23™ International System Dynamics Conference Boston MA, USA, 17.-21.07.2005
A composite optimisation-simulation model for the analysis of
the dynamic interactions in the Swiss Milk and Meat Market
Abstract
By 2011 Switzerland aims to liberalise the milk market which will result in market
changes in the basic conditions for agriculture. The impacts of the liberalisation are
investigated with a composite model obtained by combining an optimisation model for
the agricultural sector and a dynamic simulation model for the milk and meat market.
The calculations with the composite model indicate that milk price depends strongly on
the phasing out of market support, while the abolition of milk quotas in 2009 is less
decisive. An introduction of a dairy cow premium leads to a higher milk production,
especially with abolished milk quotas. In this case the European milk price level
represents the lower limit for the milk price in Switzerland. Compared to the milk
market, with falling quantities meat prices are likely to exhibit a stable development.
1. Introduction
By 2011 Switzerland aims to carry out a far-reaching liberalisation of the milk market,
the scope of which exceeds that foreseen in the EU and in other member countries of the
OECD. This plan involves three measures: 1. The current milk quota regulations will be
completely abolished in Switzerland by 2008, whereby organisations which have their
own quantity management can withdraw prematurely as per May 2006. 2. Mid-2007
will see the implementation of the bilateral agreements with the European Union
resulting in the complete liberalisation of the cheese trade between Switzerland and the
EU. 3. Parallel to the cuts in milk market support between 2004 and 2008 in the EU
(European Commission 2003), finance policy reasons will also lead to a reduction of
support for milk in Switzerland by 2007 and further cuts are to be expected up until
2011 (BLW 2003, BLW 2004). This will result in marked changes in the basic condi-
tions for Swiss agriculture in the coming years. In this environment the introduction of a
dairy cow premium to compensate income losses is discussed.
From the point of view of political and administrative decision-makers, it is very
interesting to know how farmers will react to the new basic conditions and how agri-
cultural production structures will develop in the next few years. Since the multifunc-
tional targets or functions of agriculture are defined in both the Swiss Federal Con-
stitution (Art. 104) and the laws on agriculture this is particularly relevant with regard to
the second point. For example, if it became apparent that the constitutional targets for
agriculture, such as the conservation of natural resources or the upkeep of landscape
could no longer guaranteed due to the liberalisation, it follows that the question would
arise regarding a modification or reformulation of the measures.
The impacts of liberalisation for two liberalisation scenarios are investigated by creating
price and supply forecasts with an optimisation model for the Swiss agricultural sector
and a dynamic simulation model for the milk and meat market. The composite model
obtained by combining the two methodological approaches is presented in this study. In
addition to describing and discussing the two approaches, the results relating to future
developments in the milk and meat market in Switzerland as well as developments in
sectoral land use and developments in livestock numbers are presented. The paper
closes with conclusions relating to both the content and the methodology.
2. Models
The output of an optimisation model is a statement of the best way to accomplish a goal.
Given the exogenous basic conditions indicated and the behaviour assumed in the
model, the results of a normative optimisation model provide a clear solution indicating
the best from among a well-defined set of alternatives (Sterman 1988). In the context of
this study, a sectoral optimisation model yields optimal land use and numbers of live-
stock. In addition to the limitations arising from their technical characteristics, the fun-
damental problem of sectoral optimisation models is that they need exogenous details
regarding not only product prices, but also conceming costs and the political measures
in favour of agriculture (Hazell and Norton 1986). Therefore, the models fail to capture
the interaction between farm reactions and impacts on the factor and product markets
(Zeddies 2003).
While it has been possible to use expert knowledge to obtain an exogenous assumption
of price development for the major agricultural products in the market environment in
which Swiss agriculture finds itself up to now, with its numerous domestic market
support regulations and trade policy measures, this will no longer be reasonable, or in-
deed possible in future. In particular, this applies to markets for products for which the
trade policy protection measures are reduced or completely rescinded in future and, at
the same time, the current quantity regulations governing production are abolished. In
this new market environment, the future development of prices depends directly on the
interrelationship between supply and demand. As a consequence the evaluation of future
supply developments in agriculture can no longer be based on exogenous price
assumptions.
Simulation models are primarily suitable for analysing a system. The purpose of a si-
mulation model is to mimic the real system so that its behavior can be studied. While
optimisation models are prescriptive, simulation models are descriptive. A simulation
model clarifies what would happen given a certain situation. The purpose of simulations
can be foresight (predicting how systems might behave in the future under assumed
conditions) or policy design (designing new decision-making strategies or organisa-
tional structures and evaluating their effect on the behavior of the system) (Sterman
1988). The results of the simulation in the context of this study reveal development
scope, or development prospects for the simulated system, the milk and meat market for
two liberalisation scenarios. Bouamra-Mechemache et al. (2002) describe the INRA-
Wageningen simulation system for the EU dairy sector. This simulation system consists
of two stand-alone models, one for milk and beef production on farms, the other for the
processing of milk into dairy products and their allocation between domestic and
foreign markets. The models simulate milk production, processing and market clearing
in 14 EU member states. The production model is a simulation model with econometri-
cally estimated behavioural equations. The processing and demand model breaks total
milk supplies down into fat and protein components. Market demands for milk products
drive the derived demands for the milk components, which are reconstituted at market
level. Domestic and world markets for 14 dairy products complete the model. The two
models can be run separately (to simulate quota scenarios) or interactively (no-quota
scenarios).
