
resulting in algae blooms and oxygen depletion. According to our
calculations, Swedish meat and dairy production accounts for
approximately 18.5% of total nitrogen, including nitrogen from
ammonia emissions, and 8.3% of total phosphorus released into the
Baltic Sea from Sweden. Given the large proportions of agricultural
and arable land that are needed for livestock production (worldwide
Taking population increase into account, which increased from 8 590 630 in 1990
to 9 340 682 in 2009 (Statistics Sweden, 2010).

http://dx.doi.org/10.1016/j.foodpol.2015.05.008
0306-9192/Ó 2015 Elsevier Ltd. All rights reserved.
42 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53
33% of arable land and 70% of agricultural land FAO (2006)) the real
percentage is likely to be much higher.
The environmental impacts of introducing a Pigovian tax on
beef, pork and chicken, as well as milk, fermented dairy products,
cream and cheese, were determined as consumer response to price
changes, which was assessed by econometric analyses using the
Almost Ideal Demand System (AIDS) (Deaton and Muellbauer,
1980). The analysis was conducted in two stages, with total food
demand as the first upper stage, and meat demand as one of the
second, lower stages, and dairy demand as the other lower stage
(Edgerton, 1997; Carpentier and Guyomard, 2001). Based on esti-
mated price and income elasticities, the effects of an environmen-
tal meat and dairy tax on emissions of GHG, nitrogen, ammonia
and phosphorus was determined.
Starting in the 1920s, there is a large body of literature on anal-
yses of economic instruments for combating environmental dam-
age, applied e.g. to water pollution, climate change and
biodiversity conservation (see e.g. Helfand et al., 2003 for a
review). Based on this literature, one could argue in favour of dif-
ferentiated taxes according to pollutant emissions and environ-
mental impacts by each polluter (e.g. Baumol and Oates, 1988).
However, when considering difficulties in monitoring each firm’s
pollutant emissions and associated environmental damage, uni-
form taxes on outputs can be less costly to society (e.g.
Schmutzler, 1997; Gren, 2004; Kampas and White, 2004). A deci-
sion on whether to implement a tax on the consumption or the
production side also had to be made. An argument in favour of con-
sumption taxes in a small country like Sweden is the issue of com-
petition from imported products, which would be severe if a tax
was to be imposed on the production side. Swedish production
would decrease, consumption would probably remain constant,
and emissions would be exported, as net imports of meat and dairy
products increased. The net effect of emission mitigation might
then even be zero if policies were implemented on the production
side (e.g. Van Doorslaer et al., 2015).
The design of efficient food taxes on consumption has been
investigated in numerous studies, in which the tax is applied
mainly to effect dietary changes for improvement of health (e.g.
Schroeter et al., 2008; Nnoaham et al., 2009; Nordström and
Thunström, 2011; Briggs et al., 2013). However, very few studies
have examined the design and effects of environmental taxes on
meat and dairy consumption (Wirsenius et al., 2011; Edjabou
and Smed, 2013). Wirsenius et al. (2011) considered consumption
taxes in the EU on animal products and Edjabou and Smed (2013)
designed taxes for a large set of food items in Denmark, in order to
regulate GHG emissions and promote an environmental friendly
diet. Both studies made use of estimated price elasticity and
income elasticity of the included food products for simulating
effects on GHG emission of different tax levels, an approach
adopted in the present study. The main novel contribution of our
study is the inclusion of nitrogen, ammonia and phosphorus emis-
sions, in addition to GHG, when determining tax levels for meat
and dairy, and estimating the associated impacts on pollutant
emissions. The econometric approach used followed the literature
by constructing an AIDS model for estimating demand and income
elasticity of the meat products (Deaton and Muellbauer, 1980).
There were some limitations to our analysis. While we recog-
nise the need to consider the entire food chain and a larger set of
food items, when regulating pollutant emissions, we chose to study
animal food products. Meat was a main focus, which was moti-
vated by findings by Wirsenius et al. (2011) indicating that a tax
on ruminant meat contributed most of the mitigation effect of a
GHG-weighted tax on animal food. Edjabou and Smed (2013) con-
firmed that the largest decrease in GHG emissions is obtained from
a tax on beef. However, excluding dairy products was not possible
given the high emission levels from, for example, cheese. Our aim
was to include nitrogen, ammonia and phosphorus and in our esti-
mations, and thus extending the analysis reported by Wirsenius
et al. (2011) and Edjabou and Smed (2013). Another limitation of
the study is the calculation of average damage costs of pollutants
instead of the theoretically correct marginal damage cost. The rea-
son is the lack of data.
This paper is organised as follows. In Section ‘Model specifica-
tion of meat demand, meat tax and emission reductions’, we pre-
sent the approach used for calculating effects of environmental
taxes and the AIDS model used to estimate demand and income
elasticity. In Section ‘Estimation of meat demand’, data and regres-
sion results on the demand system are presented, while Section ‘C
alculation of environmental meat tax’ presents calculations of
average emission levels and average damage costs per kg meat
and dairy products. The results are presented in Section ‘Impact
of meat taxes on pollutant emissions’, followed by a discussion
and summary in the sixth section.
Model specification of meat demand, meat tax and emission
reductions
The modelling framework and the subsequent empirical analy-
ses consisted of three main steps: (i) derivation of demand for ani-
mal products before the tax, (ii) calculation and introduction of the
tax and derivation of the new demand for meat and dairy products,
and (iii) estimation of emissions of environmental pollutants
before and after the introduction of the tax. Starting with the first
step, we used the non-linear AIDS model to calculate demand and
income elasticities on per capita level (Deaton and Muellbauer,
1980). This was conducted in two stages (e.g. Edgerton, 1997). In
the first stage, a demand system of aggregated food groups was
estimated. In the second stage, we estimated two separate systems,
one for meat products and one for dairy products. Final uncompen-
sated elasticities take both changes within the food group, and
changes to other groups of food into account.
In the model framework, share of total consumption of each
meat product in each period of time, sj,t (where j = 1, . . . , n prod-
ucts) was expressed as a function of a constant, aj and logged
prices of all commodities and total expenditure Xt. The AIDS model
was thus:
X
m
sj;t ¼ aj;t þ cj;k ln pk;t þ bj ðln X t  ln Pt Þ ð1Þ
k¼1
where k = 1, . . . , m products within the group of
P
commodities,sj,t = pjqj/Xt, X t ¼ mj¼1 pj qj is total expenditure for each
year for each commodity group, and the price index ln Pt was:
X
n
1Xn X
m
ln Pt ¼ a0 þ aj;t ln pj;t þ c ln pj;t ln pk;t ð2Þ
j¼1
2 j k j;k
The parameters are to fulfil adding-up restrictions as well as
Pn Pn
homogeneity and symmetry conditions, j aj ¼ 1 and, j bj ¼ 0
where bj shows the change in budget shares when expenditure
changes. Furthermore, cj,k indicates the change in budget share
when the price changes and the homogeneity restriction requires
P
that m j¼1 cj;k ¼ 0. Symmetry conditions imply that a change in price
of good j has the same marginal effect on the budget share of good
k as a price change of good k has on the marginal change of budget
shares of good j, i.e. that cj,k = ck,j.
Price and income elasticity calculations for the two-stage
demand system were performed using definitions in Green and
Alston (1990) and the multistage elasticities from Edgerton
(1997). In the following, the subscript M denotes uncompensated,
Marshallian elasticity, H compensated Hicksian elasticity and I
income elasticity for each state separately, which are defined as:
 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53
eIj ¼ 1 þ bj =sj
eMj;k ¼ ½ðcj;k  bj sk Þ=sj   dj;k
eHj;k ¼ ðcj;k =sj Þ þ sk  dj;k
where the Kronecker delta d ¼ 1 if j ¼ k and 0 otherwise:
Homogeneity of degree zero requires that restrictions on elasticities
P Pm H
are: (i) eIj þ m M
k¼1 ej;k ¼ 0 and (ii) j¼1 ej;k ¼ 0.
Eqs. (1)–(5) were used for estimating conditional elasticities in
both stages of the demand system, i.e. total demand for food in
Sweden (with aggregated food groups) and demand for only meat
products. Final elasticity values combined both stages.
Uncompensated income elasticity was defined as:
eIj ¼ eIj eIr
and the final uncompensated price elasticities were:
 
