Agricultural Systems 103 (2010) 585–597
Contents lists available at ScienceDirect
Agricultural Systems
journal homepage: www.elsevier.com/locate/agsy
FSSIM, a bio-economic farm model for simulating the response
of EU farming systems to agricultural and environmental policies
Kamel Louhichi a,*,1, Argyris Kanellopoulos b,c,*,1,2, Sander Janssen b,c, Guillermo Flichman d, Maria Blanco e,
Huib Hengsdijk f, Thomas Heckelei g, Paul Berentsen b, Alfons Oude Lansink b, Martin Van Ittersum c
a
INRA-UMR Economie Publique, Avenue Lucien Brétignières, 78850 Thiverval-Grignon, France
Business Economics, Wageningen University, Hollandseweg 1, 6706 KN Wageningen, The Netherlands
Plant Production Systems, Wageningen University, P.O. Box 430, 6700 AK Wageningen, The Netherlands
d
IAMM-CIHEAM, 3191 Route de Mende, 34093 Cedex 5, Montpellier, France
e
European Commission Joint Research Centre (JRC), IPTS, Edificio Expo, Inca Garcilaso 3, 41092 Sevilla, Spain
f
Plant Research International, Wageningen University, P.O. Box 616, 6700 AK Wageningen, The Netherlands
g
University of Bonn, Economic and Agricultural Policy, Nussallee 21, 53115 Bonn, Germany
b
c
a r t i c l e
i n f o
Article history:
Received 16 July 2009
Received in revised form 18 June 2010
Accepted 22 June 2010
Available online 23 July 2010
Keywords:
Bio-economic model
Integrated assessment
Environmental policy
Market liberalization
a b s t r a c t
The disciplinary nature of most existing farm models as well as the issue specific orientation of most of
the studies in agricultural systems research are main reasons for the limited use and re-use of bio-economic modelling for the ex-ante integrated assessment of policy decisions. The objective of this article
is to present a bio-economic farm model that is generic and re-usable for different bio-physical and
socio-economic contexts, facilitating the linking of micro and macro analysis or to provide detailed analysis of farming systems in a specific region. Model use is illustrated in this paper with an analysis of the
impacts of the CAP reform of 2003 for arable and livestock farms in a context of market liberalization.
Results from the application of the model to representative farms in Flevoland (the Netherlands) and
Midi-Pyrenees (France) shows that CAP reform 2003 under market liberalization will cause substantial
substitution of root crops and durum wheat by vegetables and oilseed crops. Much of the set-aside area
will be put into production intensifying the existing farming systems. Abolishment of the milk quota system will cause an increase of the average herd size. The average total gross margin of farm types in Flevoland decreases while the average total gross margin of farms in Midi-Pyrenees increases. The results
show that the model can simulate arable and livestock farm types of two regions different from a biophysical and socio-economic point of view and it can deal with a variety of policy instruments. The examples show that the model can be (re-)used as a basis for future research and as a comprehensive tool for
future policy analysis.
Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction
Governments and policy agencies attempt to assess consequences of new policies before their introduction. The European
Commission has formalized this through a mandatory ex-ante impact assessment of its new agricultural and environmental policies
(EC, 2005). Science can contribute to these governmental demands
for impact assessment by developing tools that can, in a transparent, rigorous and repeatable fashion, make impact assessments
of agricultural and environmental policies better informed. Bioeconomic farm models have been proposed for such ex-ante
* Corresponding authors. Tel.: +33 130815333; fax: +33 130815368.
E-mail addresses: klouhichi@grignon.inra.fr (K. Louhichi), argyris.kanellopoulos@wur.nl (A. Kanellopoulos).
1
These authors contributed equally to this article.
2
Tel.: +31(0)317 482141; fax: +31(0)317 484892.
0308-521X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.agsy.2010.06.006
assessments (Flichman and Jacquet, 2003; Janssen and Van Ittersum, 2007) and many recent applications (Donaldson et al.,
1995; Flichman, 1996; Judez et al., 2001;Berentsen, 2003; Veysset
et al., 2005; Onate et al., 2006; Riesgo and Gomez-Limon, 2006;
Semaan et al., 2007) assess the impacts of policy changes on economic, environmental and social indicators of agricultural systems.
If a bio-economic farm model is to be used as a basis for such exante assessments of agricultural and environmental policies at
European level, some requirements must be fulfilled, i.e. it must
be possible to upscale the model’s results (e.g. product supply) to
higher system levels (e.g. country or market); data with respect
to farm types, their locations and production activities must be
readily available throughout various regions; the model must be
applicable to different farm types including mixed farm types;
the application and calibration of the model should not require
many ad hoc steps or unjustified strict calibration constraints,
and finally it must be possible to assess many different policy
586
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
instruments. In short, it must be possible to use and apply the same
bio-economic farm model in a consistent way across the European
Union (EU).
A literature review showed that a generic model meeting the
above requirements does not exist (Janssen and Van Ittersum,
2007). Some models focus on simulating specific farm types without providing much opportunities to expand their application beyond the original target domain (e.g. Donaldson et al., 1995;
Veysset et al., 2005), while other models require extensive data
collection limiting a rapid operationalization (e.g. Riesgo and Gomez-Limon, 2006). Various model applications address very specific EU policy issues and do not allow the assessment of a range
of interrelated policy questions that EU decision-makers face (Topp
and Mitchell, 2003; Onate et al., 2006).
Each of these models (Donaldson et al., 1995; Topp and Mitchell, 2003; Veysset et al., 2005; Riesgo and Gomez-Limon, 2006) has
strengths that made them suitable to be used for specific data-sets
and applications. In trying to extend their use to other policies
questions and locations, this specificity causes problems. With
the limitations of existing approaches in mind, this article has
the following two objectives. The first objective is to present the
Farm System SIMulator (FSSIM) which aims to be a generic bioeconomic farm model that can be applied in combination with
higher level models to assess, ex-ante, a variety of policy questions
under different bio-physical and socio-economic conditions. The
second objective is to demonstrate the applicability of the model
as a stand alone tool to assess farm level impact of future policy
scenarios for different farm types in different regions. FSSIM has
been developed as part of the integrated modelling framework of
the System for Environmental and Agricultural Modelling: Linking
European Science and Society (SEAMLESS) (Van Ittersum et al.,
2008) which targets to integrated assessment of agricultural systems in the EU of 27 member states (EU27). This implies that
FSSIM can be and has been linked to other models for multi-scale
analyses (Pérez Domínguez et al., 2009).
In Section 2, the SEAMLESS context and the requirements for a
model like FSSIM are presented to justify the modelling choices. In
Section 3, FSSIM is described. In Section 4, the model is used to
simulate arable and dairy farms of Flevoland (The Netherlands)
and Midi-Pyrenees (France). In Section 5, the results of the application of FSSIM are described. Section 6 discusses the results and
concludes.
2. Model requirements following from the SEAMLESS Integrated
Framework
The main objective of the SEAMLESS Integrated Framework
(SEAMLESS-IF) is to enable ex-ante evaluation of a broad range of
agricultural and environmental policies at multiple decision making levels. This framework consists of models which operate in
an iterative way (Fig. 1). First, the Common Agricultural Policy
Regionalized Impact modelling system (CAPRI) which is an EU agricultural sector model (Britz et al., 2007) is used to estimate a set of
initial prices for the agricultural products of all EU27 regions. Second, FSSIM uses the estimated prices and calculates supply re-
sponses of farms to price shocks in a selection of EU 27 regions.
Third, EXPAMOD (Pérez Domínguez et al., 2009) is used to extrapolate results of the sample regions to all EU27 regions by means of
econometric approaches. Next, CAPRI is recalibrated with the new
supply responses coming from EXPAMOD to generate a set of market clearing prices that are used by FSSIM for the final run.
Modelling all individual farms within EU27 is not possible because of the large number of farms and the existing variation and
diversification among farming systems. Therefore, a farm typology
was developed associating economic and environmental characteristics of EU farms. This farm typology is based on the existing EU
farm typology (EEC, 1985) which classifies farms according to their
income and specialization. This farm typology has been enriched
with environmental criteria related to the land use and intensity
of farming (Andersen et al., 2007).