In order to predict future price and quantity developments simultaneously while giving
due consideration to these dependencies, the forecast of agricultural supply in Switzer-
land combines an optimisation and a simulation model. The sectoral optimisation model
described in section 2.1 uses exogenous prices to evaluate optimal land use and animal
husbandry. The simulation model described in section 2.2 endogenously calculates
prices and the interactions between the milk and meat market.
2.1 The Swiss Optimisation Model SILAS-dyn
Agroscope FAT Taenikon, the Swiss Federal Research Station for Agricultural
Economics and Engineering, has developed the Swiss agricultural sectoral information
and forecasting system (SILAS) on behalf of the Swiss Federal Office of Agriculture
since 1996. The model is used as a decision support system in connection with budget
funds planning for Switzerland's agricultural sector. The system is also used to analyse
the effects of policy measures on regional and sectoral production, factor input in agri-
culture and income (Malitius et al., 2001).
The concept of SILAS is based on regionally differentiated process analysis models, as
developed by Henrichsmeyer et al. (1996). These approaches are characterised by the
modelling of so-called "regional farms", the depiction of all the interconnecting rela-
tionships between production, outlay and production factor generation and utilisation as
well as the delimitation of the sector according to the concept of agricultural accounting
(Jacobs, 1998). The development to a recursive-dynamic model approach, SILAS-dyn,
with a 10 year forecast horizon, ensures that long-term factor utilisation decisions can
be optimised in the model (Mann et al., 2003).
The Swiss SILAS-dyn model bases the regional farms on eight agricultural areas de-
fined by increasingly difficult production and living conditions. These form the basis for
a large number of agricultural policy measures. This enables very accurate modelling of
the Swiss direct payments system, which is characterised by regionally graduated direct
payment approaches and contribution restrictions. Furthermore, the homogeneous pro-
duction potential of individual areas can be adequate represented in the model, as most
of the statistical data are available at this regional level.
The sector model depicts all the prevalent crop and animal production activities in
Switzerland. These are all differentiated according to organic and non-organic farming
methods, whereby extensive production variants are also taken into consideration in the
case of cereals, rapeseed and grassland. All the requirements a farm must meet with
regard to a balanced nutrient balance sheet are depicted by means of a fertiliser module.
A feed ration module ensures a correct, minimum cost feeding of all animals and a sec-
toral projection of commercial feed consumption and costs. A labour module optimises
the use of hired labour as a function of specific regional working time requirements and
available family labour. A building module optimises investments in stable construction
depending on the availability of existing buildings from previous years and building
requirements. Balance sheet equations at regional and sectoral levels ensure domestic
utilisation of all agricultural intermediate products. As Switzerland’s agricultural sector
is cut off from the EU market for agricultural intermediate products, no trade relations
with other countries are modelled.
The objective function maximises simultaneously the gross value added minus costs for
outside labour and the fixed costs of investments in replacements and expansion for all
eight areas, thus ensuring an optimal regional allocation of production. The positive
mathematical programming method (PMP) (Howitt, 1995) is used to calibrate the base
year to reality. To achieve this, complementary information from observation data is in-
corporated into the objective function in the form of a quadratic term (Rohm, 2003).
Therefore due to its methodological approach, the sector model lies midway between
pure normative optimisation models and positive econometric models (Cypris, 2000).
The mathematical model is:
2
Max Z, = > Pit Vit +> dX -> Victor -> Vigor Uta — > BinXin -05)) BinXin
iz ie rd [oy ye je
subject to
> AijnX jar S Da
7
Ye aX a = Ur + (1- Vo
ie
Vier Xjar Vir 2 [0]
Z= Objective function value e= Matrix technical coefficients of production
z= 1..8 (Number of regions) activities for using investment resources
p= Product price vector B= Matrix of parameters associated with the
d= Vector of direct payments quadratic term (PMP)
v= Vector of investment cost = Reciprocal value of useful life expectancy of
x= Vector of production activities (crop and livestock) investment activities
u= Vector of investment activities j 1...n (Number of production activities)
y= Vector of buying and selling activities i (Number of buying and selling activities)
c= Variable cost vector of production activities k (Number of investment activities)
a= Vector of parameters associated with the linear term 1 (Number of investment resources)
(PMP) t ...11 (Number of years)
a= Matrix technical coefficients of production activities t-1 = Previous year
for using production resources
b= Vector of available resources
The SILAS-dyn sector model represents a pure supply model for Swiss agriculture. It
optimises agricultural factor utilisation and production with exogenous, specified pro-
duct and factor prices, the development of which has up till now been periodically
estimated in advance by experts from the agricultural administration on the basis of
specialised knowledge. Variables relating to developments due to technical progress are
predicted using trend extrapolation. As already discussed in the Introduction, the far-
reaching reform and liberalisation of agricultural markets will make expert forecasts far
more difficult. Therefore the sector model for the milk and beef sectors is combined
with a partial market model which simulates future price development on the basis of
market development. Other product and factor prices up to 2011 are obtained using
estimates provided by exogenous experts from the agricultural administration sector.
These anticipate that the Swiss price index for agricultural products will fall steadily by
15 to 20% between the initial year 2001 and 2011 due to further reforms of the markets.
In contrast, the price index for factor prices is only expected to fall by about 3% since it
is unlikely that there will be any great changes in the consumer environment if
Switzerland goes it alone.
2.2. Dynamic Simulation Model for the Milk and Meat Market
A dynamic simulation model is used for the simulation of future price and quantity
developments in the milk and meat market under differing basic agricultural policy
conditions. Comparable applications for system dynamics in the market development
sector are to be found primarily in the relevant A merican literature. One example of this
is the work of Pagel et al. (2002) which investigates the influence of agricultural policy
measures on the development of the American milk production and milk market using a
dynamic simulation model. By way of comparison, the work of Nicholson and
Fiddaman (2003) focuses on the volatility of the milk price in the American milk
market and the identification of those factors which exert the greatest influence factors.