eM M I M 0
j;k ¼ dr;u ej;k þ ej sk dr;u þ er;u
eH M I
j;k ¼ ej;k þ ej sk
where dr,s = 1, indices j and k denote within the meat group and
indices r and u denote the aggregated groups (Edgerton, 1997).
Taxes for each commodity j were calculated as the average
damage cost (ADi) from pollutant i = 1, . . . , h, times the average
emissions, ei,j, per kg meat. The environmental tax on animal pro-
P
duct j was then written as taxj ¼ hi ei;j ADi . Change in demand from
implementation of the tax was calculated as:
DQ j Xm
DPk M
¼ e
Qj k
Pk j;k
where DQj/Qj was the percentage change in demand of each meat
after tax and DPk/Pk the percentage increase in price of each com-
modity with tax.
Finally, emissions of a pollutant i, Ei, were calculated as the sum
of emissions over all meat and dairy products. For each commod-
ity, emissions amount was obtained by multiplying consumption
by emissions per unit consumption. The effect of the tax was then
estimated as the difference in emissions with and without the tax,
which was written as:
X
m
DE i ¼ ei;j DQ j
j
The application of Eqs. (3)–(5) are presented for both stages in
the demand system in the following section, as well as (30 )–(50 )
and calculations according to Eqs. (6) and (7) are presented in
the ‘Results’ section.
Estimation of meat demand
As noted in the previous section, the first step in our calcula-
tions of the environmental impacts of an environmental tax on
meat consisted of econometric estimates of demand for different
meat products in Sweden. This was done in a two-stage frame-
work, where the upper stage consisted of the aggregated food
groups: meat, dairy, vegetables, fruit and grain. Fish had to be
excluded due to lack of data on fresh fish consumption after
1995. The second, lower stages was (i) carcass weight consumption
of beef, pork and chicken and (ii) milk, fermented dairy products,
cream and cheese. Carcass weight on meat products was chosen
43
ð3Þ because data on direct consumption of meat products would have
included processed meats, which would have created difficulties in
ð4Þ relating appropriate environmental damage costs to mixed meats.
All data used were taken from the Swedish Board of
Agriculture’s statistical database, showing per capita consumption
ð5Þ
1980 to 2012. The selection of 1980 as the starting date was based
on Rickertsen (1995) for Norwegian meat consumption and Lööv
and Widell (2009) for Swedish food consumption, both of which
found a structural change in meat consumption around 1980.
Beef, pork and chicken accounted for 80.3 kg/person of total meat
consumption of 85.7 kg/person in 2012. Less frequently consumed
products such as mutton and lamb, horse meat and wild game
were excluded. The same data source was used for the demand
system of dairy products.
Between 1980 and 2012, per capita consumption of beef, pork
0
ð3 Þ and chicken in Sweden increased by almost 36%. The main increase
was in chicken consumption, which more than tripled, followed by
beef, which increased by almost 37% (the relatively high increase
in chicken consumption follows a global pattern (e.g. Fabiosa,
2011). Consumption of pork remained rather constant. During
ð4 Þ
the study period, prices also increased and the real price index in
2012 relative to 1980 = 1 was 2.8 for beef, 2.7 for pork and 1.5
0
ð5 Þ for chicken. The price of beef and pork was highest in 2012, but
increased only slightly from 1990 (when the index was 2.4 and
2.6 for beef and pork, respectively), while the peak in chicken
prices occurred in the 1990s. This decrease in relative price in
recent years could be one explanation for the large increase in
chicken consumption (see Fig. 1). All prices are from the Swedish
Board of Agriculture’s statistical database, showing consumer price
indexes on agricultural products (1980 = 1).
The increasing trend seen in meat was also found in dairy prod-
ucts, with the exception of drinking milk, where consumption
decreased during the whole time period. In 1980, the average
ð6Þ
Swede used 162 l of milk per year. In the end of the time period,
this number had decreased to less than 90 l. In Figs. 2 and 3, the
consumption of milk, fermented products, cream and cheese is
presented. Consumption of fermented products and cream
increased with close to 50% per person from 1980 to 2012, and
cheese consumption with 34% during the same time period.
During the period, prices of milk and cheese have followed each
other quite closely, price index of both commodities was 3.6 in
the end of the time period, and the average index in the study per-
iod was 2.5. Cream had the smallest price increase and in 2012
index was on 1.7. The highest price on cream was in 1990.
Consumption of dairy products is presented in Figs. 2 and 3.
Descriptive statistics of the data used in the demand systems is
ð7Þ
presented in Table 1.
The demand systems presented in Section ‘Model specification
of meat demand, meat tax and emission reductions’ were esti-
mated using SURE (Seemingly Unrelated Regression Equations).
The model was tested with lagged variables, including an autocor-
relation parameter. LM tests indicated no problem with autocorre-
lation. The models showed good statistical fit as measured by R2,
which varied between 0.78 and 0.99 depending on model. Test
results are presented in Appendix, Table A1.
However, expenditures on the meat and dairy products can be
endogenous and dependent on total income and other macroeco-
nomic variables. The SURE estimate can then create biased and
inconsistent estimates. A difficulty when accounting for such endo-
geneity is to find valid and sufficiently strong instruments, which
should be correlated with expenditures on the respective food
products, but not directly with the dependent variable (e.g.
Angrist and Krueger, 2001). Lack of such instruments may create
more problems with biased and consistent estimates than when
the endogenous explanatory variable is treated as exogenous (e.g.
Murray, 2006). We therefore follow the literature and treat
44 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53

40
Kilo
35

30

25
Beef consumpon
20 Pork consumpon
15 Chicken consumpon

10

0
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
Fig. 1. Per capita consumption (kg carcass weight) of beef, pork, chicken and all meat in Sweden 1980–2012. Per capita consumption, kilo’ on y axis. Source: Swedish Board of
Agriculture, statistical database.

Kilo 180
Table 1
160
Descriptive statistics of meat and dairy product in Sweden 1980–2012. Q refers to
140 quantity and P to Price index. Source: Swedish Board of Agriculture, statistical
database.
120
100 Min Max Mean Std dev

80 Beef Q 15.8 26.2 20.6 3.8

Milk
Beef P 1 277.9 203.2 41.0
60
Pork Q 29.4 37.3 33.9 2.5
40
Pork P 1 271.0 219.1 39.2
20
Chicken Q 4.5 18.9 10.5 5.1
0 Chicken P 1 190.0 142.5 22.4
1980
1985
1990
1995
2000
2005
2010

Milk Q 89.4 162.1 122.7 21.3

Milk P 1 3.6 2.5 0.7
Fig. 2. Per capita consumption of milk in Sweden 1980–2012. Per capita Fermented Products Q 22.8 34.8 29.1 3.4
consumption, kilo’ on y axis. Source: Swedish Board of Agriculture, statistical Fermented products P 1 3.2 2.4 0.6
database.
Cream Q 7.7 11.8 9.7 1.2
Cream P 1 2.0 1.7 0.3
Cheese Q 14.1 19.5 16.7 1.4
Kilo 40 Cheese P 1 3.6 2.6 0.7
35
30
Fermented All estimated own price elasticities were negative, which is in
25
products accordance with theory, and all but two of the compensated are
20
Cream significant. The level of these elasticities was of the same order
15 of magnitude for all three meat products, though lower in absolute
10 Cheese terms than the values obtained by Lööw and Widell (2009). We
used data from the same source, only adding some observations
5
for later years. However, while we used carcass weight consump-
0 tion on meat, Lööw and Widell (2009) use direct consumption add-
1980
1986
1992
1998
2004
2010

ing variables to the demand system. Also, we used a non-linear

model instead of the linear version of the AIDS model.
Fig. 3. Per capita consumption of fermented dairy products, cream and cheese in The sign of most cross-price elasticity values for meat products
Sweden 1980–2012. Per capita consumption, kilo’ on y axis. Source: Swedish Board was negative, showing that commodities are complementary prod-
of Agriculture, statistical database.
ucts. This is not an uncommon finding, although it might be ques-
tioned whether meats and dairy products are complements or
expenditures as exogenous (e.g. Rickertsen, 1995; Lööv and Widell, substitutes within the groups. Rickertsen (1995) estimated elastic-
2009; Nordström and Thunström, 2009). ities for Norway and Scotland and reported similar results.
Regression results on group elasticity are presented in Table 2 Regarding dairy, the commodities was less sensitive than meat,
and were used to calculate final uncompensated elasticities in both in prices and in income changes. Also, more substitutes are
Table 3. found within the dairy system.
Result shows that as a group, meat consumption was more sen-
sitive to price and income changes, than what dairy consumption Calculation of environmental meat tax
was found to be. In Table 3, both compensated, within group elas-
ticities (Eqs. (3)–(5) in Section ‘Model specification of meat The calculation of the environmental meat tax constituted the
demand, meat tax and emission reductions’) and final uncompen- second step in our calculations. The tax level for each meat product
sated elasticities (Eqs. (30 )–(50 )) are presented. Parameter values for was derived from the environmental damage caused by GHG,
all demand systems are presented in Appendix, Table A2. nitrogen, ammonia and phosphorus at the production stage of that
 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53 45