A spatial allocation procedure was developed to geo-reference
farm types allowing the aggregation of model results at farm type
level to both natural (territorial) and administrative regional level
(Elbersen et al., 2006; Hazeu et al., 2010). FSSIM is used to simulate
an ‘‘average farm” which is a virtual (not observed in reality) farm
derived by averaging data from the Farm Accountancy Data Network (FADN) of farms that are grouped in the same farm type. It
is assumed that the ‘‘average farm” represents all farms that belong
to the same farm type. Structural changes in the sector are related
to interactions between farms (land market) and alternative income sources and can only be taken into account by using another
model (Zimmermann et al., 2009) of SEAMLESS-IF. However, policy
makers can compare the gross margin of an average farm calculated by FSSIM with the estimated income from non-farming activities to draw conclusions on the viability of the particular average
farm.
The general context of SEAMLESS and the variety of policy questions that FSSIM should be able to address leads to a number of
model requirements. First, FSSIM must be integrated with the
other models of SEAMLESS-IF. The integration with components
at field and market level must be streamlined in terms of methodology (e.g. temporal and spatial scales), concepts and scenarios
being used and software. Second, the conceptual design of FSSIM
should be ‘‘generic” so that the model can be easily modified and
used for assessing different policies under various socio-economic
and bio-physical conditions (e.g. different farm types and different
regions). Third, production activities and available technology
must be described in an explicit and transparent way to improve
the explanatory power of the model. Fourth, the data needs of
FSSIM should be preferably limited to those data available at
EU27 level minimizing the resource demanding process of data
collection. The model must be robust enough to function with data
like those from FADN. Moreover, the model should be capable to
exploit more detailed data that is not currently available but might
become available in the future. Finally, FSSIM should be easily
adaptable and re-usable (modularity). This will allow model users
to easily change it to account for different regions, farming systems, and policies.
3. Model description
3.1. Model specification
Fig. 1. SEAMLESS model chain (Pérez Domínguez et al., 2009). See text for
explanation.
FSSIM is an optimization model which maximizes a farm’s total
gross margin subject to a set of resource and policy constraints. Total gross margin is defined as total revenues including sales from
agricultural products and compensatory payments (subsidies)
minus total variable costs from crop and animal production. Total
variable costs include costs of fertilizers, costs of irrigation water,
costs of crop protection, costs of seeds and plant material, costs
of animal feed and costs of hired labour. A quadratic objective
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
function is used to account for increasing variable costs per unit of
production because of inadequate machinery and management
capacity and decreasing yields due to land heterogeneity (Howitt,
1995). The general mathematical formulation of FSSIM is presented below:
maximize Z ¼ w0 x x0 Qx
subject to Ax 6 b;
xP0
ð1Þ
where Z is the total gross margin, w is the n 1 vector of the parameters of the linear part of the activities’ gross margin, Q is the n n
matrix of the parameters of the quadratic part of the activities’ gross
margin, x is a n 1 vector of the simulated levels of the agricultural
activities, A is a m n matrix of technical coefficients, and b is a
m 1 vector of available resources and upper bounds to the policy
constraints.
A different model formulation has already been implemented
and can be used if detailed agro-management information is available or if it is important to account for the risk averse attitude of
the farmer explicitly. In this model formulation the farmer’s utility
is maximized. Utility is defined as gross margin minus risk. For this
specification a linear gross margin function is assumed.
maximize U ¼ w0 x ur
subject to Ax 6 b;
xP0
ð2Þ
where u is the risk aversion parameter that assumes constant absolute risk aversion (Hazell and Norton, 1986) and r is the standard
deviation of the total gross margin.
FSSIM consists of four major components, i.e. arable production,
livestock production, policies and regulations and the calibration
and forecasting component which are described below.
3.1.1. Arable production
In FSSIM, arable agricultural activities are defined as crop rotations grown under specific soil and climate conditions and under
well-defined management describing major field operations in detail. It is assumed that in each year, all crops of a rotation are grown
on equal shares of the land. A model solution can include several
crop rotations. The concept of crop rotations allows to account
for temporal interactions between crops. The agricultural management of arable activities describes operations associated with fertilization, soil preparation, sowing, harvesting, irrigation and pest
management of crops and results in different inputs and outputs.
FSSIM uses information available in FADN. This data source
lacks detail in agro-management information which is needed to
assess the environmental aspects of production. Therefore, a simple survey was performed within SEAMLESS to identify and quantify current production activities (Borkowski et al., 2007; Zander
et al., 2010). For operational purposes and due to resource limitations the survey was conducted for a sample of 16 NUTS2 regions
from the EU27 (NUTS: Nomenclature of Territorial Units for Statistics). Experts from the sampled regions were asked to specify the
most important rotations and related management which are currently used by arable farms in their region. In total 87 rotations of
21 different crops were identified in the sampled regions.
The agricultural management component of FSSIM (FSSIM-AM)
and the Agricultural Production Externalities Simulator (APES)
(Janssen et al., 2009b) can be used to quantify externalities of current activities (e.g. N-leaching) and complete sets of discrete input
and outputs coefficients (e.g. costs, labour requirements, input of
agrochemicals, yields, externalities) for alternative activities which
have improved performance in one or more criteria. Alternative
arable activities may include new crops and rotations, changes in
crop management or their combination resulting in activities with
different technical coefficients. Alternative activities are used to account for technological innovations in agriculture (e.g. new varie-
587
ties, modern agricultural practices) and effects of future changes
to bio-physical and climatic conditions (e.g. effects of climate
change or soil degradation to production).
Arable farmers face a number of resource scarcities that affect
their decision making. These resource scarcities have been taken
into account in FSSIM by means of constraints. The available arable
land constraint is specified per soil type and ensures that the sum
of the area of the activities on a certain soil does not exceed the
available farm land for this soil type. The available land is derived
from FADN and hence imposed exogenously. Selling or buying of
land is not considered in FSSIM. However, pre-determined scenarios with more (in case of buying) or less (in case of selling) available land can be tested. The available irrigated land constraint
ensures that the area with irrigated activities does not exceed
the available irrigable land. The available amount of irrigation
water constraint ensures that the total volume of water required
for the irrigated activities does not exceed the available water volume. Finally, the labour constraint is used to calculate the number
of hours of hired labour, given the labour requirements of different
activities and the availability of family labour. Hired labour is considered as an additional cost, the price of which is equal to the
average region-specific wage rate. Allocation of family labour to
off-farm activities is not considered in FSSIM. Scenarios can be
used to assess consequences of allocating family labour to off-farm
activities by changing the availability of family labour for agricultural activities.
3.1.2. Livestock production
Three different animal activities are modelled in FSSIM, i.e. dairy,
beef, and small ruminants (sheep and goats). The core element of a
dairy activity is a productive cow, a bull and their off-springs. A
replacement rate is based on the actual milk production per cow
and sets the share of young animals in a dairy activity i.e. calves
and heifers. For example, a typical dairy activity in Flevoland may
consist of 60.5% cows, 17.5% heifers, 20.8% calves and 1.2% bulls.
Increasing the activity level by 1 unit will cause an increase in the
number of all animals so that the share of animals in the activity remains constant. Feed requirements of different animal types and
decisions on the length of the grazing period are also taken into account in a dairy activity. The feed requirements of the herd in terms
of fibre, energy and protein are covered by roughage produced onfarm (fresh, hay or silage), purchased roughage (hay or silage), concentrates produced on-farm or purchased concentrates. Feed crops
like grass and fodder maize are grown either in a rotation with other
crops or as mono-crop activities. The quantities of on-farm produced
and purchased feed depend mainly on prices of crop product (including feed) and input prices. Beef activities are modelled in a similar
way. Two distinct methods of raising animals for beef production
are available i.e. a suckler system comprising a cow and its offsprings, and a fattening system, which merely fattens purchased
young animals till the moment of selling. The small ruminant activities for meat and milk production are modelled in a way similar to
dairy and beef activities. The milk and meat production is used to
determine an appropriate replacement rate and the feed requirements of different animals (Thorne et al., 2009).