The structure of the partial simulation model for the milk and meat market corresponds
to that of the FAPRI model (Fapri 2004). The model, as in the case of FAPRI, is a non-
spatial policy model. Price formation for the partial markets is realised via market
clearing, whereby due consideration is likewise given to foreign trade and the associated
restrictions as well as meat imports within the tariff quota. In contrast to the FAPRI or
GTAP models which, applied to the milk market, distinguish between butter, cheese and
milk (Fapri 2004a) or between milk production and milk processing (GTAP; Lips
2002), the simulation model permits a detailed depiction of demand and the associated
relevant support and any changes it undergoes.
The simulation model consists of the following eight subsystems:
1. Herd model including feeding and feed balance
2.- 4. | Meat supply, Meat demand and Meat market (veal and beef)
5.- 7. Milk supply, Demand dairy products and Milk market
8. Decisions
Figure 1 presents the aggregate structure of the simulation model. The figure does not
include the subsystem ‘decisions’. It also does not show the feedback loops between
prices on the two markets and the remaining six subsystems. These relationships are
shown in figure 2.
stock meat Supply veal and Demand meat
production beef imports
Breeding 5 Domestic demand Demand veal and
meat beef
; Domestic demand Demand milk
Livestock milk dairy products
production
|
Milk production Supply Export demand Demand mitk
traded milk dairy products imports
Figure 1: Subsystem diagram of the simulation model for the milk and meat market
In addition to the development of the milk and meat market, the mutual dependencies of
these two partial markets are of special interest. To this end, figure 2 provides a detailed
illustration of the two partial markets for milk as well as for beef and veal. The figure
illustrates the relationship between prices, proceeds and cost developments and numbers
of livestock on the one hand, while on the other hand it shows the market dependencies
between supply and demand. Imports and exports of milk products are endogenously
incorporated for the milk market.
Figure 2 shows that the interactions within and between the milk and meat market are
driven by a series of balancing feedback loops. These loops control the reinforcing
feedback loop in the herd model that allocates livestock units to meat or milk
production. In the absence of the price control mechanisms animal husbandry would
concentrate entirely on either of the two categories given an incremental disturbance of
an initial equilibrium. The balancing feedback loops operate on several levels such as
domestic and export markets. This constitutes a major source of instability because the
adaptation times on these markets and within the herd model are very different. The
concem about oscillatory behaviour on the two markets arises especially in a liberalised
market environment where the feedback loops controlling the ratio between domestic
production and import will become much more dominant.
domestic
demand meat y,
pA
tariff quota
atts import
ed supply meat ~ domestic
(. price meat (5 4 i
de a
import meat ~v.
direct " ¥ :
payments livestock units +
meat production,
¢ + import tariffs
av -
proceeds-cost Aig
ratio milk/meat -
wh > import dairy
i * products
livestock units A
production mpilk production 8
costs + +4 demand,
he traded milk y
. F + domestic demand
supply rN dairy products
traded milk 8 aoc
4s aN 8
price milk export dairy
<) products
Figure 2: Demand, supply and price formation on the milk and meat market
The development of supplies of milk and meat is depicted by means of an aggregated
herd model for the entire sector (not shown in figure 2). On the consumption side, the
aggregated demand for beef and veal at wholesale trade level is relevant. The cal-
culation of demand is based on four factors: 1. the prices estimated in the model, 2. the
initial market balance 2001, 3. price developments for the competitor products pork and
poultry, and 4. demand elasticity. In addition, it is assumed in the model that additional
imports will compete against domestic produce for demand. In the case of the milk
market, the model considers seven distinct product branches from the consumption
point of view: drinking milk, fresh milk products and milk specialities, cream, butter,
cream and soft cheeses, hard and semi-hard cheeses as well as long-life milk products.
In these partial markets, demand for domestic products is depicted as well as imports
and exports, whereby the simulations for the milk market are realised at wholesale trade
level. Thus, the processors' demand for milk for a product branch is calculated in the
model, whereby demand depends on price development, initial balance 2001 and de-
mand elasticities. As in the case of domestic demand, demands for cheese exports into
the EU and non-EU countries as well as for cheese imports are subject to iso-elastic de-
mand functions. With the exception of the impact generated by competition from
imports, changes in demand in the simulation only occur as the result of price adjust-
ments at wholesale trade level. Price developments also depend on support of the milk
market and border protection measures.
The simulation of the price for beef, veal and milk as illustrated in figure 2 represents a
central element of the model. It is assumed that the quantities of milk and meat pro-
duced must correspond to medium and long term demand. A short term supply surplus
leads to a price slump. On the other hand, additional demand generates a rise in prices
and possibly an increase in production, whereby the numbers of livestock must be ad-
justed accordingly. While the price for beef and veal is based on overall supply and
aggregated demand, the price formation for milk is obtained separately for the four
partial markets, drinking milk and milk in fresh milk products, milk for cheese-making,
milk in butter as well as milk in long-life milk products. In the model, price differences
are offset by transferring milk to partial markets with a higher price. From a medium
and long term point of view, the resulting adjustment of milk utilisation leads to a price
balance within the partial markets.