Table 2
Regression results on compensated group elasticity for prices on meat and dairy, and
income in Sweden, 1980–2012. All values are based on medium values.
Meat/dairy Income
Meat 0.614*** 0.783***
(0.158) (0.216)
Dairy 0.275*** 0.522***
(0.077) (0.125)
Standard deviation in brackets. Underlying budget equations are concave, with the
Slutsky matrix being negative-definite (negative eigen values).
***
Significance at 1% level.
meat and dairy product in Sweden. The choice of these pollutants
was based on data availability and the possibility of calculating
environmental damage in monetary terms.
Many previous studies have considered measurement of envi-
ronmental damage in monetary terms (see Turner et al., 2003 for
a review). A common approach in these previous studies is to
assess environmental problems of concern, such as degraded water
quality, climate change and biodiversity loss. However, the causes
of these problems, such as pollutant emissions and land use
change, are seldom identified and quantified. This creates difficul-
ties with respect to the calculation of environmental damage by
specific pollutants in monetary terms. In this study we used
revealed preference by Swedish politicians for assessing the dam-
age costs of GHG and nutrients. The revealed preference of damage
costs for GHG is in terms of a tax on carbon dioxide, which
amounts to 1000 SEK/ton CO2. Corresponding preferences for
reductions in nitrogen, ammonia and phosphorus are shown by
the willingness to pay (WTP) for the cost of entering and comply-
ing with international agreements on nutrient reductions. The
abatement costs at the margin of obtaining the nutrient reduction
targets for the Baltic Sea were then used as the revealed marginal
damage cost (Gren et al., 2008).
Data on emissions of GHG for the different meat products, mea-
sured in CO2/e, were obtained from Cederberg et al. (2009) who
estimated emissions using Life Cycle Analysis (LCA) (Table 4). All
values of GHG emissions are corrected upwards according to new
global warming potential weights of methane, given in the most
recent IPCC report (Myhre et al., 2013).
The damage caused by nitrogen and phosphorus emissions is
mainly on water quality, and data are available on the impacts
on the Baltic Sea of emissions from different locations in Sweden
(Elofsson, 2003). In this study we recalculated the average emis-
sions per kg meat for six different regions in Sweden to give total
national loads to average local regions as well as the Baltic Sea,
with the shares of emissions reaching the Baltic Sea giving the
weighting for the final damage cost calculations (Gren et al.,
2008). Nitrogen emissions were available in several studies, but
they show a large variation (e.g. Elofsson, 2000; Schou et al.,
2006; Leip et al., 2010; Pierer et al., 2014). As a reference we used
nitrogen emissions from Leip et al. (2010) which are calculated as
nitrogen in ammonia emissions per kilo of output. Approximately
40% of total nitrogen that is emitted from livestock is released into
the air as ammonia and the remaining 60% is deposited on land for
further transports into soil and water (Elofsson, 2000). Thus the
values in Leip et al. (2010) was assumed to be 40% of total emis-
sions of nitrogen, presented as average ammonia emissions per
kilo produce in Table 4. The remaining 60% was used to calculate
the land and water borne leaching (Nleakage in Table 4). Ammonia
was added as separate posts, where one is total emissions from
the sector, and the other is the amount of ammonia that is depos-
ited on the Baltic Sea. The deposition on the sea is assumed to be
16% of total emissions (Gren et al., 2008). Once in the sea, ammonia
was treated as all nitrogen, creating the same abatement cost per
kilo. Phosphorus was assumed to have the same ratio to nitrogen

Table 3
Regression results on the demand system, compensated elasticity and calculated final uncompensated elasticity for beef, pork, chicken and income in Sweden, 1980–2012.

Beef Pork Chicken Income

Compensated elasticities, within group Beef 0.661*** 0.368* 0.024 1005***
(0.150) (0.159) (0.101) (0.173)
Pork 0.197** 0.562*** 0.175** 0.933***
(0.069) (0.103) (0.057) (0.112)
Chicken 0.020 0.770*** 0.435** 1.225***
(0.230) (0.287) (0.235) (0.369)
Milk Fermented products Cream Cheese Income
Milk 0.709*** 0.112* 0.065*** 0.134*** 1.020***
(0.051) (0.070) (0.021) (0.031) (0.026)
Fermented products 0.427 0.574 0.112 0.146 0.973***
(0.283) (0.678) (0.170) (0.307) (0.0843)
Cream 0.743*** 0.347 0.176 0.311 0.955***
(0.261) (0.508) (0.227) (0.220) (0.112)
Cheese 0.900*** 0264 0.185 0.480* 0.930***
(0.222) (0.544) (0.130) (0.274) (0.077)

Beef Pork Chicken Income

Final uncompensated elasticities Beef 0.538 0.161 0.083 0.786
Pork 0.082 0.370 0.120 0.731
Chicken 0.130 0.518 0.363 0.959
Milk Fermented products Cream Cheese Income
Milk 0.205 0.011 0.023 0.064 0.532
Fermented products 0.054 0.457 0.072 0.213 0.508
Cream 0.271 0.231 0.137 0.376 0.499
Cheese 0.440 0.377 0.223 0.415 0.485

Standard deviation in brackets. Underlying budget equations are concave, with the Slutsky matrix being negative-definite (negative eigen values).
*
Significance at 10% level.
**
Significance at 5% level.
***
Significance at 1% level.
46 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53

Table 4
Emissions of carbon dioxide equivalents (CO2e), nitrogen (N), ammonia (NH3) and phosphorus (P), produced nationally (N/NH3/Pcountry) after leakages are discounted for (N/
Pleakage) and entering the Baltic Sea where also retention levels are accounted for (N/Pbaltic), per kg beef, pork and chicken produced in Sweden, 1980–2012.

CO2/e Nland Nleakage Nbaltic NH3country NH3baltic Pcountry Pleakage Pbaltic

Beef 24.29a 0.142 0.023 0.014 0.095b 0.015 0.03985c 0.00065 0.00027
Pork 3.74a 0.044 0.007 0.004 0.029b 0.005 0.02801c 0.00046 0.00019
Chicken 1.93a 0.028 0.005 0.003 0.019b 0.003 0.01103d 0.00017 0.00007
Milk 1.29a 0.009 0.001 0.001 0.006b 0.001 0.00215c 0.00003 0.00001
Fermented products 1.29f 0.009 0.001 0.001 0.006 0.001 0.00215 0.00003 0.00001
Cream 4.79e 0.032 0.005 0.003 0.022 0.003 0.00795 0.00013 0.00005
Cheese 9.58e 0.065 0.011 0.006 0.043 0.007 0.01590 0.00026 0.00011

‘‘Land’’ is the land bound nitrogen, 60% of total emissions, ‘‘Country’’ is the total amount of NH3 and P emitted, ‘‘Leakage’’ is average local emission levels of nitrogen and
phosphorus after leakages are accounted for. ‘‘Baltic’’ is the emissions that leach into the Baltic Sea, after retention levels are extracted.
a
Cederberg et al. (2009) where 85% of total emissions from dairy cows is allocated to milk.
b
Leip et al. (2010) Annex 6.
c
Schou et al. (2006).
d
Elofsson (2000).
e
Röös (2012).
f
Assumed to be the same as milk.