FADN data are used to identify the predominant livestock activities across the regions of EU, and to derive related animal shares,
production levels and replacement rates. The SEAMLESS survey
(Borkowski et al., 2007) and a feed evaluation and animal nutrition
system proposed by Jarrige (1989) were used to quantify the technical coefficients of animal activities like yields, total production
costs, costs of feed, feed nutrient values and feed requirements
(Thorne et al., 2009).
A number of constraints were used to model the on-farm availability of resources, the feed production and the animal’s diet. Constraints relating feed availability to feed requirements are used to
588
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
secure that the total requirements of energy, protein and fibre are
met by the produced (on-farm) and purchased quantities of feed
and concentrates. Another constraint (maximum amount of concentrates) is used to set an upper bound to the share of concentrates in the animal’s diet to prevent animal diseases related to
high amounts of concentrate. The available amount of roughage
constraint restricts the grazing period to a region-specific maximum. Finally, the milk quota constraint restricts the produced
quantity of milk to the available milk quota. Any milk production
exceeding the milk quota is penalized. This constraint is the main
limiting factor for a dairy farm and for that reason it is mentioned
here as a resource constraint.
3.1.3. Policies and regulations
FSSIM is able to simulate a broad range of agricultural and environmental policy instruments, some of which have been already
implemented in practice while others might be of interest to policy
makers in the future. These policies are modelled as additional
constraints and variables in a generic way to account easily for various products or region-specific policy implementation. The policy
instruments which are currently modelled in FSSIM can be classified in a number of groups.
The first group of policies modelled in FSSIM includes the EU
compensation payments which are taken into account as part of
revenues in the objective function of the model. Existing compensation payments related to rain-fed and irrigated land, historical
yield but also the degree of the payments that is linked to production (coupling) are taken into account in order to calculate the total
amount of received payments according to the existing regime.
Two farm support policies are already programmed in FSSIM, the
farm support policy under the Agenda 2000 (CEC, 1999a,b) and
the reform of the Common Agricultural Policy of 2003 (CAP reform
of 2003) (CEC, 2003; OECD, 2004). The first CAP reform of 1992
(CEC, 1991) and the market liberalization led to a reduction of
product prices. Therefore, a regime of direct payments was developed to compensate farm income within the general context of the
Agenda 2000. These direct payments were given to the highly affected arable and livestock sectors of the EU and they were linked
either to production or to the area of different crops. The direct
payments are financed by the EU and administered by the ministry
or department of agriculture of each member state. Modelling the
regional specific implementation of the Agenda 2000 requires two
pieces of information: the way the payment was given (i.e. per
activity level, per unit of main output) and the amount of the payment (basic premium) per hectare, slaughtered animal or tonne of
product. The CAP reform of 2003 replaces the Agenda 2000 regime
and involves mainly the partial (or total for some crops) decoupling of subsidies from production. To calculate subsidies under
the CAP reform of 2003 in FSSIM, the subsidies received under
the Agenda 2000 were (partially or totally) detached from production. To achieve this, the new coupling degree of each product was
used. The decoupled part of the payment is based on the historical
reference land and the total amount of subsidies received over the
years 2000–2002. The coupled and the decoupled payment of each
activity were used to calculate the total received subsidies per
hectare of activity under the CAP reform of 2003.
The second group of policy instruments that has been modelled
in FSSIM relates to quota based policies which are currently used in
many EU countries to regulate the price and supply of certain products like milk and sugar beet. This kind of regulation was also used
under Agenda 2000. In FSSIM quota based policies are taken into
account with additional constraints. The part of production that
exceeds the pre-determined quotas gets a lower price according
to the specificities of the regulation. The same structure of the constraint set is used for all products that are currently under a quota
regulation (or might be in the future).
Another policy that has been included in FSSIM is the obligatory
set-aside policy which was introduced by the EU in 1988 (i) to reduce the large and costly cereal surpluses produced under the
guaranteed price system of the CAP reform of 1992 and (ii) to provide environmental benefits following considerable damage to
agro-ecosystems and nature as a result of the intensification of
agriculture. Although the implementation of the set-aside policy
differs across the EU, in general, the measure entails the obligation
to leave a proportion of the farm land uncultivated or assigned to
non-food purposes for a certain period in exchange for subsidy
payments. The obligatory set-aside policy is taken into account in
FSSIM by setting a lower bound to the area which is left as setaside and by adding an extra source of revenues in the objective
function for each hectare of set-aside. If the area of set-aside is less
than 10% of the area of Cereals, Oil seed and Protein (COP) crops a
subsidy cut is assumed.
The last group of policies modelled in FSSIM is related to the
environmental conditions and cross-compliance regulations which
aim at sustaining various agro-environmental conditions that must
be respected to avoid reduced farm support payment under the
CAP reform of 2003. Cross-compliance regulations must be in line
with a number of well-defined standards determined at EU level
and cover environmental, food safety, crop protection, animal
health and animal welfare issues. Cross-compliance regulations
are taken into account mainly by additional constraints while in
some cases binary variables are needed transforming the model
into a Mixed Integer Non-Linear Programming (MINLP) model.
In addition to the above described policy instruments, a number
of environmental indicators (e.g. total nitrogen use, water use, pesticide use), indicators related to biodiversity and multi-functionality (e.g. number of crop species on the farm), and socio-economic
indicators (e.g. labour use per hour) are assessed. Those indicators
can be easily used to evaluate future environmental policies.
3.2. Calibration and forecasting
A Positive Mathematical Programming (PMP) based approach is
used to calibrate the model and guarantee exact reproduction of
the observed (base year) situation without using additional calibration constraints which are difficult to justify in a way consistent
with existing economic theory (Heckelei, 2003). PMP is a generic
and fully automated procedure which means that it can be easily
adapted and used for different regions and farm types without
additional site specific information.
In PMP calibrated models, the observed activity levels of farm
types are used to calculate unobserved non-linear costs which
are omitted from the linear cost function of LP models because of
data limitations and simplification purposes. Non-linear costs are
related to issues like managerial capacity, fixed costs (e.g. machinery, buildings) and risk. PMP uses a two step approach. In the first
step, a number of calibration constraints are added to the model, to
ensure that the observed activity levels of the base year are reproduced. In the second step, the calibration constraints are taken out
and their shadow prices are used to specify and include the nonlinear costs in the objective function. Since the first introduction
of PMP to bio-economic modelling by Howitt (1995) a number of
PMP variants have been developed based on different assumptions
resulting in different model forecasts (Heckelei and Wolff, 2003;
Röhm and Dabbert, 2003; Kanellopoulos et al., 2010). The appropriateness of PMP variants is case specific and depends on the
available data and policy question. In FSSIM a number of PMP variants are programmed providing users with various options.
A different calibration procedure is used for the model presented in (2) where the risk aversion coefficient is the only unknown parameter. The risk aversion parameter is estimated in an
iterative process that involves multiple model runs. In each model
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
run a different value of the risk aversion coefficient is used; the value of the risk aversion coefficient that gives the best fit in terms of
crop allocation is selected for simulations. In this case, exact calibration is not guaranteed.
After the model has been calibrated it can be used for forecasting. Inflation of input and output prices is considered, while exogenous to the model information on yield and price trends are used
to account for possible technological innovations and price–supply
fluctuations.
To facilitate the analysis of policy scenarios, FSSIM is set-up in
such a way that policy makers and model users can easily access
and adapt the constraint set and the parameters of the model.
New policy scenarios can be incorporated into the model by: (i)
varying the available farm resources, (ii) changing the input and
output coefficients for activities, (iii) abolishing base year policies,
and (iv) including new policies, constraints and parameters. A set
of general policies has been pre-programmed and is ready to use
after having provided the required data.
Fig. 2 shows a simple presentation of the model set-up for a
simulation of an arable farm in year 2003, where gross margin is
maximized (risk aversion is not taken into account) subject to a
number of resource and policy constraints. This presentation reveals the general structure of the model and summarizes the required information that is stored in an integrated database
developed within SEAMLESS (Janssen et al., 2009a). Switching on
and off different components of the model allows different simulation of the same or a different farm type (e.g. the livestock component is switched on in the case where a livestock farm type is
simulated).