2.3 Description and Discussion of the Composite Model
The composite model presented in this paper consists of the sectoral forecasting system
SILAS-dyn and the simulation model for the milk and meat market. It permits a genuine
improvement in forecasting developments in agricultural prices and supplies in Switzer-
land. The main reason for this is that the two model approaches, both of which are
characterised by specific strengths and weaknesses, combine to complement one
another. The optimisation model, with its depiction of the competition for scarce pro-
duction factors, facilitates exact forecasts regarding supply developments in animal
husbandry and arable farming. However, the forecasts are based on exogenous price
assumptions, which remain constant regardless of the respective supply reaction. On the
other hand, the price forecasts which are obtained from the simulations with the market
model for the milk and meat market only cover the supply and demand changes in the
two partial markets. Therefore, modifications resulting from changing competitive
forces between animal husbandry and arable farming or activities within the scope of
ecological compensation are not covered. However, the latter is the strong point of the
sectoral optimisation model, which optimises future supply developments under the
fixed production factors available and the respective basic conditions.
Forecasting of future price and supply developments in Swiss agriculture is therefore
carried out using an iterative combination of the two models (see Figure 3): To start
with, the market model simulates price forecasts only (1st run). These represent initial
values which are used for forecasts with the sectoral optimisation model and form the
basis for the formulation of new milk and beef supply forecasts. The supply forecasts
for milk and beef are then re-entered in the market model in order to formulate new
price forecasts (2nd run), whereby iterative price and supply forecasts are formulated
with both models until prices and quantities no longer change. Due to the interchange of
results between the simulation model and SILAS-dyn, medium and long term market
balances can be delimited with a relatively high degree of precision. This applies in
particular to medium and long term developments in production quantities, as these can
be calculated much more exactly with SILAS-dyn than with the simulation model. On
the other hand, the simulation model is highly effective for the calculation of future
price developments in the milk and meat market.
Price forecast (1st run)
for milk and beef
Supply forecast (1st run)
for milk and beef
Price forecast (2nd run)
for milk and beef
Supply forecast (2nd run)
for milk and beef
Figure 3: Structure of the iterative combi
3. Development of the Milk and Meat Market in Switzerland up until 2011
The investigation of the impact of the liberalisation on sectoral supplies of milk and
beef as well as domestic prices for milk and beef was carried out using calculations with
the composite model. The calculations are based on two liberalisation scenarios which
were elaborated in collaboration with decision-makers within the agricultural admini-
stration and observing a time horizon up until 2011. The first scenario (Scenario I)
studies a 25% reduction in milk market support in Switzerland by 2007. The implemen-
tation of the bilateral agreements means that support payments for the export of cheese
into the EU must be phased out completely. There are no further cuts between 2007 and
2011. After a slight increase between 2001 and 2007 milk quotas are abolished with no
transitional phase as per 2008. On the other hand, the second scenario (Scenario II)
considers the complete discontinuation of milk market support by 2007, the abolition of
milk quotas as per 2008 as well as the introduction of a dairy cow premium amounting
to CHF 700 per LSU from 2008. The funds which are no longer needed to support the
milk market are to be used to finance the dairy cow premium. With this adaptation, the
premium is paid for all cattle. Up to now, dairy were excluded.
3.1 Milk and Meat Market in Switzerland up until 2011 in Scenario I
Table 1 and Figure 4 show the price forecasts for Scenario I with a 25% withdrawal of
market support and the abolition of milk quotas as per 2008. In the first run, which was
performed with no link-up to the sectoral optimisation model (see Figure 3), milk price
forecasts obtained using the simulation model show a 19% fall in prices by 2007. In
spite of the abolition of milk quotas, there are no further noteworthy price changes
between 2007 and 2011. Price development can be explained by two factors: 1. In spite
of the fall in milk prices, the export price for cheese (hard and semi-hard cheeses)
increases due to the elimination of contributions for cheese exports and the reduction of
market support. 2. The liberalisation of the cheese market and the decreasing market
support in the EU lead to a rise in cheese imports into Switzerland resulting in
increasing price pressure on the domestic cheese market. In the case of beef, the market
model shows a steady 13% decline in prices from 2004 up until 2011. This fall in prices
is due to increased supplies, whereby imports nevertheless remain constant.
Table 1: Domestic and Export Prices (in CHF per tonne; 1 Euro = 1.55 CHF) for Milk
Products in 2001 and 2011 for Scenario I (1st Run)
Products Domestic price Export price Domestic price Export price
2001 2001 EU 2011 2011 EU
Drinking milk 800 657
Fresh milk products 800 657
Cream 748 619
Butter 608 516
Hard and semi hard cheeses 568 407 486 486
Cream and soft cheeses 517 410 486 486
Long-life milk products 645 544
Milk price CH 800 657
The supply forecasts based on the prices obtained from the 1st run with the sector model
SILAS-dyn show that the number of dairy cows producing traded milk will fall by 11%
in the first half of the forecast up until 2007 due to declining milk prices (-19%). After
this, numbers will rise again by 5% with no changes in the price of milk up until 2011.
Finally, by 2011 numbers will settle at a level of 94% of those registered in 2001. The
increasing number of dairy cows in the second forecast period indicates that, with prices
in general on the decline and stable milk prices, the competitivity of keeping dairy cows
rises slightly. This development in the number of animals will result in a decline in milk
quantities to 3.2 million tonnes by 2007. Due to the growing number of dairy cows and
increasing milk yields, milk quantities can be expected to rise to roughly 3.55 million
tonnes between 2008 and 2011. Furthermore, the calculations with the agricultural sec-
tor model SILA S-dyn show that there is a close competitive relationship between dairy
and suckler cow farming. On the one hand, milk price developments lead to an increase
in cows used to produce traded milk, while on the other hand suckler cows are ousted.