as found in Schou et al. (2006) and Elofsson (2000). In those papers,
emissions of nitrogen and phosphorus per animal and year are pre-
sented. The nitrogen levels presented in those papers are total
nitrogen, thus phosphorus levels relate to total levels of nitrogen,
including both land and air emissions (Nland and NH3country in
Table 4). For fermented products, cream and cheese, nitrogen,
ammonia and phosphorus were given the same weight as GHG
gases, compared to milk. Cheese emits almost eight times as much
GHG gases per kilo as milk. The nitrogen and phosphorus emis-
sions of cheese are thus eight times higher than of milk.
In Table 4, average emissions are presented. For nitrogen,
ammonia and phosphorus, both local emissions and emissions that
reach the Baltic Sea are shown. Local levels are the emissions at the
source that leak out in the local areas, while the emissions that
reach the sea also depend on nutrient retention levels in each
region. Losses and retention levels and local production shares
are shown in Appendix Table A3.
Beef was the main emitter of all pollutants. Pork had the sec-
ond highest phosphorus emissions and cheese generated the sec-
ond highest average emissions of GHG and nitrogen. Pierer et al.
(2014) calculate nitrogen footprints for food products in Austria.
It is not clear what the values of nitrogen in that paper contain,
however, they are lower than the ones given in manure in Leip
et al. (2010). Thus it was reasonable to assume that not all nitro-
gen emissions were included. The ratio of phosphorus and nitro-
gen was the same in those calculations, hence the total emissions
of both nitrogen and phosphorus were lower than in Leip et al.
(2010). Still we used the estimations from that paper in the sen-
sitivity analysis in Section ‘Impact of meat taxes on pollutant
emissions’, excluding the more detailed analysis including
ammonia.
With respect to the damage costs of GHG, Stern (2006) refers to
studies reporting GHG emissions costs to range between $0 and
$400 per ton in year 2000 prices, with an average of $85 per ton.
This corresponds to between 0 and 2.8 SEK per kg CO2e.
Wirsenius et al. (2011) use a cost of €60 per ton and Edjabou and
Smed (2013) use several different costs which range between 1.2
and 6.9 DKK per kilo (the highest level corresponded to more than
SEK 8). With such uncertainties, the politically revealed cost of
GHG emissions in Sweden was used instead and the cost of carbon
equivalents was set to SEK 1 per kg which was slightly higher than
the Stern average (which corresponded to 0.55 SEK per kilo). Data
on the marginal damage costs of nutrients were taken from Gren
et al. (2008), who found that marginal abatement of one kg of
nitrogen in the Baltic Sea costs SEK 252 at a 50% reduction level
and phosphorus SEK 3279 per kg at a 60% reduction level. Again,
since no damage functions are available, these cost estimates were
used here as a proxy for the damage costs of nutrients. Average
damage costs (ADi) and final tax levels are presented in Table 5.
According to the results presented in Table 5, the inclusion of
environmental damage costs on nutrients raised the tax by the
most with approximately 50% for pork and chicken. The tax on
dairy products was increased by approximately 30%. With beef,
the inclusion of nutrients had the lowest effect, around 25%
increase of the costs compared to only including GHG emissions
in the calculations. In total, one kg of beef had the highest damage
costs and milk and fermented dairy products the lowest. Nguyen
et al. (2011) tried to estimate the environmental costs of pig pro-
duction and found that €1.9, corresponding to close to SEK 17
per kg of pork, would cover the costs of negative external effects.
This is higher than our estimates. However, their model also
included a human health quality measure (Weidema, 2009). As

Table 5
Average damage costs (ADi) in SEK/kg of carbon dioxide equivalents (CO2e), nitrogen (N), ammonia (NH3) and phosphorus (P) final tax level, initial price per kg and price change
(%) for meat and dairy products in Sweden, 1980–2012.

CO2/e Nbaltic NH3baltic Pbaltic Total, taxj Initial price DP/P

Beef 24.29 3.52 3.81 0.87 32.50 97.7 33.3
Pork 3.74 1.13 1.19 0.62 6.67 59.3 11.3
Chicken 1.93 0.73 0.75 0.22 3.63 41 8.9
Milk 1.29 0.22 0.23 0.05 1.79 8 22.4
Fermented products 1.29 0.22 0.23 0.05 1.79 10.3 17.4
Cream 4.79 0.80 0.87 0.17 6.63 35 18.9
Cheese 9.58 1.60 1.74 0.35 13.26 76.4 17.4

Initial price levels on meat products are from prices (€/kg) in Wirsenius et al. (2011), corrected for exchange rate and price index in Sweden. Dairy product prices are from
Statistics Sweden (2013).
 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53 47

mentioned above, it is likely that our values of nitrogen and phos-
phorus are too low.
Impact of meat taxes on pollutant emissions
As stated in Section ‘Model specification of meat demand, meat
tax and emission reductions’, the impact of the meat tax on emis-
sions was calculated by taking account of the difference in emis-
sions before and after introduction of the tax. It was assumed
that the emissions of pollutants per unit meat consumption were
the same for domestic production and imports, but not the effects
of nutrients on the Baltic Sea. The fraction of imports in total con-
sumption was 0.39, 0.22 and 0.29 for beef, pork and chicken,
respectively. For dairy products the fraction of imports of milk
and cream was only 0.03, fermented products 0.18 and cheese
0.39 (Lukkarinen and Lannhard Öberg, 2012). We therefore present
emissions from domestic production and imports before the intro-
duction of the tax, see Table 6 (imports are then assumed to have
similar emission levels, which could be a very strong assumption
depending on the country of origin). However, most imported meat
in Sweden originates from the EU, (for an overview of net imports
of meat and dairy products in Sweden see Lukkarinen and Eklöf,
2011; Lukkarinen et al., 2011; Lukkarinen and Lannhard Öberg,
2011, 2012) where emissions from livestock are on average similar
to emissions from Swedish livestock (Leip et al., 2010).
Consumption of animal products in Sweden accounted for
approximately 12.7% of total emissions of GHG from production
in Sweden, and for lower shares of nutrients and ammonia.
Keeping in mind that all emission levels per kilo has been corrected
upwards with the new measure for methane, the total share of
GHG emissions in relation to the total Swedish emissions levels
is too high, since the national inventories has not been updated.
As can also be seen from Table 6, at least two-thirds of the
emissions from consumption were created in Sweden. Beef stands
out as the main polluter, even on aggregate level. Regarding GHG
emissions from Swedish production, milk come second due to
the large quantities consumed, followed by pork meat and cheese.
For nutrients, pork is second to beef with both nitrogen and phos-
phorus emissions.
The effect on estimated demand for the meat and dairy prod-
ucts of introducing an environmental tax varied by approximately
1.8–13.1% depending on product and inclusion of price elasticities,
see Table 7.
For cheese and fermented products, the substitution effects
reduced the effect when considering all price effects. While for
all other products the opposite occurred because of complementar-
ity in demands. The reduction of beef and pork would take Sweden
back to the levels of consumption recorded at the end of the 1990s.
This was in contrast to chicken consumption, where the impact
was small. For milk, this reduction would be in line with decrease
that has been observed the whole time period. Consumption of fer-
mented products would be at the same level as 10–15 years ago,
while cream and cheese consumption would be back to the
observed levels in the beginning of the time period.
Emissions impacts in the reference case
Calculated impacts of the taxes in Table 5, on emissions are pre-
sented in Table 6. It was assumed that demand for domestic and
imported products would be reduced in proportion to the alloca-
tion in 2009. Calculated emission reductions (Eq. (7) in
Section ‘Model specification of meat demand, meat tax and emis-
sion reductions’) are shown in Fig. 4.
Within the livestock producing sector, emission reductions can
reach 12%, with the largest decrease in GHG emission, followed by
local nitrogen and nitrogen reaching the Baltic Sea. The smallest

Table 6
Aggregate levels (kton) of carbon dioxide equivalents (CO2e), nitrogen (N), phosphorus (P) emissions from Swedish animal products, production and imports in 2009.

CO2/e Nitrogen Ammonia Phosphorus

Prod Import Prod/leakage Import Baltic Prod Import Baltic Prod/leakage Import Baltic
Beef 3449 1352 3.32 1.30 1.99 13.43 5.26 2.15 0.092 0.036 0.038
Pork 979 217 1.90 0.42 1.17 7.69 1.71 1.23 0.121 0.027 0.049
Chicken 218 66 0.56 0.17 0.33 2.10 0.64 0.34 0.019 0.006 0.008
Milk 1124 33 1.25 0.04 0.75 5.05 0.15 0.81 0.030 0.001 0.012
Fermented products 333 61 0.37 0.07 0.22 1.50 0.28 0.24 0.009 0.002 0.004
Cream 486 14 0.54 0.02 0.32 2.19 0.06 0.35 0.013 0.000 0.005
Cheese 1005 388 1.12 0.43 0.67 4.51 1.74 0.72 0.027 0.010 0.011
Total 7594 2132 9.06 2.44 5.44 36.47 9.84 5.83 0.310 0.082 0.127
Sweden, total 60000a 85.9 60.9d 50.1c 60.9d 2.08b 1.54b
% of total in Sweden 12.7 10.5 8.9 72.8 9.6 14.8 8.3
a
Swedish Environmental Protection Agency (2012).
b
Brandt et al. (2008).
c
Staaf and Bergström (2011).
d
Total Nitrogen loads to the Baltic Sea, from both Ammonia and Nitrogen from land (Brandt et al., 2008).

Table 7
Reduction in demand (% and kilo/capita/year) for meat and dairy products in Sweden after the introduction of an environmental tax. All price effects and own price effects.