Fig. 2. Set-up of FSSIM for simulating an arable farm type under Agenda 2000.
Switching on (j) and off (h) components and constraints results in different
simulations of a different farm type. Summary of the required information stored in
the SEAMLESS database.
589
4. Set-up of the calculations
Here, we present the application of FSSIM to arable and dairy
farm types in Flevoland (The Netherlands) and Midi-Pyrenees
(France). The bio-physical (climate and soil types) and socio-economic (different size, intensity and specialization of arable and
livestock farms) conditions of these two regions differ substantially. We have chosen these regions to illustrate the applicability
of the model under different bio-physical and socio-economic conditions and demonstrate the generic features of the model enabling
the simulation of other farm types within the EU. For this exercise,
we did not use a bio-physical model to estimate externalities because of data limitations and simplification purposes. Instead, we
used total nitrogen input and total irrigation water input as environmental indicators. The model specification with a non-linear total gross margin function, described in (1), was selected for this
exercise because exact calibration is guaranteed. This model specification is used for higher level analysis where data is limited and
calibration only on the risk aversion parameter is not adequate to
reproduce what is observed in reality. FSSIM was calibrated for the
base year (2003) with the PMP variant proposed by Kanellopoulos
et al. (2010), using activity specific supply elasticities from the literature (Jansson, 2007). For this exercise we used exogenous base
year prices and consequently we did not use the full procedure described in Fig. 1.
We use the four digits codes of the SEAMLESS farm typology to
distinguish between the different farm types. The first digit of the
farm type code refers to the farm size: (3) Large farms, i.e. size > 40
European Size Units (ESU), (2) Medium farms (16 ESU 6 size 6
40 ESU), (1) small farms (size < 16 ESU). The second digit refers to
farm intensity: (3) High intensity (output > 3000 €/ha), (2) Medium
intensity (500 €/ha 6 output 6 3000 €/ha), (1) Low intensity (output < 500 €/ha). The two last digits refer to farm specialization:
(08) dairy cattle/others, (07) dairy cattle/land independent, (06)
dairy cattle/temporary grass, (05) dairy cattle/permanent grassland,
(04) arable/other, (03) arable specialized crops, (02) arable/fallow,
and (01) arable/cereal. The set of constraints, used for the base year
(2003) to simulate arable farm types consists of the resource constraints (available land, available irrigated land and labour) and policies (sugar beet quota regime and the obligatory set-aside). For
dairy farms the constraints relate to the feed availability, the maximum amount of concentrates in animals’ diet, and the grazing period
were added. The data requirements for the base year simulations include the available farm resources (i.e. available farm land characterized by soil and climatic conditions, available irrigated land and
available family labour) the inputs and outputs of current activities,
the observed cropping patterns, the herd composition (Tables 1 and
2), the economic data (i.e. variable costs of inputs, output prices and
wages) and the policy data (i.e. compensation payments under
Agenda 2000, quotas for sugar beet and milk production).
The calibrated model is used to predict changes in total gross
margin, agricultural supply and environmental indicators as a consequence of the implementation of the 2003 CAP reform in a context of market liberalization. The time horizon of the simulation is
the year 2013 and takes into account (i) new exogenous prices generated by the CAPRI agricultural sector model (Britz et al., 2007)
under the market liberalization scenario, (ii) abolishment of the
existing quota for sugar beet and milk, (iii) abolishment of the
obligatory set-aside policy, and (iv) new subsidies calculated under
the CAP reform of 2003.
The data used for policy scenarios are yields and price trends for
year 2013 as calculated by CAPRI (Britz et al., 2007) in a market liberalization scenario. The market liberalization scenario in CAPRI
assumes abolishing the export tariffs. It should be noted that this
market liberalization scenario within CAPRI does not include abolishment of quota or of the obligatory set-aside policy. Input and
590
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
Table 1
Farm specific data of farm types in Flevoland in 2003, and, observed crop areas and animal numbers that are included in the current activities. Source: FADN.
Arable farms
Farm specific data
Total available land
Irrigated land
Family labour
Milk quota
Costs of hired labour
Farms represented
(ha)
(ha)
(h)
(tons)
(€/h)
(%)
Crop areas
Grass (permanent)
Maize (silage
Onions
Potatoes
Set-aside
Sugar beet
Wheat (soft)
Other crops (not simulated)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
Animals
Bulls
Calves
Cows
Heifers
Total
(heads)
(heads)
(heads)
(heads)
(heads)
Dairy farms
FT 2303
FT 3203
FT 3303
FT 3304
Average farm
FT 3205
FT 3305
FT 3307
FT 3308
Average farm
17.9
66.3
68.7
33.9
56.4
49.7
44.6
33.1
48.9
44.6
3156
2997
5403
7641
4754
7.5
13
7.5
29
7.5
44
7.5
15
7.5
3325
437
7.5
8
4293
555
7.5
78
4440
488
7.5
6
3933
571
7.5
8
4196
543
7.5
1.5
2.2
4.8
1.7
3.1
2.7
2
2
3.2
17.9
1.8
11.2
10.4
19.8
0.6
9.7
24.8
1.3
9.1
11.5
11.7
0.1
23.4
3.5
0.9
1.3
2
2.8
1
8.9
17.1
1.4
7.8
8.7
11.5
45
3.7
35.9
7.7
18.8
9.2
0.2
0.4
3
0.3
1.2
19
16.4
0.6
1
5.4
1.6
1.7
3.1
34.2
8.2
0.1
0.3
0.8
0.4
0.2
0.6
1
23
65
19
107
2
31
76
22
131
2
25
72
21
120
Table 2
Farm specific data of farm types in Midi-Pyrenees in 2003, and observed crop areas
and animal numbers that are included in the current activities. Source: FADN.
Arable farms
Farm specific data
Total available land
Irrigated land
Family labour
Milk quota
Cost of hired labour
Farms represented
Crop areas
Barley
Grass (permanent)
Maize (grain)
Maize (silage)
Peas
Rape seed
Set-aside
Soya
Sunflower
Wheat (durum)
Wheat (soft)
Other crops (not
simulated)
Total
Animals
Bulls
Calves
Cows
Heifers
Total
FT
3201
FT
3202
FT
3304
Average
farm
141.2
41.8
2902
123.8
30.4
3260
173.1
16.5
3179
148.7
30.9
3067
7.5
46
7.5
20
7.5
34
7.5
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
(ha)
4.1
1.6
2.4
3
35.1
0.3
3.7
1.7
9.3
3
14.3
17.3
13.1
39.3
25.1
0.5
3.6
1
18.9
3.6
12.6
11.4
12.3
33.2
3.6
0.7
6.4
1.6
9.4
7.8
33.9
31.6
13.2
62.4
22.3
0.5
4.6
1.5
11.2
4.8
20.7
21.1
13
46
(ha)
141.2
123.8
173.1
148.7
(ha)
(ha)
(h)
(tons)
(€/h)
(%)
(heads)
(heads)
(heads)
(heads)
(heads)
Dairy
farm
FT
2206
41.6
0.8
2152
171
7.5
100
2
28.4
0.1
6.3
0.3
0.5
0.9
3.3
41.7
1
10
29
7
47
output prices are inflated with a constant inflation rate (1.9%) in
both regions for a period of 10 years. Historical yields, subsidy levels (as those determined in the CAP reform of 2003) and regionspecific decoupling of subsidies from production, were also
considered.