From an overall point of view, the number of cows remains relatively constant up until
2011. This also means that no great changes are to be expected in beef quantities up to
this date.
New price forecasts (2nd run) are formulated using the simulation model taking the
milk and beef production determined with the sectoral optimisation model as their basis.
Figure 3 shows that the increase in milk production calculated for the second forecast
period leads to a lower milk price in the price simulations, while the price for beef is
slightly higher than in the first run. In the second run, the milk price will amount to
roughly 0.60 CHF per kg in 2011 while prices for beef will rise slightly.
The supply forecasts for milk and beef based on the new price forecasts (2nd run) show
that they are clearly lower than the first quantity estimates. New price forecasts based
on these supply quantities from the 2nd run reveal that prices for milk and beef then lie
at an intermediate level. Thus it may be expected by 2011 that the price for milk will lie
between 0.60-0.65 CHF per kg. According to the SILAS-dyn forecasts, milk production
will amount to 3.3 - 3.5 million tonnes by 2011. In addition, it is clear that in spite of
the abolition of milk quotas, a partial liberalisation of the milk market will lead to a
10
decline in the number of dairy cows and that this will not be offset by an increase in the
number of suckler cows. Therefore, a steady fall by about 9% in beef supplies is to be
expected in Scenario I.
Milk price Beef price
850 8000
a a pe
g 750 7000
5 700 La i i,
5 6509 }—__“@—» , |
bo | PE] i
5 600 5
550 5500
500 5000
2000 2002 2004 2006 2008 2010 2012 2000 2002 2004 2006 2008 2010
—@ Price forecasts Ist run —t— Price forecasts 2nd run > Price forecasts 3rd run
2012
Figure 4: Price forecasts for milk and beef using the simulation model for Scenario I
Dairy cows
50000 ny. Suckler cows
200000
600000
150000 +
q 550:000 + g
2 500000 § tooo
450000 50000
400000 0
2000 2002 2004 +2006 «2008-2010 2012 3000 2002 2004 ©2008-2008 20102012
Milk production Beef production
38 160
155 +
bas 2 | bi aS
& $ us
8 : 40
2 334 135
| = = 130
125
3.0 120
2000 2002 2004 2006 2008 2010 2012 2000 = 2002 2004 2006 2008 2010 2012
+ SILAS-dyn forecasts using price forecasts Ist run
—# SILAS-dyn forecasts using price forecasts 3rd nu
“ae SILAS-dyn forecasts using price forecasts 2nd run
Figure 5: Number of cows plus milk and beef supplies using the sector model for
Scenario I
3.2 Milk and Meat Market in Switzerland up until 2011 in Scenario II
The main difference between Scenario II and Scenario I is that market support for milk
is phased out gradually by 2008 and a new dairy cow payment is granted.
Figure 6 shows that price forecasts obtained using the market simulation without link-
up to the sectoral optimisation model (1st run) in Scenario II amount to 0.51 CHF per
kg for milk in 2011. Although the foreseen abolition of market support leads to a fall in
the price of milk, this is relatively low due to the reduced milk quantities in the simula-
11
tion. In contrast, the optimisation model indicates that, at this price, a considerably
higher number of dairy cows is to be expected due to the abolition of milk quotas and
the payment of a dairy cow contribution (CHF 700 per cow) from 2008 onwards. Given
an average milk yield of 6000 kg per cow and year, the payment corresponds to a
compensation of 0.12 CHF per kg. Therefore, compared to Scenario I, there is a
noticeably higher number of dairy cows with dairy cow payments than with product
price support, even though milk prices are similar. Evidently, when milk prices exceed
0.50 CHF per kg, animal-specific payments represent an incentive to expand milk
production; especially after the abolition of quotas and slightly lower payments for
keeping other bovines (suckler cows or cattle for fattening). On the other hand, due to a
shortage of funds no increase is to be expected in the number of suckler cows and their
numbers will even out at the 2001 level. Therefore, for the sector as a whole, the beef
production in the 1st run remains constant up until 2011. Even when milk quotas are
abolished, it would be possible to achieve a marked increase in the quantity of traded
milk, namely 3.9 million tonnes, with the same number of cows as in 2001 merely by
exploiting higher milk yields and by limiting the amount of milk used for feeding
purposes.
If milk quantities rise to over 3.9 million tonnes, the calculations obtained with the
simulation model indicate that milk prices can be expected to fall to the EU level in
2011 (see Table 2). In the case of beef, the simulation model indicates constant market
prices up until 2011 with unchanged production. Given these milk and beef prices, and
in spite of the payment of a dairy cow contribution, the SILAS-dyn calculations show a
steady 14% decline in the number of dairy cows to about 530000 animals by 2010, after
which only a slight increase is to be expected. At the same time, the decline in numbers
of dairy cows on the one hand and stable beef prices on the other lead to a 25% increase
in the number of suckler cows. Therefore, given a milk price of 0.40 CHF per kg, the
milk production calculated with the sectoral optimisation model SILAS-dyn falls to
roughly 3.3 million tonnes in the 2nd run and is thus somewhat lower than the quantity
produced in 2001. The decline in the number of cows in the sector would lead to a 7%
fall in beef production.
In the 3rd run of the market simulation, the decline in milk quantities forecast in the 2nd
run with SILAS-dyn results in a new milk price of 0.49 CHF per kg in 2011 which,
combined with a dairy cow payment of CHF 700 per LSU and the abolition of milk
quotas, leads to a long term build up in the numbers of animals. In this case, the
abolition of milk quotas also results in an increase in the quantity of milk produced and
this would clearly put pressure on the milk market.