All price effects Own price effects

% Kilo per capita % Kilo per capita
Beef 19.0 4.7 17.9 4.5
Pork 8.0 2.9 4.2 1.5
Chicken 4.7 0.8 3.2 0.6
Milk 5.9 5.7 4.6 4.4
Fermented products 4.4 1.5 7.9 2.7
Cream 6.1 0.7 2.6 0.3
Cheese 6.3 1.2 7.2 1.3
48 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53

14.00
%
12.00 Reducons from
the livestock sector
10.00
Reducons in
relaon to total
8.00
emissions from
Swedish
6.00
producon
4.00

2.00

0.00
Co2/e N leakage N balc Nh3 land Nh3 balc P leakage P balc

Fig. 4. Reductions in emissions of carbon dioxide equivalents (CO2e), nitrogen (N) and phosphorus (P) produced nationally (N/Pcountry) and entering the Baltic Sea (N/Pbaltic)
from introduction of an environmental tax on meats and dairy products (% from initial emissions in Table 5 as well as the Swedish total emission from all domestic production
which are also shown in Table 5). All price effects are included.

Table 8
Allocation of reductions in emissions of carbon dioxide equivalents (CO2e), nitrogen (N) and phosphorus (P) produced nationally (N/Pcountry), and entering the Baltic Sea (N/
Pbaltic) among meat and dairy products, all elasticities included.

CO2/e Nleakage Nbaltic NH3country NH3baltic Pleakage Pbaltic

Beef 0.713 0.629 0.626 0.630 0.630 0.533 0.535
Pork 0.085 0.151 0.155 0.152 0.152 0.295 0.294
Chicken 0.011 0.026 0.026 0.025 0.025 0.027 0.027
Milk 0.073 0.074 0.074 0.074 0.074 0.055 0.055
Fermented products 0.016 0.016 0.016 0.016 0.016 0.012 0.012
Cream 0.033 0.033 0.033 0.033 0.033 0.025 0.025
Cheese 0.069 0.070 0.070 0.070 0.070 0.052 0.052
Total 1.0 1.0 1.0 1.0 1.0 1.0 1.0

effect was on phosphorus. The relative contribution from each
meat product on reductions in emissions is shown in Table 8.
The results presented in Table 8 show that reductions in beef
was most important to reduce the environmental impacts from
GHG and nutrient emissions. The shares of reductions from beef
are between 0.535 and 0.713. Beef is followed by pork for reduc-
tion in all pollutants. With respect to GHG emissions 8.5% of the
reductions were from pork, while the reductions from milk were
7.3%. In nitrogen and phosphorus, the shares of reductions from
pork were between 0.151 and 0.295 and the shares of milk 0.055
and 0.074 respectively.
Sensitivity analysis
Because of high levels of uncertainty regarding the estimated
levels of nitrogen and phosphorus emissions, sensitivity analyses
with other levels of N and P were conducted. The levels of nitrogen
emissions from beef, pork, poultry and milk, as calculated in Pierer
et al. (2014) were then used instead of the values given in Leip
et al. (2010). The new levels affected not only average emissions
per kilo of produce, but also tax levels and shares of aggregated
emissions compared to the total emission levels from all Swedish
production. Tax levels on meat product increased from the refer-
ence case, while the levels on dairy products decreased.
Another issue of concern is the assigned damage costs of the
pollutants. A reduction to SEK 0.55 per kg CO2e, as used in
Wirsenius et al. (2011) and an increase to the maximum level in
Stern (2006) of SEK 2.8 per kg CO2e were used to recalculate tax
levels and estimate new reduction levels. Results for alternate cost
of GHG emissions as well as the scenario with new nitrogen emis-
sions levels are presented in Fig. 5, where reductions in 1000 tons
are shown (see Tables A4–A7 in the appendix for all price
increases, new tax levels and changes in demand).
In the 0.55 SEK/kilo scenario, the decrease in emissions of the
pollutants range between 7.2% and 8.2%. In the scenario where
one kilo of CO2/e cost 2.8 SEK the reductions of GHG emissions
was just above 27%, nitrogen emissions reduced with almost 24%
and phosphorus emissions reduced by close to 23%, from the live-
stock sector. With the increase in the price, the relevance of beef
reductions increased, while all other decreased. However, the
higher level of damage costs gave such large increases in prices
that the calculated elasticity estimates could be less useful.
In the scenario of other values for nitrogen emissions, the
impacts on reductions in total emissions were not very large,
though the relative change from the reference scenario is. The
effects of the higher tax levels on meat products are seen mainly
on the importance of each product. In Table 9, the new shares of
reductions among products are presented. The main finding is that
the shares of pork and chicken increased, while the shares of beef
decreased. Pork was the most important for reducing phosphorus
in this scenario.
Discussion and summary
One way of dealing with environmental problems arising from
meat production is to introduce a Pigovian tax which covers mar-
ginal damage costs. This study calculated the impacts of introduc-
ing such a tax on seven different products (beef, pork, chicken,
milk, fermented products, cheese, and cream) in terms of emis-
sions of three pollutants (GHG, nitrogen, and phosphorus). It was
found that the marginal emission costs of Swedish meat from these
pollutants were between 1.8 and 32.5 SEK per kg produce. These
tax levels corresponded to 8.9% and 33.3% of initial prices. Not all
environmental damage was included in the costs. For example,
only costs of nutrient emissions to the Baltic Sea were included
and not local effects of nutrient emissions due to lack of data.
 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53 49

Fig. 5. Reductions in emissions of carbon dioxide equivalents (CO2e), nitrogen (N) and phosphorus (P) produced nationally (N/Pcountry) and entering the Baltic Sea (N/Pbaltic)
from introduction of an environmental tax on meats and dairy products. Reference case is the reduction in 1000-tons from Section ‘Emissions impacts in the reference case’.
New values for N show the reductions when emissions of N from Pierer et al. (2014) are used. The last two scenarios show the results of a change in the damage costs of GHG
from a min level = SEK 0.55 per kg CO2e to a max level = SEK 2.8 SEK per kg CO2e. All price effects are included. Observe the scaling for the different pollutants.

However, the main novelty in this paper was to include more pol- Table 9
Relative proportions of emissions of carbon dioxide equivalents (CO2e), nitrogen (N)
lutants than GHG gases. On the other hand, positive external
and phosphorus (P) produced nationally (N/Pcountry), and entering the Baltic Sea (N/
impacts from the agricultural landscape were not included. Pbaltic) from introduction of an environmental tax on meat and dairy products, all
However, given the reduction levels found from introduction of elasticities included.
the environmental taxes, it is not likely that all grazing animals
CO2/e Ncountry Nbaltic Pcountry Pbaltic
would disappear from Sweden. Only a proportion of all ruminants
Beef 0.721 0.599 0.595 0.444 0.446
graze on traditional fields, where they have a positive impact on
Pork 0.092 0.259 0.265 0.443 0.442
biodiversity. Chicken 0.014 0.078 0.077 0.071 0.071
The effects on demand and, hence, emissions were calculated by Milk 0.066 0.025 0.024 0.016 0.016
estimating a non-linear demand system for the seven products Fermented products 0.015 0.005 0.005 0.004 0.004
based on time series data from 1980. The results revealed rela- Cream 0.030 0.011 0.011 0.007 0.007
Cheese 0.062 0.023 0.023 0.015 0.015
tively low and inelastic (in absolute terms) own price elasticities
and higher income elasticities. Meat products were found to be Total 1.0 1.0 1.0 1.0 1.0

more sensitive in both prices and income than dairy products.