Weighted average economic and policy data (prices, yields subsidies, costs and gross margins) for the base year and the 2003 CAP
0.6
0.2
0.2
0.3
0.1
0.3
3
24
58
17
102
2
25
74
21
122
0.1
0.3
reform for Flevoland and Midi-Pyrenees under the market liberalization scenario are presented in Tables 3 and 4, respectively. The
weights are determined from the share of activities observed in
each farm type and the share of farm types in the regions. Based
on this information, in Flevoland the expected gross margins of silage maize, onions and soft wheat are projected to increase in 2013
while the expected gross margins of potatoes and sugar beet are
projected to decrease substantially. With the CAP reform of 2003
silage maize receives a larger subsidy than other crops. The main
reason for this is that most of the silage maize area is at dairy farms
which receive a larger subsidy per ha because of the decoupled animal production. Grass products are assumed to be non-tradable
products and thus have no price in the model. The expected gross
margin decrease of grass is due to increasing costs because of inflation. The large decrease in the price of milk is associated with the
market liberalization scenario and it is the reason for the lower expected gross margin per herd unit. In Midi-Pyrenees, the expected
gross margins of most crops increase due to higher prices and subsidies. An exception is durum wheat for which the subsidy decreases by almost 67% resulting in a substantial decrease of
expected gross margin. Inflation of the costs is the main reason
for the lower expected gross margin of grass while the lower subsidy for set-aside is the main reason for lower expected gross margin of the fallow activity. Similar to Flevoland the average expected
gross margin of a herd unit is reduced due to projected lower milk
price.
Three model runs were designed to analyse the effects of the
different changes during the period 2003–2013 (see Table 5). In
the first model run (price–yield change) we included only price
and yield changes and inflated input prices for year 2013, assuming
market liberalization. In the second model run (set-aside & quota
abolishment) we added the abolishment of the obligatory set-aside
policy and the quota regimes for both sugar beet and milk. In the
third model run (CAP 2003) we added the CAP reform of 2003. In
this model run we recalculated subsidies according to the CAP reform of 2003 where decoupling of subsidies from production was
decided. Notice that only model run 3 can be considered as a complete policy scenario (all interrelated changes are taken into account simultaneously), the other model runs serve to analyse the
effects of the individual changes during period 2003–2013.
591
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
Table 3
Crop product and animal product prices, yields, subsidies, costs and gross margins in 2003 and 2013 in Flevoland. Source: FADN, SEAMLESS survey and CAPRI model.
Price
(€/tonne)
Crop products
Maize fodder
Onions
Potatoes
Set-aside
Soft wheat (spring)
Soft wheat (winter)
Sugar beet
a
Subsidy
(€/ha or €/head)
Costs
(€/ha or €/head)
Gross margin
(€/ha or €/head)
2003
2013
2003
2013
2003
2013
Change (%)
448
7
91
1098
2158
2252
100
527
524
1150
1329
2611
2725
121
638
634
1392
126
3098
1838
198
707
892
3763
567
4100
340
121
836
1082
2018
350
32
81
161
18
21
46
267
323
267
323
21
749a
906
720
633
12
2003
2013
2003
2013
30
90
100
34
109
74
40.8
58.4
40.9
42.9
61.4
40.5
120
130
75
142
154
48
7.8
8.6
65.5
8.7
9.6
70.6
6.0
4.0
6.6
4.4
0.0
0.0
0.2
7.5
0.0
0.0
0.2
8.9
Grassland
Grass dry matter (grazed)
Grass dry matter (silage)
Animal products
Bull (meat)
Calves (meat)
Cows (meat)
Cows (milk)
Herd unit
Yield
(tons/ha or tons/head)
700
108
650
320
695
143
645
275
298
298
298
31
234
234
59
Average costs before calibration, feed costs are not included.
Table 4
Crop product and animal product prices, yields, subsidies, costs and gross margins in 2003 and 2013 in Midi-Pyrenees. Source: FADN, SEAMLESS survey and CAPRI model.
Price
(€/tonne)
Crop products
Barley
Maize (grain)
Maize (silage)
Peas
Rape seed
Set-aside
Soya
Sunflower
Wheat (durum)
Wheat (soft)
a
Subsidy
(€/ha or €/head)
Costs
(€/ha or €/head)
Gross margin
(€/ha or €/head)
2003
2013
2003
2013
2003
2013
2003
2013
Change (%)
304
304
452
431
423
448
443
156
450
451
198
444
340
859
860
385
582
411
1039
1041
466
704
331
294
421
430
401
356
509
520
434
765
988
385
171
304
424
521
846
628
567
993
1657
526
451
156
1027
871
546
879
23
23
40
27
62
95
59
40
55
29
72
87
72
87
21
405a
490
1023
800
22
2003
2013
94
120
120
133
204
101
152
132
150
318
5.0
11.0
15.4
3.5
2.2
5.2
10.5
17.3
3.6
2.2
196
213
135
116
318
323
148
137
2.3
2.4
5.0
6.5
3.1
2.4
5.8
7.0
2.3
3.1
4.6
2.5
3.4
5.0
0.0
0.0
0.2
6.0
0.0
0.0
0.2
7.3
Grassland
Grass dry matter (grazed)
Grass dry matter (hay)
Grass dry matter (silage)
Animal products
Bull (meat)
Calves (meat)
Cows (meat)
Cows (milk)
Herd unit
Yield
(tons/ha or tons/head)
1200
110
600
320
1191
146
595
258
304
304
304
304
304
592
304
30
31
Average costs before calibration, feed costs are not included.
5. Results
In this section weighted average results of different farm types
in the two regions are presented; the weights are determined from
the relative share (based on number of farms represented) of the
farm types in the region, i.e. first, the average farm of each farm
type is simulated and then the results were used to calculate
weighted average values of arable and dairy farms in each region.
The regional average simulated crop levels, the regional average
economic results and the calculation of the regional average nitrogen use of arable farms in Flevoland are presented in Fig. 3. Because of the PMP calibration, the simulated crop levels for the
base year are exactly the same as the actual levels observed in
FADN (Table 1). In the price–yield change model run, the gross margin increase of maize silage, onions and wheat causes a substantial
increase in the areas of these crops in arable farming. The gross
margin decrease of potatoes and sugar beet causes a decrease in
the area of these crops. The decrease of the area of sugar beet is
also because of the yield trend (8% increase). Less area of sugar beet
are needed to produce the same quota. The average total gross
margin of arable farms decreases with more than 28%. The shift
of crop production from spring soft wheat to winter soft wheat is
the main reason for the increase of the total nitrogen use per ha
in all farm types of Flevoland.
In the set-aside & quota abolishment model run for arable farms,
the abolishment of the sugar beet quota system and the obligatory
set-aside policy, are the reasons for the increase of the area of sugar
beet and the decrease of the area of set-aside (compared to the
simulated levels of these activities in the price–yield change model
run). Putting the set-aside area in production causes an increase of
the average total gross margin of arable farms. The total nitrogen
use increased in all farm types because of the decrease of the area
592
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
Table 5
Definition of the base year and the model runs (price–yield change, set-aside & quota abolishment and CAP 2003).
Exogenous
assumptions
Base year (2003)
Price and yield
Set-aside and quota
policies
EU compensation
payment
2003 price and yield
With obligatory set-aside
and quota
Agenda 2000 (direct
payment)
Price–yield change
(2013)
Inflation rate of
1.9% per year
Projection in prices and yields from 2003 to 2013
accounting for market liberalization
With obligatory set-aside
and quota
Agenda 2000 (direct
payment)
Set-aside & quota
abolishment (2013)
Inflation rate of
1.9% per year
Projection in prices and yields from 2003 to 2013
accounting for market liberalization
Abolishing set-aside
obligation and quota
Agenda 2000 (direct
payment)
CAP 2003 (2013)
Inflation rate of
1.9% per year
Projection in prices and yields from 2003 to 2013
accounting for market liberalization
Abolishing set-aside obligation and quota
2003 CAP reform
(decoupled payment)
Fig. 3. Simulated results for the base year (2003) simulation and 3 model runs (2013) for an average arable farm of Flevoland (price–yield change: yield and price trend,
inflation of input prices, set-aside & quota abolishment: price–yield change + abolishment of obligatory set-aside policy and quotas, CAP 2003: price–yield change + set-aside &
quota abolishment + CAP reform of 2003).
of set-aside and the increase of the area of the more nitrogen
demanding winter soft wheat.
In the CAP 2003 model run, overall effects on crop allocation are
very modest compared to the set-aside & quota abolishment, model
run and the associated effect on the total gross margin negligible.