12
Milk price Beef price
1000 10000
wn OSS
[ 600 : 6000
a a
400 + 4000
‘a &
5 &
200 2000
0 0
2000 © 20022004» «2006-2008. « «20102012 2000 2002-2004» 2006 »«« 2008-2010 «2012
-@-Price forecasts Istrun = Price forecasts 2ndrun a Price forecasts 3rd run
Figure 6: Price forecasts for milk and beef using the simulation model for Scenario II
Dairy cows Suckler cows
650000 150000
a 100000
550000 i
A casa sown
450000
400000 %
2000 ©2002-2004 ©2006.» 2008-2010 2012 2000 2002 2004 2006 2008 2010 2012
43 Milk production Beef production
160
i “0 g 190
§ 38 g
g 2 140
£354 8
East = 130
30 120
2000 2002-«-2004-~««-2006-=«-2008-«2010-»«o1.-~«=«2000-«2002«2004-««-2006. «2008S 20102012
—@- SILAS-dyn forecasts using price forecasts Istrun <a SILAS-dyn forecasts using price forecasts 2nd run
= SILAS. dyn forecasts using price forecasts 3
Figure 7: Number of cows plus milk and beef supplies using the sector model for
Scenario II
3.3 Discussion of the Development of the Milk Market in Switzerland
Cheese exports into the EU and cheese imports play a major role in the development of
milk prices in Switzerland. The first depend strongly on the relationship between the
milk price for Swiss export cheese and the European price for milk for cheese-making.
Table 2 shows the development of the milk price in Switzerland and the export price for
cheese in both Scenarios. In addition, the anticipated price for milk for cheese-making
in the EU is also indicated. Thus, a comparative price can be calculated via the Swiss
allowances and the preferences of Swiss consumers for domestic products.
The direct comparison of prices shows that the export price for cheese in Scenario I
exceeds the European price for milk for cheese-making (price EU-5). On the other hand,
13
the domestic milk price obtained from the market simulation is only slightly higher than
the comparative price (price EU-5 plus allowances and preference). A further decline in
cheese exports is to be expected due to export price differences and this will increase
pressure on milk prices in Switzerland. Pressure on prices is influenced by domestic
milk production and by the preferences of Swiss consumers for home produce. If milk
production in Switzerland continues at today's level and preferences for Swiss products
are low, the milk price will fall below the comparative price plus allowances of 0.56
CHF per kg by the year 2011.
In Scenario II, the calculations using both the SILAS-dyn optimisation model and the
market model indicate that the milk price and domestic milk production will be highly
volatile. The abolition of milk support by 2008 means that price developments depend
directly on domestic milk production: If lower quantities of milk are produced than in
the initial year there is less pressure on prices and it can then be expected that prices in
Switzerland are slightly higher than in the EU, whereby the magnitude of the difference
depends primarily on the preference of Swiss consumers for milk products "made in
Switzerland". On the other hand, if it is assumed that milk production remains
practically the same as in 2001, the price of milk in Switzerland would probably come
quite close to the European price. Due to the dairy cow premium foreseen in Scenario
II, the quantity of milk produced in 2011 will be the same as in the initial year, even if
the price for milk settles at 0.45 CHF per kg.
Table 2: Development of the milk price and the price of cheese for export into the EU as
well as the European comparative price
Milk Price (CHF per tonne)
2001 | 2002 | 2003 | 2004 | 2005 | 2006 | 2007 | 2008 | 2009 | 2010 | 2011
Scenario I
Milk price CH 800 | 782 | 729 | 686 | 666 | 657 | 650 | 652 | 653 | 657 | 657
Export price for cheese 407 | 417 | 439 | 465 | 475 | 476 | 476 | 477 | 481 | 485 | 486
Milk price EU-5" 507 [| 469 | 475 [| 478 | 457 | 434 | 412 | 395 | 394 | 394 | 304
Milk price EU-5 plus
allowances CH 739 | 689 | 683 | 675 | 642 | 607 | 582 | 565 | 564 | 564 | 564
Milk price EU-5 plus
allowances CH and
domestic preference” 815 | 765 | 759 | 751 | 718 | 683 | 658 | 641 | 640 | 640 | 640
Scenario II
Milk price CH 800 | 782 | 729 | 686 | 666 | 645 | 591 | 553 | 520 | 505 | 515
Export price for cheese 407 | 417 | 439 | 465 | 475 | 473 | 467 | 473 | 490 | 504 | 513
Milk price EU-5" 507 | 469 | 475 | 478 | 457 | 434 | 412 | 395 | 394 | 394 |) 394
Milk price EU-5 plus
allowances CH 739 | 689 | 683 | 675 | 642 | 573 | 504 | 441 | 394 | 394 | 304
Milk price EU-5 plus
allowances CH and
domestic preference” 815 | 765 | 759 | 751 | 718 | 649 | 580 | 517 | 470 | 470 | 470
Remark: ! Milk price EU-5: anticipated milk price for Germany, France, Italy, Austria and Belgium or the
Netherlands.
? The domestic preference amounts to CHF 76 per tonne raw milk.