Taxing all seven products simultaneously could result in reduc-
tions up to 1.5% of all the included pollutants, compared to total
emissions in Sweden, and up to 12.1% from the livestock sector.
If reductions in demand mainly affected imported products, the
environmental gain in Sweden might be zero, while if reductions
affected only meat and dairy produced in Sweden the gain could
be important for the Swedish environment. If the latter is the out-
come, emission reductions could be even larger if change in land
use can take place (see Wirsenius et al., 2011). The most realistic
scenario in which reductions would take place is not discussed in
this paper. However, imposing the same taxes on all meat,
Swedish or imported, would most likely shift relative prices in
favour of Swedish meat because of the lower emissions per kg
meat consumed. There is also an ongoing discussion in Sweden
about eating locally produced meat, which might make the
Swedish population more hesitant to buy imported meat if the
price increased.
In the past, policy makers have shown resistance to imposing
taxes on food products, regardless of recommendations to reduce
meat consumption in order to mitigate GHG emissions, improve
food security or population health (e.g. UNEP, 2009; Röös and
Tjärnemo, 2011; WCRF, 2014). The reduction of meat and dairy
products in Sweden as a result of taxes, would serve to improve
the health for Swedish consumers. Especially regarding red meat
where consumption levels are much higher than what is recom-
mended (see for example Bjerselius et al., 2014; Larsson and
Wolk, 2012).
Consumption taxes as a means of achieving environmental tar-
gets should also be compared with other ways of reducing emis-
sions. One reason for having a meat tax on the consumption side
of the market is because of the political difficulties in taxing the
agricultural sector in many countries. However, even if we con-
sider only the costs of the policies, measured as decreases in con-
sumer surplus for a meat consumption tax and farmers’ costs of
50 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53

introducing measures targeting production, a consumption tax Table A1 (continued)

may have a relative advantage because of its effects on several pol-
Equation for Equation: EQAIDS2
lutants and thereby environmental targets. In general,
fermented dairy Dependent variable: S2
cost-effective solutions for achieving environmental targets are products Mean of dep. var. = .170264
calculated for one target at a time, without much consideration Std. dev. of dep. var. = .026388
of the eventual effects of these measures on other environmental Sum of squared
targets, such as studies on the costs of reaching EU climate targets residuals = .428237E03
Variance of residuals = .138141E04
(e.g. Böhringer et al., 2009) and Baltic Sea international agreements
Std. error of
(e.g. Elofsson, 2003). When accounting for the multi-pollutant regression = .371673E02
effects of a consumption tax on meat and dairy, it can have a rela- R-squared = .979593
tive cost advantage, in particular for environmental targets where LM het. test = 2.00380 [.157]
Durbin–Watson = 2.46163
livestock products account for a relatively large share of the total
Equation for cream Equation: EQAIDS3
emissions. This is not shown to be the case for nutrient emissions Dependent variable: S3
from meat and dairy to the Baltic Sea, though as mentioned above, Mean of dep. var. = .056326
we have reasons to assume the nutrient values included in this Std. dev. of dep. var. = .389366E02
paper were too low. The taxes investigated in this study would Sum of squared
residuals = .919804E04
contribute approximately 2.5% to achievement of the Swedish
Variance of residuals = .296711E05
Baltic Sea phosphorus reduction target of 530 ton (Helcom, Std. error of
2013), and almost 6.5% of the national commitment to reduction regression = .172253E02
of nitrogen emissions in the reference case and up to 14.8% when R-squared = .797806
LM het. test = .031368 [.859]
costs of CO2/e was 2.8 SEK/kilo (Swedish Environmental
Durbin–Watson = 1.74604
Protection Agency, 2014).
First stage Equation for meat Equation: EQAIDS1
aggregated Dependent variable: S1
food groups Mean of dep. var. = .150588
Appendix A Std. dev. of dep. var. = .652337E02
Sum of squared
residuals = .229109E03
See Tables A1–A7. Variance of residuals = .739060E05
Std. error of
regression = .271857E02
Table A1 R-squared = .820776
Test results for all stages in the demand system. The final equation in each system is LM het. test = .069042 [.793]
dropped. Durbin–Watson = 2.59913

Second stage Equation for beef Equation: EQAIDS1 Equation for dairy Equation: EQAIDS2
meat system Dependent variable: S1 Dependent variable: S2
Mean of dep. var. = .316795 Mean of dep. var. = .372837
Std. dev. of dep. var. = .017513 Std. dev. of dep. var. = .028384
Sum of squared Sum of squared
residuals = .189030E02 residuals = .473689E03
Variance of residuals = .590718E04 Variance of residuals = .152803E04
Std. error of Std. error of
regression = .768582E02 regression = .390900E02
R-squared = .808189 R-squared = .981186
LM het. test = .100458 [.751] LM het. test = .368503 [.544]
Durbin–Watson = 2.43598 Durbin–Watson = 2.03460

Equation for pork Equation: EQAIDS2 Equation for fruit Equation: EQAIDS3
Dependent variable: S2 Dependent variable: S3
Mean of dep. var. = .531004 Mean of dep. var. = .199469
Std. dev. of dep. var. = .044338 Std. dev. of dep. var. = .017930
Sum of squared Sum of squared
residuals = .214383E02 residuals = .117805E02
Variance of residuals = .669947E04 Variance of residuals = .380016E04
Std. error of Std. error of
regression = .818503E02 regression = .616454E02
R-squared = .965195 R-squared = .878049
LM het. test = .519872 [.471] LM het. test = .053345 [.817]
Durbin–Watson = 1.78869. Durbin–Watson = 2.07850
Equation for Equation: EQAIDS4
Second stage Equation for milk Equation: EQAIDS1 vegetables Dependent variable: S4
dairy system Dependent variable: S1 Mean of dep. var. = .130940
Mean of dep. var. = .676486 Std. dev. of dep. var. = .015093
Std. dev. of dep. var. = .043424 Sum of squared
Sum of squared residuals = .470180E03
residuals = .569860E03 Variance of residuals = .151671E04
Variance of residuals = .183826E04 Std. error of
Std. error of regression = .389450E02
regression = .428749E02 R-squared = .931756
R-squared = .989946 LM het. test = 1.27330 [.259]
LM het. test = .349012 [.555] Durbin–Watson = 2.81127
Durbin–Watson = 2.31316
 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53 51

Table A2
Parameter values for all demand systems.

Meat system a=a Number of observations = 32

b=b Log likelihood = 223.883
c=c Schwarz B.I.C. = 207.247
1 = beef
2 = pork Standard
3 = chicken Parameter Estimate Error t-statistic P-value
C11 .107970 .047167 2.28907 [.022]
C12 .115739 .035350 3.27407 [.001]
C22 .214159 .038836 5.5144 [.000]
A1 .370090 .111213 3.32777 [.001]
A2 .308453 .336571 .916459 [.359]
RHO .979152 .02962 33.0536 [.000]
B1 .014566 .054961 .026503 [.979]
B2 .035426 .059783 .592574 [.553]
B3 .033969 .055798 .608789 [.543]
A3 .321457 .261610 1.22876 [.219]
C13 .077695 .030930 .251201 [.802]
C23 .098420 .028989 3.39502 [.001]
C33 .090650 .034827 2.60284 [.009]
Dairy system a=a Number of observations = 31
b=b Log likelihood = 435.510
c=c Schwarz B.I.C. = 406.048
1 = milk
2 = fermented products Standard
3 = cream Parameter Estimate Error t-statistic P-value
4 = cheese C11 .207905 .032410 6.41477 [.000]
C12 .074400 .047340 1.57162 [.116]
C13 .043444 .013988 3.10568 [.002]
C22 .070279 .112797 .623062 [.533]
C23 .019892 .028452 .699169 [.484]
C33 .046130 .012720 3.62649 [.000]
A1 .319489 .307607 1.03863 [.299]
A2 .337703 .149250 2.26267 [.024]
A3 .119271 .057728 2.06608 [.039]
RHO .981718 .015767 62.2631 [.000]
B1 .013679 .017729 .771557 [.440]
B2 .045139 .014073 .320742 [.748]
B3 .025181 .063162 .398684 [.690]
B4 .066470 .073251 .907426 [.364]
A4 .223537 .109472 2.04195 [.041]
C14 .090062 .02107 4.27390 [.000]
C24 .024013 .051476 .466487 [.641]
C34 .017206 .012399 1.38768 [.165]
C44 .048843 .026207 1.86372 [.062]
Aggregated system a=a Number of observations = 31
b=b Log likelihood = 536.577
c=c Schwarz B.I.C. = 490.785
1 = meat
2 = dairy Standard
3 = fruit Parameter Estimate Error t-statistic P-value
4 = vegetables C11 .053013 .023454 2.26034 [.024]
5 = grain products C12 .029847 .014836 2.01174 [.044]
C13 .089234 .078903 1.13093 [.258]
C14 .057468 .010259 .560184 [.575]
C22 .204985 .018895 10.8488 [.000]
C23 .095526 .010517 .908270 [.364]
C24 .054015 .099333 5.43781 [.000]
C33 .046027 .016203 2.84067 [.005]
C34 .054179 .010312 .525406 [.599]
C44 .079142 .015715 5.03595 [.000]
A1 .193211 .023109 8.36074 [.000]
A2 .294962 .037892 7.78431 [.000]
A3 .148143 .036720 4.03442 [.000]
A4 .191484 .030161 6.34883 [.000]
RHO .967862 .010391 93.1473 [.000]
B1 .032548 .032433 1.00354 [.316]
B2 .179846 .047122 3.81661 [.000]
B3 .208078 .069190 3.00736 [.003]
B4 .016952 .044091 .384484 [.701]
B5 .012637 .042708 .295896 [.767]
A5 .172201 .025877 6.65446 [.000]
C15 .084963 .023493 .361661 [.718]
C25 .111571 .019348 5.76663 [.000]
C35 .022133 .010002 2.21286 [.027]
C45 .013962 .013715 1.01798 [.309]
C55 .156162 .032194 4.85063 [.000]
52 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53