The regional average weighted results of the simulated dairy
farms in Flevoland are presented in Fig. 4. The produced feed reported in Fig. 4 corresponds to on-farm feed production that is
used on-farm excluding sold quantities of on-farm produced feed.
Similar to the base year simulation of the arable farms and because
of the PMP calibration, the observed activity levels of crops and
animals of the dairy farms are reproduced exactly. In the price–
yield change model run the total number of animals of the herd decreases because of the increased milk production per cow and the
given quota. The area of permanent grassland decreases but the
on-farm feed production of grass increases because of the assumed
yield increase. The amount of silage maize sold increases because
of the price increase. The share of grass in the diet increases and
as a result the amount of concentrates also increases to fulfil the
animals energy requirements while respecting their intake capacity. The gross production decreases mainly because of the decrease
in the price of milk. The total costs increase because of the higher
input and feed prices and the increased feed requirements. As a result, the total gross margin decreases by almost 35%. The total
nitrogen use remains almost the same in all dairy farm types of
Flevoland.
In the set-aside & quota abolishment model run where the milk
quota is abolished, the total number of animals increases by 1.7%
compared with the base year simulation and by almost 13% from
the price–yield change model run. The increased feed requirements
are covered by increasing purchases of concentrates and silage
maize. The total gross margin increases by almost 16% from the total gross margin of the price–yield change model run while the total
nitrogen use remained almost the same.
In the CAP 2003 model run, the CAP reform of 2003 and mainly
the large increase of the subsidy for maize silage causes a shift of
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
593
Fig. 4. Simulated results for the base year (2003) simulation and 3 model runs (2013) for an average dairy farm of Flevoland (price–yield change: yield and price trend,
inflation of input prices, set-aside & quota abolishment: price–yield change + abolishment of obligatory set-aside policy and quotas, CAP 2003: price–yield change + set-aside &
quota abolishment + CAP reform of 2003).
production from grass to maize silage. The received premiums under the CAP reform of 2003 for dairy farms increase substantially,
causing a modest increase of the farm’s total gross margin.
The regional average weighted results from the application of
FSSIM to the arable farms of Midi-Pyrenees are presented in
Fig. 5. In the price–yield change model run, the predicted changes
for 2013 of gross margins resulted in an increase of the areas of
soya, rape seed and silage maize and a decrease in the areas of barley and peas. The average total gross margin of arable farms increases by 24%. The main reason for this is the large price
increase of oil seed crops.
In the set-aside & quota abolishment model run the set-aside
obligation of arable farms is abolished putting almost 70% of the
set-aside area of the price–yield change model run into production.
The set-aside land is allocated to all other crops. The intensification
of production caused a large increase of the average total gross
margin of arable farms but also a substantial increase of the total
nitrogen use compared to the price–yield change model run.
The recalculation of subsidies according to the CAP reform of
2003 caused a large increase of the received subsidies for most
crops. Exceptions are the subsidies for durum wheat and set-aside
land, which decrease by 67 and 49%, respectively, causing a decrease of the average area of these activities. The total gross margin
decreased by 1.5% compared with the set-aside & quota abolishment, model run and increased by 24% and 54% compared with
the price–yield change model run and base year, respectively.
Regional average weighted results from the application of
FSSIM to the dairy farms of Midi-Pyrenees are presented in
Fig. 6. Similar to the dairy farm of Flevoland, the produced feed re-
ported in Fig. 6 corresponds to on-farm feed production that is
used on-farm, excluding sold quantities of on-farm produced feed.
In the price–yield change model run, the substantial increase of
feeding costs and input prices; and the price decrease of milk
caused a small decrease in the average herd size in Midi-Pyrenees.
The area of permanent and temporary grasslands decreases and it
is substituted mainly by silage maize and barley. On-farm produced grass and the more expensive purchased concentrates in
this model run are substituted by cereals and silage maize to cover
the animal’s feed requirements. The gross margin decreases by 6%.
Abolishment of the milk quota policy in the set-aside & quota
abolishment model run increases the average herd size back to
the level of the base year. Labour availability becomes a binding
constraint and therefore the number of animals does not exceed
the number of animals observed in the base year. The grassland
area increases compared to the price–yield change model run
substituting the area with silage maize which is not fed to the animals. In both Flevoland and Midi-Pyrenees it is expected that the
yield of milk in 2013 will have increased with 19% and 22%, respectively, causing an increase in the animals’ feed requirements. To
cover the additional requirements for feed, more grass silage and
purchased concentrates are needed. Abolishment of the milk quota
caused an increase to the farm’s total gross margin.
The effects of the CAP reform 2003 (tested in the CAP 2003 model run) relative to the results of the set-aside & quota abolishment
model run are marginal. The large increase of subsidies on maize
silage in the CAP 2003 model run caused an increase of the area
of silage maize and a decrease of the area of cereals (mainly barley)
and grassland compared to the set-aside & quota abolishment model
594
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
Fig. 5. Simulated results for the base year (2003) simulation and 3 model runs (2013) for an average arable farm of Midi-Pyrenees (price–yield change: yield and price trend,
inflation of input prices, set-aside & quota abolishment: price–yield change + abolishment of obligatory set-aside policy and quotas, CAP 2003: price–yield change + set-aside &
quota abolishment + CAP reform of 2003).
run. The decreased amount of barley fed to the animals is compensated by purchased concentrates. A small amount of hired labour is
needed to cover the additional labour requirements of maize.
Water and nitrogen use remain almost the same in all model runs.
6. Discussion and conclusion
In this article, a bio-economic farm model has been presented
that is modular and can be used to simulate the responses of farms
to agricultural and environmental policies in a broad range of contexts that may occur in the EU27. This was achieved by: (i) separating model and data and creating a consistent European database
for farm types, their locations and production activities, (ii) designing the model in a modular way, that allows switching on and off
modules, constraints or calibration methods, (iii) providing adequate documentation, and (iv) ensuring public availability. The
arable and dairy farms of two regions that differ substantially from
a bio-physical and socio-economic point of view were simulated
successfully, using information mainly available in a large EU-wide
database (i.e. FADN) and a relatively simple survey conducted
within SEAMLESS for a sample of regions representative for the
EU27. The PMP based calibration of FSSIM does not require additional region-specific knowledge and detailed information on specific constraints to guarantee exact calibration. Nevertheless,
availability of this kind of information could be easily exploited
and used to improve the forecasting performance of the model.
The market of land, possibilities for off-farm labour and structural changes are usually issues exogenous to the system definition
of bio-economic farm models (Janssen and Van Ittersum, 2007).
This is how these issues were also treated in the model presented
in this article, but we have indicated in Section 2 and 3 how they
can be partially dealt with, using FSSIM and a scenario approach.
To simulate farm structural change and land markets more comprehensively FSSIM needs to be combined with other models that
account for market and sector level changes, as has been attempted
in the SEAMLESS modelling framework (Pérez Domínguez et al.,
2009; Zimmermann et al., 2009).
In the present article we illustrate the standalone value of
FSSIM using applications in two regions and different farming systems. The applications raise a number of discussion points because
of a number of decisions concerning the set-up of the model. First,
the presented applications were based on data available in the
Farm Accountancy Data Network and a simple survey on agricultural management. This led to a restricted set of environmental
indicators, i.e. the total amount of water used for irrigation, and
the total amount of nitrogen used. This hinders a comprehensive
overview of the environmental implications of the market liberalization under the CAP reform of 2003. The use of a bio-physical
model to calculate technical coefficients that can easily be
exploited in FSSIM would increase the number of environmental
indicators and thus improve the overall assessment of the environmental impact of the tested scenario. However, this requires detailed agro-management data (timing and precise quantities of
inputs per crop) that are not available in pan-European data-sets.
Second, we used an average farm type in our simulations to ensure that all important crop products that are produced by farms of
a specific farm type will be part of the simulated production plan.