14
4, Summary and Conclusions
There will be major changes in the basic conditions for Swiss agriculture in the coming
years. These changes are not only due to the liberalisation of agricultural market fore-
seen within the scope of current WTO negotiations, but are also attributable to the
implementation of the bilateral agreements with the EU on the one hand and Swiss
financial policy on the other. Shifts in agricultural production structures will be one
direct result of these changes in basic conditions, whereby the liberalisation of the milk
market will probably be particularly significant. As medium and long term flexibility
rises, both the basic conditions for agriculture and technical progress are relevant for the
decision-making process from the farmers' point of view. The development of agricul-
tural production structures confronts decision-makers at political and administrative
levels with the question of ensuring the multifunctional targets of agriculture and the
possible modification of agricultural policy measures. Decisions are based on ex-ante
forecasts of future developments in agriculture under the expected basic conditions. In
this paper, a composite model consisting of a sectoral optimisation model and a simu-
lation model for the milk and meat market serves as the methodological instrument used
to provide this scientifically-based policy advisory service. The main task of this com-
posite model is to assess future developments in the milk and meat market, sectoral land
utilisation and numbers of livestock up until the year 2011 in the event of an abolition
of milk quotas and a possible reallocation of the current milk market support in
Switzerland.
From a content point of view, the calculations with the composite model indicate that
the development of milk and meat prices depends strongly on the degree of liberali-
sation to which the milk market is subjected and, consequently, on the phasing out of
market support in Switzerland. The abolition of milk quotas is less decisive. A partial
liberalisation (25% withdrawal of market support) would lead to milk prices in a range
of 0.60-0.65 CHF per kg and thus considerably higher than the prices for 2011
calculated on the basis of the price forecasts obtained from the FAPRI model (FAPRI,
2004b), the Model AGLINK (OECD, 2004) and forecasts of the European Commission
(2004). These forecasts indicate a price of about 0.41 CHF per kg. In addition to the
factors already mentioned, the development of milk prices in Switzerland is highly
dependent on the preferences of consumers both at home and abroad for Swiss dairy
produce. If we assume a low degree of preference, the milk price in Switzerland will
come close to the European price. In the case of a partial liberalisation, the allowances
which continue to be disbursed will make a difference, whereby their support effect
depends directly on domestic milk production. Compared to the milk market, meat
prices are likely to exhibit a stable development. Basically, with falling meat quantities
and the specified imports (WTO tariff quota), the calculations in the composite model
indicate that meat prices can be expected to remain steady or even to rise. However,
under the current organisation of the import system, it is probable that additional
imports would be permitted if meat prices were to go up. As in the case of milk, the
amount of these additional imports depends on the Swiss consumers’ preference for
Swiss meat. However, since the origin of meat is a decisive factor in the decision to buy
(BAG 2003), imports of meat may not be increased arbitrarily.
The model calculations indicate that milk prices and milk production would exhibit a
relatively high degree of volatility in the event of a complete liberalisation of the milk
market and the transformation of current support into a dairy cow premium. In this case
15
the cow premium represents an incentive to expand milk production, especially in
combination with the abolition of milk quotas. Given free access to the cheese market,
the European price level represents the lower limit for the milk price. This leads to a
milk price of at least 40 CHF per kg in the year 2011. On condition that the milk
quantity does not change by 2011, the market simulations indicate a price of 47 CHF
per kg; however at this price, the sectoral calculations with SILAS result in a slightly
higher milk quantity as the competitivity of milk production is enhanced by the intro-
duced dairy cow payments. Consequently, the milk price is likely to amount to roughly
40 to 45 CHF per kg, whereby the influence of consumer preferences is a decisive factor
in this case as well.
From the methodological point of view, the conclusions pertain to the choice of a
suitable methodological procedure and the combination of a sectoral optimisation model
with a dynamic simulation model as presented in this paper. Basically, it must be stated
that sectoral optimisation models, such as the SILAS-dyn model presented here, are
ideal instruments for a scientifically based policy advice. In particular, this aspect gains
in importance within the context of changing basic conditions and the associated mo-
difications in production structures which must be implemented in future. The principal
strength of sectoral optimisation models lies in the detailed assessment of development
in agricultural supplies and production structures. Furthermore the optimisation also
indicates the optimal allocation of production factors which are in short supply. Given
the exogenous specified basic conditions, factor allocation is realised on the basis of the
competitive strength of the activities and their changes. However, the latter depend
heavily on the exogenous assumed product prices. Due to the limitations of the
optimisation model on the supply side, there is no interaction between supply and price
developments, which is detrimental to the validity of the forecasts.
The demand side must be included to mitigate this weak point; in this paper, this aspect
of the milk and meat market is solved by using a dynamic simulation model. The funda-
mental strength of the market model lies in the model-endogenous simulation of future
price developments giving due consideration to demand and supply developments. The
latter can be taken over from the sectoral optimisation model. The results presented here
show that an iterative link-up of the two models for price and quantity forecasts is pos-
sible and that future price and quantity developments can be delimited in this way. An
extension of this procedure to other product markets should therefore be investigated.
The main advantage of linking the two model approaches together lies in the deliberate
combination of their individual strengths; on the one hand, the detailed depiction of
supply development obtained from the optimisation model and on the other hand, the
price simulation based on an exact assessment of supply and demand developments. In
this way, by using a combination of the two model approaches, it is possible to forecast,
or rather delimit, future market and production structure developments in agriculture
with a higher degree of precision. This is an important point since price and supply fore-
casts will be far more difficult in a liberalised market environment. However, political
decision-makers require the most precise assessment possible regarding future develop-
ments.
From the methodological point of view, the market simulation requires further research,
in particular concerning the demand function and elasticities used for assessing demand.
The empirical bases which are available today are founded entirely on investigations
16
carried out largely in a state-controlled market environment. Thus, the question arises
conceming the extent to which the consumers' reactions in a liberalised product market
can be explained on the basis of previous demand behaviour. The most vital aspect is
the future development of imports and exports in a completely liberalised market, and
thus the consumers' preferences for Swiss dairy products.