Table A3 Appendix B. Supplementary material

Losses and retention levels for nitrogen (N) and phosphorus (P) to the Baltic Sea, as
well as local production share of beef, pork and chicken, in six regions in Sweden.
Supplementary data associated with this article can be found, in
Production share Losses Retention the online version, at http://dx.doi.org/10.1016/j.foodpol.2015.05.
Beef Pork Chicken N P N P 008.
0.042 0.047 0.044 0.051 0.019 0.23 0.40
0.163 0.176 0.048 0.085 0.025 0.27 0.40 References
0.229 0.259 0.228 0.164 0.013 0.60 0.47
0.238 0.161 0.296 0.164 0.013 0.60 0.47 Angrist, J., Krueger, A., 2001. Instrumental variables and the search for
0.063 0.067 0.077 0.276 0.016 0.30 0.00 identification: from supply and demand to natural experiments. J. Econ.
0.267 0.290 0.307 0.207 0.016 0.20 0.40 Perspect. 15, 69–85.
Bajželj, B., Richards, K.S., Allwood, J.M., Smith, P., Dennis, J.S., Curmi, E., Gilligan, C.A.,
2014. Importance of food-demand management for climate mitigation. Nat.
Clim. Change 4 (2014), 924–929.
Baumol, J.W., Oates, W.E., 1988. The Theory of Environmental Policy. Cambridge
Table A4 University Press, UK.
New tax calculations for sensitivity analysis regarding nitrogen levels. Nitrogen levels Bjerselius, R., Brugård Konde, Å., Sanner Färnstrand, J., 2014. Konsumtion av rött
from Pierer et al. (2014). kött och charkuteriprodukter och samband med tjock- och ändtarmscancer-
risk – och nyttohanteringsraport. National Food Agency Sweden Report 20-
Costs per kilo CO2/e Nbaltic Pbaltic Total SEK P0 % change 2014.
Böhringer, C., Löschel, A., Moslener, U., Rutherford, T., 2009. EU climate policy up to
Beef 24.29 3.32 0.49 28.10 97.70 28.77
2020: an economic impact assessment. Energy Econ. 31 (Suppl. 2), S295–S305.
Pork 3.74 1.79 0.59 6.12 59.3 10.31
Brandt, M., Ejhed, H., Rapp, L., 2008. Näringsbelastning på Östersjön och
Chicken 1.93 1.72 0.31 3.96 41 9.65 Västerhavet 2006. Swedish EPA Report 5815.
Milk 1.29 0.08 0.01 1.38 8 17.28 Briggs, A., Kehlbacher, A., Tiffin, R., Garnett, T., Rayner, M., Scarborough, P., 2013.
Fermented products 1.29 0.08 0.01 1.38 10.3 13.43 Incorporating the societal cost of greenhouse gases into the price of foods could
Cream 4.79 0.29 0.04 5.12 35 14.62 save lives from cardiovascular disease and cancer in England: a comparative
Cheese 9.58 0.58 0.08 10.23 76.4 13.39 risk assessment modelling study. BMJ Open 3, e003543.
Carpentier, A., Guyomard, H., 2001. Unconditional elasticities in two-stage demand
systems: an approximate solution. Am. J. Agric. Econ. 83, 222–229.
Cederberg, C., Sonesson, U., Henriksson, M., Sund, V., Davis, J., 2009. Greenhouse Gas
Emissions from Swedish Production of Meat, Milk and Eggs 1990 and 2005. SIK
Table A5 Report 793. ISBN: 987-91-7290-284-8.
Cederberg, C., Hedenus, F., Wirsenius, S., Sunesson, S., 2013. Trends in greenhouse
New tax calculations for sensitivity analysis regarding costs of CO2/e. One kilo of CO2/
gas emissions from consumption of animal food products – implications for
e is here assumed to cost 0.55 SEK.
long-term climate targets. Animal 7, 330–340.
Costs per kilo CO2/e Nbaltic NH3baltic Pbaltic Total SEK P0 % change Deaton, A., Muellbauer, J., 1980. An almost ideal demand system. Am. Econ. Rev. 70,
312–326.
Beef 13.36 3.52 3.81 0.87 21.57 97.7 22.07 Edgerton, D.L., 1997. Weak separability and the estimation of elasticities in
Pork 2.06 1.13 1.19 0.62 4.99 59.3 8.41 multistage demand systems. Am. J. Agric. Econ. 79, 62–79.
Chicken 1.06 0.73 0.75 0.22 2.76 41 6.74 Edjabou, L.D., Smed, S., 2013. The effect of using consumption taxes on foods to
Milk 0.71 0.22 0.23 0.05 1.21 8 15.11 promote climate friendly diets. The case of Denmark. Food Policy 39, 84–96.
Fermented 0.71 0.22 0.23 0.05 1.21 10.3 11.74 Elofsson, K., 2000. Cost Efficient Reductions of Stochastic Nutrient Loads to the
products Baltic Sea. Swedish University of Agricultural Sciences, Department of
Cream 2.63 0.80 0.87 0.17 4.48 35 12.79 Economics. Working Paper Series 6.
Elofsson, K., 2003. Cost effective control of stochastic agricultural loads to the Baltic
Cheese 5.27 1.60 1.74 0.35 8.95 76.4 11.72
Sea. Ecol. Econ. 47, 1–11.
Fabiosa, J., 2011. Globalization and trends in world food consumption. In: Lusk, J.,
Roosen, J., Shogren, J. (Eds.), The Oxford Handbook of the Economics of Food
Consumption and Policy. Oxford University Press, New York, pp. 591–611.
FAO, Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., Rosales, M., De Haan, C., 2006.
Livestock’s Long Shadow – Environmental Issues and Options. FAO, Agriculture
Table A6
and Consumer Protection Department.
New tax calculations for sensitivity analysis regarding costs of CO2/e. One kilo of CO2/
Galloway, J., Rownsend, A., Erisman, J., Bekunda, M., Cai, Z., Freney, J., Martinelli, L.,
e is here assumed to cost 2.8 SEK. Setizinger, S., Sutton, M., 2008. Transformation of the nitrogen cycle: recent
trends, questions, and potential solutions. Science 320, 889–892.
Costs per kilo CO2/e Nbaltic NH3baltic Pbaltic Total SEK P0 % change
Gerber, P.J., Steinfeld, H., Henderson, B., Mottet, A., Opio, C., Dijkman, J., Falucci, A.,
Beef 68.01 3.52 3.81 0.87 76.22 97.7 78.01 Tempio, G., 2013. Tackling Climate Change through Livestock – A Global
Pork 10.48 1.13 1.19 0.62 13.41 59.3 22.61 Assessment of Emissions and Mitigation Opportunities. Food and Agriculture
Chicken 5.40 0.73 0.75 0.22 7.10 41 17.32 Organization of the United Nations (FAO), Rome.
Milk 3.62 0.22 0.23 0.05 4.12 8 51.47 Green, R., Alston, J.M., 1990. Elasticities in AIDS model. Am. J. Agric. Econ. 72, 442–
Fermented 3.62 0.22 0.23 0.05 4.12 10.3 39.98 445.
Gren, I.M., 2004. Uniform or discriminating payment for environmental production
products
on arable land under asymmetric information. Eur. Rev. Agric. Econ. 31, 61–76.
Cream 13.41 0.80 0.87 0.17 15.25 35 43.58
Gren, I.M., Jonzon, Y., Lindqvist, M., 2008. Costs of Nutrient Reductions to the Baltic
Cheese 26.82 1.60 1.74 0.35 30.50 76.4 39.93
Sea-Technical Report. Swedish University of Agricultural Sciences, Department
of Economics. Working Papers Series 2008: 1 ISSN: 1401-4068.
Hedenus, F., Wirsenius, S., Johansson, D.A., 2014. The importance of reduced meat
and dairy consumption for meeting stringent climate change targets. Clim.
Table A7 Change 124, 79–91. http://dx.doi.org/10.1007/s10584-014-1104-5.
Reduction in 1000-tons for each scenario. Helcom, 2013. HELCOM Copenhagen Ministerial Declaration. <http://www.helcom.
fi/Documents/Ministerial2013/Ministerial%20declaration/2013%
1000 tons Total ton Total ton Total ton Total ton Total 20Copenhagen%20Ministerial%20Declaration%20w%20cover.pdf> (accessed
before reference Austrian N 0.55 ton 2.8 30.03.14).
Helfand, G., Perck, P., Maull, T., 2003. The theory of pollution policy. In: Mäler, K-G.,