This is very important for the type of analysis that requires full representation of agricultural production to determine equilibrium
between supply and demand, such as in SEAMLESS (Van Ittersum
et al., 2008). However, simulating the average farm has also important drawbacks. An average farm and an average farmer do not
actually exist and consequently, an average activity pattern also
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
595
Fig. 6. Simulated results for the base year (2003) simulation and 3 model runs (2013) for an average dairy farm of Midi-Pyrenees (price–yield change: yield and price trend,
inflation of input prices, set-aside & quota abolishment: price–yield change + abolishment of obligatory set-aside policy and quotas, CAP 2003: price–yield change + set-aside &
quota abolishment + CAP reform of 2003).
does not exist. The activity pattern of the average farm is much
more diversified than that of individual farms. Reproducing such
a cropping pattern using an LP model would require a large number of binding constraints. It is possible that such constraints do
not even exist in reality and consequently they are difficult to define (e.g. rotational constraints of an ‘‘average” production plan). In
such cases, calibration of the LP model is necessary for reproducing
the observed activity levels and often calibration will dominate the
simulations. It is possible that the impact of calibration on the results of the model would be reduced substantially if a number of
individual farms were simulated instead of a single average farm.
However, this would also have increased the computational
requirements and individual farm data would have to be available
which is usually not the case (individual farm data are usually confidential and not available for research).
Finally, we assume a yield trend (based on forecasts of the sector model CAPRI) to represent technological innovation. However,
the rapid changes in the socio-economic and the bio-physical environment might lead to a broader variety of alternative activities
that will become available to farmers in the future with even completely different inputs and outputs. Such alternative activities
cannot be ignored and should be taken into account in ex-ante
evaluation of agricultural and environmental policies. Offering
alternative activities in FSSIM is possible from a technical point
of view. The difficulty is to identify a consistent and feasible set
of alternative activities for all regions across the EU.
Apart from using our intuition to assess the model’s forecasting
performance, it is very difficult to evaluate the results in a quanti-
tative and more objective way because they refer to future events
and they use simulated data to account for price and yield trends.
The quality of the results of FSSIM has been previously evaluated
and assessed in ex-post experiments that demonstrate the capacity
of the model to simulate the future behaviour of the farmer (Kanellopoulos et al., 2010). Even though, the results of such ex-post
exercises cannot be generalized they do increase the confidence
in the model’s predictions.
A well calibrated and tested bio-economic farm model can be
used for ex-ante assessment of the impacts of new policies. Different farming systems across EU can be affected in different ways
and consequently farmers respond differently when they are confronted with market and policy changes. This was confirmed by the
results presented in this article. For example, price and yield
changes are the main factor explaining the gross margin decrease
of farms in Flevoland. In Midi-Pyrenees, simulated price and yield
changes have the opposite effect on the total gross margin of arable
farms, and for this region the abolishment of obligatory set-aside
has an additional positive effect on the total gross margin of arable
farms. In Flevoland farms showed an increase in premiums under
the CAP 2003 reform scenario, whereas in Midi-Pyrenees the CAP
2003 scenario did not further increase the already high level of premiums. The variation in farm’s behaviour should be taken into account for efficient and effective policy assessment. Bio-economic
modelling can be a useful tool for exploring this variation.
FSSIM has been set-up such that it can readily simulate farm
types in very different contexts (climate, soils and socio-economic
conditions) and for different purposes. The presented examples in
596
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
this paper show a fairly detailed analysis for the farm types of two
regions. The reusability of the model was confirmed by the significant number of applications that have been published (Louhichi
et al., 2008, 2009; Kanellopoulos et al., 2009; Majewski et al.,
2009; Mouratiadou et al., 2010; Traoré et al., 2009). Pérez Domínguez et al. (2009) show how results of the model can be used for
linking micro and macro level analysis of market changes. The
model is available under an Open Source license (www.seamlessassociation.org) and through its broader use it can be further tested
and new modules can be added.
Acknowledgements
The work presented in this publication was funded by the
SEAMLESS Integrated Project, EU 6th Framework Programme for
Research Technological Development and Demonstration, Priority
1.1.6.3 Global Change and Ecosystems (European Commission,
DG Research, Contract No. 010036-2).
References
Andersen, E., Elbersen, B., Godeschalk, F., Verhoog, D., 2007. Farm management
indicators and farm typologies as a basis for assessment in a changing policy
environment. Journal of Environmental Management 82, 352–362.
Berentsen, P.B.M., 2003. Effects of animal productivity on the costs of complying
with environmental legislation in Dutch dairy farming. Livestock Production
Science 84, 183–194.
Borkowski, N., Hecker, J.-M., Louhichi, K., Blanco Fonseca, M., Janssen, S., Van
Ittersum, M.K., Stokstad, G., Zander, P., 2007. Surveying crop management data
for bio-economic farm models. In: Hatfield, J., Donatelli, M., Rizzoli, A. (Eds.),
Farming Systems Design 2007: An International Symposium on Methodologies
for Integrated Analysis of Farm Production Systems, Catania, Sicily, Italy, vol. 1,
pp. 33–34.
Britz, W., Pérez, I., Zimmermann, A., Heckelei, T., 2007. Definition of the CAPRI Core
Modelling System and Interfaces with other Components of SEAMLESS-IF.
SEAMLESS Report No. 26, SEAMLESS Integrated Project, EU 6th Framework
Programme, Contract No. 010036-2, p. 114. <www.seamlessassociation.org>.
CEC, 1991. The Development and Future of the Common Agricultural Policy. COM
(91) 258 Final.
CEC, 1999a. Agricultural Council: Political Agreement on CAP Reform. DG
Agriculture Newsletter, March 11, 1999, Brussels (special edition).
CEC, 1999b. Berlin European Council: Agenda 2000, Conclusions of the Presidency.
DG Agriculture Newsletter, April 1999, Brussels.
CEC, 2003. Reform of the Common Agricultural Policy: Medium-term Prospects for
Agricultural Markets and Income in the European Union, December, Brussels.
<http://www.europa.eu.int/comm/agriculture/publi/caprep/prospects2003b/
fullrep.pdf>.
Donaldson, A.B., Flichman, G., Webster, J.P.G., 1995. Integrating agronomic and
economic models for policy analysis at the farm level: the impact of CAP reform
in two European regions. Agricultural Systems 48, 163–178.
EC, 2005. Impact Assessment Guidelines. European Commission, Brussels. p. 99.
EEC, 1985. Commission Decision of the 7th of June 1985, Establishing a Community
Typology for Agricultural Holdings (OJ L 220, 17.8.1985, p. 1).
Elbersen, B., Kempen, M., Van Diepen, K., Hazeu, G., Verhoog, D., Andersen, E., 2006.
Protocols for Spatial Allocation of Farm Types. SEAMLESS Report No. 19,
SEAMLESS Integrated Project, EU 6th Framework Programme, Contract No.
010036, p. 99. <www.seamlessassociation.org>.
Flichman, G., 1996. L’evaluation micro-economique de la reforme de la PAC par un
modele bio-economique: une illustration sur la region de Toulouse. In:
Reyniers, F.N., Benoit-Cattin, M. (Eds.), Linking-Agricultural-Models: SeminarProceedings, Montpellier, France. CIRAD-CA, IAM, Montpellier, France, pp. 71–
75.
Flichman, G., Jacquet, F., 2003. Couplage des modèles agronomiques et
économiques: intérêt pour l’analyse des politiques. Cahiers d’économie et
Sociologie Rurales 67, 1–20.
Hazell, P., Norton, R., 1986. Mathematical Programming for Economic Analysis in
Agriculture. Macmillan, New York.
Hazeu, G., Elbersen, B., Andersen, E., Baruth, B., Van Diepen, K., Metzger, M., 2010. A
bio-physical typology for a spatially-explicit agri-environmental modeling
framework. In: Brouwer, F., Van Ittersum, M.K. (Eds.), Environmental and
Agricultural Modelling: Integrated Approaches for Policy Impact Assessment.
Springer Academic Publishing, Dordrecht.
Heckelei, T., 2003. Positive Mathematical Programming: Review of the Standard
Approach, CAPRI Working Papers 97-03. <http://www.agp.uni-bonn.de/agpo/
rsrch/capri/wrkpap_e.htm>.