By way of contrast, surveys on the development of agricultural production structures
reveal clearly that not only economic, but also numerous non-economic factors are
highly relevant to farmers when they make decisions regarding production and invest-
ments. However, these factors are, to a large extent, not included in the sectoral optimi-
sation model or are entered as constant over time with the positive mathematical pro-
gramming approach. It thus becomes apparent that there is a need for research to pro-
vide an adequate, reality-orientated depiction of decision-making behaviour at the farm
level in the optimisation model. A system dynamics modelling approach can contribute
significantly to this task.
References
BAG Bundesamt fiir Gesundheit (ed.), 2003. Die Herkunft der Lebensmittel beeinflusst
das Kaufverhalten. Bulletin fiir Gesundheit 35/03.
BLW Bundesamt fiir Landwirtschaft (ed.). 2003, 2004. Agrarbericht des Bundesamtes
fiir Landwirtschaft. Bern.
Bouamra-Mechemache, Z., Requillart, V., Burrell, A., Guyomard, H., Jongeneel, R.,
2002. The INRA-Wageningen simulation system for the EU dairy sector. Abstract in:
Salamon, P., 2002. Proposal for a Thematic session “Modelling the phasing out of milk
quotas”. In: CD, containing the papers and abstracts of the EAAE Congress at Zaragoza
28-31 August 2002.
Cypris, C. 2000. Positive Mathematische Programmierung (PMP) im Agrarsek-
tormodell RA UMIS. Forschungsgesellschaft fiir Agrarpolitik und Agrarsoziologie e. V.,
Bonn.
European Commission. 2003. Reform of the Common agricultural policy - A long-term
Perspective for Sustainable Agriculture: Impact analysis, Brussels
(http://europa.eu.int/comm/agriculture/publi/reports/reformimpact/rep_en.pdf, Stand 10.
January 2005).
European Commission, Directorate General for Agriculture 2004. Prospects for
Agricultural Markets 2004-2011 Update for EU-25, Brussels.
FAPRI Food and Agricultural Policy Research Institute 2004a. Structure of the
FAPRI/CARD International Dairy Model and Intemational Livestock and Poultry
Model, Ames IOWA (http://www.fapri.iastate.edu/models/, Stand 10. January 2005).
FAPRI Food and Agricultural Policy Research Institute 2004b. Agricultural Outlook
2004; (http://www.fapri.iastate.edu/outlook2004/, Stand 15. November 2004).
Hazell, P.B.R.. Norton, R. D., 1986. Mathematical Programming for Economic
Analysis in Agriculture. Macmillan Publishing Company, New York. Collier
Macmillan Publishers London.
17
Henrichsmeyer, W., Isermeyer, F., Neander, E. and Manegold, D. (eds) 1996. Weiter-
entwicklung des gesamtdeutschen Agrarsektormodells RAUMIS96, Endbericht,
Bonn/Braunschweig.
Howitt, R.E., 1995. Positive Mathematical Programming. American Journal of
Agricultural Economics 77: 329-342.
Jacobs, A., 1998. Paralleler Einsatz von Regionen- und Betriebsgruppenmodellen in der
Agrarsektoranalyse. In: BMELF (ed.) Angewandte Wissenschaft, Reihe A, Heft 470,
Landwirtschaftsverlag, Minster-Hiltrup.
Lips, M., 2002. Die Auswirkungen der Neuen Agrarhandelsrunde der Welt-
handelsorganisation auf die Schweiz. Eine Anwendung des allgemeinen
Gleichgewichtsmodells GTAP. Shaker Verlag. Aachen.
Malitius, O., Mack, G., Moresino, M., 2001. The Swiss agricultural model SILAS: an
example of quantitative decision support systems for policy makers. In Heckelei, H.,
Witzke, P. and Henrichsmeyer W. (eds), Agricultural Sector Modelling and Policy
Information Systems. Kiel: Wissenschaftsverlag V auk Kiel KG, 169-176.
Mann, S., Mack, G., Ferjani, A., 2003. Kénnen Produktionsentscheidungen als
Investitionsentscheidungen modelliert werden? Agrarwirtschaft 52 (7): 333-341.
Nicholson, C., Fiddaman, T. 2003. Modeling Dairy Price Volatility. Paper presented at
the 21st International Conference of the System Dynamics Society, New Y ork City,
July 20-24, 2003.
OECD (2004): OECD Agricultural Outlook 2004-2013. OECD, Paris.
Pagel, E., Nicholson, C., Newton, P. (2002): Structural Changes in U.S. Dairy Farming
and the Role of Government Policy. Paper presented at the System Dynamics Society
Annual Meetings, Palermo, Italy, July 29-31, 2002.
Rohm, O., Dabbert, S., 2003. Integrating Agri-Environmental Programs Into Regional
Production Models: An Extension of Positive mathematical Programming. American
Journal of Agricultural Economics 85 (1): 254-265.
Sterman J. D. 2000. Business Dynamics. Systems thinking and modeling for a complex
world. McGraw-Hill, Boston.
Sterman J.D., 1988. A skeptic’s guide to computer models. In: Barney G.O., Kreutzer
W.B., Garrett MJ. (eds.). Managing a nation: The microcomputer software catalog.
Westview Press, Boulder CO: 209-229.
Zeddies, J., 2003. Modellgestitzte Politikberatung in der Agrar- und Agrarumwelt-
politik. Agrarwirtschaft 52 (4): 173-174.
18