Beef 233.5 44.3 38.15 29.59 103.15 Vincent, J. (Eds.), Handbook of Environmental Economics, vol. 1. North-Holland,
Pork 336.3 26.8 24.70 19.31 56.74 Amsterdam, The Netherlands.
Chicken 162.5 7.7 8.30 6.40 12.78 Kampas, A., White, B., 2004. Administrative cost and instrument choice for
Milk 895.8 53.2 41.07 35.92 122.36 stochastic non-point source pollutants. Environ. Resour. Econ. 27, 109–133.
Fermented 315.7 13.9 10.75 9.39 31.98 Larsson, S.C., Wolk, A., 2012. Red and processed meat consumption and risk of
products pancreatic cancer: meta-analysis of prospective studies. Brit. J. Cancer 106,
Cream 104.6 6.4 4.97 4.34 14.79 603–607.
Cheese 170.9 10.8 8.29 7.26 24.74 Leip, A., Weiss, F., Wassenaar, T., Perez, I., Fellmann, T., Loudjani, P., Tubiello, F.,
Grandgirard, D., Monni, S., Biala, K., 2010. Evaluation of the Livestock Sector’s
 S. Säll, I.-M. Gren / Food Policy 55 (2015) 41–53
Contribution to the EU Greenhouse Gas Emissions (GGELS) – Final Report.
Annex ch 6. European Commission, Joint Research Centre.
Lööv, H., Widell, L.M., 2009. Konsumtionsförändringar vid ändrade matpriser och
inkomster. Elasticitetsberäkningar för perioden 1960–2006. Swedish Board of
Agriculture, Report 2009:8.
Lukkarinen, J., Eklöf, P., 2011. Marknadsöversikt- nöt-. och kalvkött. Rapport
2011:32. Swedish Board of Agriculture.
Lukkarinen, J., Lannhard Öberg, Å., 2011. Marknadsöversikt- griskött. Rapport
2011:41. Swedish Board of Agriculture.
Lukkarinen, J., Lannhard Öberg, Å., 2012. Marknadsöversikt- mjölk och
mejeriprodukter. Report 2012:7. Swedish Board of Agriculture.
Lukkarinen, J., Burman, C., Johansson, T., Lannhard, Å., Jirskog, E., 2011.
Marknadsöversikt- fågelkött och ägg. Rapport 2011:31. Swedish Board of
Agriculture.
McMichael, A.J., Powels, J.W., Butler, C.D., Uauy, R., 2007. Food, livestock production,
energy, climate change, and health. Lancet 370, 1253–1263.
Murray, M., 2006. Avoiding invalid instruments and coping with weak instruments.
J. Econ. Perspect. 20, 111–132.
Myhre, G., Shindell, D., Bréon, F.-M., Collins, W., Fuglestvedt, J., Huang, J., Koch, D.,
Lamarque, J.-F., Lee, D., Mendoza, B., Nakajima, T., Robock, A., Stephens, G.,
Takemura, T., Zhang, H., 2013. Anthropogenic and natural radiative forcing. In:
Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., Nauels,
A., Xia, Y., Bex, V., Midgley, P.M. (Eds.), Climate Change 2013: The Physical
Science Basis. Contribution of Working Group I to the Fifth Assessment Report
of the Intergovernmental Panel on Climate Change. Cambridge University Press,
Cambridge, United Kingdom and New York, NY, USA.
Nguyen, T.L., Hermansen, J.E., Mogensen, L., 2011. Environmental costs of meat
production: the case of typical EU pork production. J. Clean. Prod. 28, p. 168(9)
ISSN: 0959-6526 (2012 June).
Nnoaham, K.E., Sacks, G., Rayner, M., Mytton, O., Gray, A., 2009. Modelling income
group differences in the health and economic impacts of targeted food taxes
and subsidies. Int. J. Epidemiol. 38, 1324–1333.
Nordström, J., Thunström, L., 2009. The impact of tax reforms designed to encourage
healthier grain consumption. J. Health Econ. 29, 622–634.
Nordström, J., Thunström, L., 2011. Can targeted food taxes and subsidies improve
the diet? Distributional effects among income groups. Food Policy 36, 259–271.
Pierer, M., Winiwarter, W., Leach, A.M., Galloway, J.N., 2014. The nitrogen footprint
of food products and general consumption patterns in Austria. Food Policy 49
(2014), 128–136.
Pigou, A.C., 1957. A Study in Public Finance, third ed. Macmillan (1947).
Rickertsen, K., 1995. Structural change and the demand for meat and fish in
Norway. Eur. Rev. Agric. Econ. 23, 316–330.
Röös, E., 2012. Mat-klimat-listan version 1.0 Rapport/Report 040. ISSN: 1654-9406.
Röös, E., Tjärnemo, H., 2011. Challenges of carbon labeling of food products: a
consumer research perspective. Brit. Food J. 113, 982–996.
53
Schmutzler, A., 1997. The choice between emission taxes and output taxes under
imperfect monitoring. J. Environ. Econ. Manage. 32, article no: EE960953.
Schou, J.S., Neye, S.T., Lundhede, T., Martinsen, L., Hasler, B., 2006. Modelling Cost-
efficient Reductions of Nutrient Loads to the Baltic Sea – Concept, Data and Cost
Functions for the Cost Minimisation Model. NERI Technical Report No. 592.
Schroeter, C., Lusk, J., Tyner, W., 2008. Determining the impact of food price and
income changes on body weight. J. Health Econ. 27, 45–68.
Staaf, H., Bergström, J., 2011. Emissions of Ammonia to Air in Sweden in 2009.
Statistics Sweden, ISSN: 1403-8978.
Statistics Sweden, 2010. Population Statistics Unit. Statistics Sweden, Tables on the
Population in Sweden 2009. ISSN: 1654-4358.
Statistics Sweden, 2013. Yearbook of Agricultural Statistics 201. ISSN: 1654-4382.
Steinfeld, H., Gerber, P., 2010. Livestock production and the global environment:
consume less or produce better? Proc. Natl. Acad. Sci. U.S.A. 107, 18237–18238.
Stern, N., 2006. Stern Review on the Economics of Climate Change. Cambridge
University Press.
Swedish Board of Agriculture, statistical database – konsumtion av livsmedel,
totalkonsumtion av vissa varor. <http://statistik.sjv.se/Database/
Jordbruksverket/databasetree.asp> (accessed 13.09.12).
Swedish Environmental Protection Agency, 2012. National Inventory Report
Sweden 2012.
Swedish Environmental Protection Agency, 2014. Begränsande utsläpp av
gränsöverskridande luftföroreningar i Europa. <http://www.miljomal.se/Hur-
nar-vi-malen/luftfororeningar/Begransade-utslapp-av-gransoverskridande-
luftfororeningar-i-Europa/> (accessed 31.03.14).
Turner, K., Paavola, J., Cooper, P., Farber, S., Jessamy, V., Georgiou, S., 2003. Valuing
nature: lessons learned and future research direction. Ecol. Econ. 46, 493–510.
UNEP, Nellmann, C., MacDevette, M., Manders, T., Eickhout, B., Svihus, B., Prins, A.G.,
Katlenborn, B.P., 2009. The Environmental Food Crisis – The Environment’s Role
in Averting Future Food Crises. UNEP/ GRID- Arendal. ISBN: 978-82-7701-054-
0.
Van Doorslaer, B., Witzke, P., Huck, I., Weiss, F., Fellmann, T., Salputra, G., Jansson, T.,
Drabik, D., Leip, A., 2015. An Economic Assessment of GHG Mitigation Policy
Options for EU Agriculture. EUR – Scientific and Technical Research Report.
ISBN: 978-92-79-45416-5.
WCRF, 2014. World Cancer Research Fund International. <http://www.wcrf.org/int/
research-we-fund/cancer-prevention-recommendations/animal-foods>
(accessed 16.01.15).
Weidema, B.P., 2009. Using the budget constraint to monetarise impact assessment
results. Ecol. Econ. 68, 1591–1598.
Wirsenius, S., Hedenus, F., Mohlin, K., 2011. Greenhouse gas taxes on animal food
products: rationale, tax scheme and climate mitigation effects. Clim. Change
108, 159–184.