Heckelei, T., Wolff, H., 2003. Estimation of constrained optimisation models for
agricultural supply analysis based on generalized maximum entropy. European
Review of Agricultural Economics 30, 27–50.
Howitt, E.R., 1995. A calibration method for agricultural economic production
models. Journal of Agricultural Economics 46, 147–159.
Janssen, S., Van Ittersum, M.K., 2007. Assessing farm innovations and responses to
policies: a review of bio-economic farm models. Agricultural Systems 94, 622–
636.
Janssen, S., Andersen, E., Athanasiadis, I., Van Ittersum, M., 2009a. A database for
integrated assessment of European agricultural systems. Environmental Science
& Policy 12, 573–587.
Janssen, S., Van Ittersum, M.K., Meuter, E., Louhichi, K., Flichman, G., Blanco, M.,
Zander, P., Borkowski, N., Hecker, M., Stokstad, G., Romstad, E., Kanellopoulos,
A., Berentsen, P., Hengsdijk, H., Van Keulen, H., Thorne, P., Thornton, P., Li, H.,
Rizzoli, A.E., Heckelei, T., 2009b. Integration of All FSSIM Components within
SEAMLESS-IF and a Standalone Graphical User Interface for FSSIM. SEAMLESS
Report No. 38, SEAMLESS Integrated Project, EU 6th Framework Programme,
Contract No. 010036-2. <www.seamlessassociation.org>.
Jansson, T., 2007. Econometric Specification of Constrained Optimization Models.
PhD dissertation. Bonn University. <http://www.ilr1.uni-bonn.de/agpo/rsrch/
capri/wrkpape.htm> (electronic publication).
Jarrige, R., 1989. Ruminant Nutrition. Recommended Allowances and Feeding
Tables. INRA, Paris, France.
Judez, L., Chaya, C., Martinez, S., Gonzalez, A.A., 2001. Effects of the measures
envisaged in ‘‘Agenda 2000” on arable crop producers and beef and veal
producers: an application of Positive Mathematical Programming to
representative farms of a Spanish region. Agricultural Systems 67, 121–138.
Kanellopoulos, A., Hengsdijk, H., Janssen, S., Van Ittersum, M.K., 2009. Ex-ante
assessment of the abolishment of the EU set aside policy: results from a bioeconomic farm analysis. In: Proceedings of the Conference on Integrated
Assessment of Agriculture and Sustainable Development Setting the Agenda of
Science and Policy (AgSAP 2009). Egmond aan Zee, The Netherlands, 10–12
March 2009. Wageningen University and Research Centre, Wageningen, pp.
478–479.
Kanellopoulos, A., Berentsen, P.B.M., Heckelei, T., Van Ittersum, M.K., Oude Lansink,
A.G.J.M., 2010. Assessing the forecasting performance of a generic bio-economic
farm model calibrated with two different PMP variants. Journal of Agricultural
Economics 61, 274–294.
Louhichi, K., Belhouchette, H., Flichman, G., Therond, O., Wery, J., 2008. Impact
assessment of the 2003 CAP reform and the nitrate directive on the arable
farming system in the Midi-Pyrénées region: bio-economic modelling at field,
farm and regional levels. In: 109th Seminar, November 20–21, 2008, Viterbo,
Italy.
Louhichi, K., Hengsdijk, H., Janssen, S., Bigot, G., Perret, E., 2009. EU dairy farming in
the face of change: an exploration using a bio-economic farm model. In:
Proceedings of the Conference on Integrated Assessment of Agriculture and
Sustainable Development Setting the Agenda of Science and Policy (AgSAP).
Egmond aan Zee, The Netherlands, 10–12 March 2009. Wageningen University
and Research Centre, Wageningen, pp. 262–263.
Majewski, E., Was, A., Belhouchette, H., Louhichi, K., Mouratiadou, I., 2009. Impact
Assessment of policy changes on the arable sector using the FSSIM model: case
study of the Zachodniopomorskie NUTS region. In: Proceedings of the
Conference on Integrated Assessment of Agriculture and Sustainable
Development Setting the Agenda of Science and Policy (AgSAP). Egmond aan
Zee, The Netherlands, 10–12 March 2009. Wageningen University and Research
Centre, Wageningen, pp. 484–485.
Mouratiadou, I., Russell, G., Topp, C., Louhichi, K., Moran, D., 2010. Modelling
common agricultural policy–water framework directive interactions and costeffectiveness of measures to reduce nitrogen pollution. Water Science &
Technology 61, 2689–2697.
OECD, 2004. Analysis of the 2003 CAP Reform. OECD Publication. 53p.
Onate, J.J., Atance, I., Bardaji, I., Llusia, D., 2006. Modelling the effects of alternative
CAP policies for the Spanish high-nature value cereal-steppe farming systems.
Agricultural Systems 94, 247–260.
Pérez Domínguez, I., Bezlepkina, I., Heckelei, T., Romstad, E., Oude Lansink, A.,
Kanellopoulos, A., 2009. Capturing market impacts of farm level policies: a
statistical extrapolation approach using biophysical characteristics and farm
resources. Environmental Science & Policy 12, 588–600.
Riesgo, L., Gomez-Limon, J.A., 2006. Multi-criteria policy scenario analysis for public
regulation of irrigated agriculture. Agricultural Systems 91, 1–28.
Röhm, O., Dabbert, S., 2003. Integrating agri-environmental programs into regional
production models: an extension of positive mathematical programming.
American Journal of Agricultural Economics 85, 254–265.
Semaan, J., Flichman, G., Scardigno, A., Steduto, P., 2007. Analysis of nitrate pollution
control policies in the irrigated agriculture of Apulia Region (Southern Italy): a
bio-economic modelling approach. Agricultural Systems 94, 357–367.
Thorne, P.J., Hengsdijk, H., Janssen, S., Louhichi, K., Van Keulen, H., Thornton, P.K.,
2009. Modelling Livestock Component in FSSIM, SEAMLESS Report No. 35,
SEAMLESS Integrated Project, EU 6th Framework Programme, Contract No.
010036-2, p. 73. <www.seamlessassociation.org>.
Topp, C.F.E., Mitchell, M., 2003. Forecasting the environmental and socio-economic
consequences of changes in the common agricultural policy. Agricultural
Systems 76, 227–252.
Traoré, B., Louhichi, K., Rapidel, B., 2009. A bio-economic model to analyse the
performance of the cotton sector in Mali. In: Proceedings of the Conference on
Integrated Assessment of Agriculture and Sustainable Development Setting the
Agenda of Science and Policy (AgSAP). Egmond aan Zee, The Netherlands, 10–12
March 2009. Wageningen University and Research Centre, Wageningen, pp.
276–277.
Van Ittersum, M.K., Ewert, F., Heckelei, T., Wery, J., Alkan Olsson, J., Andersen, E.,
Bezlepkina, I., Brouwer, F., Donatelli, M., Flichman, G., Olsson, L., Rizzoli, A.E.,
K. Louhichi et al. / Agricultural Systems 103 (2010) 585–597
van der Wal, Wien, J.E., Wolf, J., 2010. Integrated assessment of agricultural
systems – a component-based framework for the European Union (SEAMLESS).
Agricultural Systems 96, 150–165.
Veysset, P., Bebin, D., Lherm, M., 2005. Adaptation to Agenda 2000 (CAP
reform) and optimisation of the farming system of French suckler cattle
farms in the Charolais area: a model-based study. Agricultural Systems 83,
179–202.
597
Zander, P., Borkowski, N., Hecker, J.-M., Uthes, S., Stokstad, G., Rørstad, P.K.,
Bellocchi, G., 2010. Procedure to Identify and Assess Current Activities,
SEAMLESS Report No. 50, SEAMLESS Integrated Project, EU 6th Framework
Programme, Contract No. 010036-2. <www.seamlessassociation.org>.
Zimmermann, A., Heckelei, T., Pèrez Dominguez, I., 2009. Modelling farm structural
change for integrated ex-ante assessment: review of methods and
determinants. Environmental Science & Policy 12, 601–618.