Download European Simulation Model (ESIM)

European Simulation Model (ESIM):
Documentation of the Model Code
(Draft version, 30. January 2010)
Martin Banse,1 Harald Grethe,2 Andre Deppermann,2 Stephan Nolte3
Hohenheim, Brunswick, 2010
1
vTI, Brunswick.
2
University of Hohenheim.
3
University of Gent.
Content
List of Tables............................................................................................................................ 3
1
Introduction ...................................................................................................................... 4
1.1
Model Documentation................................................................................................ 4
1.2
Model Type ................................................................................................................ 4
1.3
Model History ............................................................................................................ 7
1.4
File Structure .............................................................................................................. 8
1.5
Installation................................................................................................................ 13
2
Description of Core Model ............................................................................................ 14
2.1
Overview .................................................................................................................. 14
2.2
Price Mechanism ...................................................................................................... 21
2.2.1
Basic Challenge ................................................................................................ 21
2.2.2
Approach Chosen in ESIM (This Section refers to a Former Base Period; The
Basic Mechanisms have not Changed since, but the Actual Price and Policy
Data) ................................................................................................................. 22
2.2.3
Determination of the Upper and Lower Bounds of the Price Transmission
Functions .......................................................................................................... 26
2.2.4
Changes in TRQ and Export Subsidy Levels ................................................... 27
2.2.5
The Modelling of Delayed Integration of Selected Regional Agricultural
Markets into the Single Market ........................................................................ 27
2.3
The Land Market...................................................................................................... 28
2.4
Feed Model............................................................................................................... 35
2.5
Processing Models.................................................................................................... 37
2.5.1
Oilseed Processing............................................................................................ 37
2.6
The Bioenergy Market ............................................................................................. 37
2.6.1
Overview .......................................................................................................... 37
2.6.2
Supply of Biofuel Inputs................................................................................... 38
2.6.3
Production of Biofuels and Biofuel Byproducts............................................... 39
2.6.4
Demand for Biofuels ........................................................................................ 39
2.6.5
Biofuel Policies................................................................................................. 39
2.6.6
Data................................................................................................................... 40
2.7
The Sugar Market..................................................................................................... 40
2.7.1
Introduction ...................................................................................................... 40
2.7.2
Individual Sugar Supply Curves for EU-27 Member States ............................ 41
2.7.2.1 Shape of the Sugar Beet Supply Function ........................................................ 42
2.7.2.2 Level of the Sugar Beet Supply Function......................................................... 44
2.7.2.2.1 Shadow Prices for Sugar ........................................................................ 44
2.7.2.2.2 The Processing Margin between Sugar Beet and White Sugar.............. 45
2.7.3
The Export Supply Function for Preferential EU Sugar Imports ..................... 47
2.7.3.1 Introduction ...................................................................................................... 47
2.7.3.2 Choice of Explaining Variable ......................................................................... 48
2.7.3.3 Estimating Price Responsiveness of Preferential Suppliers to the EU Market 49
2
2.7.3.3.1 ACP Countries and India........................................................................ 49
2.7.3.3.2 LDCs ...................................................................................................... 49
2.7.3.3.3 CXL, Balkan and SPS ............................................................................ 51
2.7.3.4 Estimation of the Annual Aggregate Preferential Export Supply Curve.......... 51
2.7.4
Implementation of Changes in the Model Code ............................................... 52
2.7.4.1 Individual Sugar Supply Curves for EU-27 member states ............................. 52
2.7.4.2 Price Responsive Preferential Export Supply Function for Sugar ................... 53
2.8
Stochastic elements in the model analysis ............................................................... 54
2.8.1
Overview .......................................................................................................... 54
2.8.2
Methodology..................................................................................................... 54
2.8.3
Grouping of Countries ...................................................................................... 55
2.8.4
Generation of Stochastic Terms θ..................................................................... 55
2.8.5
Gaussian Quadrature and Numerical Integration ............................................. 56
2.8.6
Implementation of Quadrature Points in ESIM................................................ 56
3
Behavioural Parameters ................................................................................................ 57
4
Base Data for Model Calibration.................................................................................. 57
4.1
Quantities ................................................................................................................. 57
4.2
Prices and Policies.................................................................................................... 57
5
References ....................................................................................................................... 57
List of Tables
Table 1.1: Product Coverage and Activities in ESIM................................................................ 6
Table 1.2: Overview of ESIM File Structure ............................................................................. 9
Table 2.2: Land Rental Prices in EU Member States in the Base Period (in Euro)................. 33
Table 2.3: Total Land Demand, Potentially Available Land, and Annual Change in Area..... 34
Table 2.4: Shadow Prices, Intercepts and Supply Elasticities for Sugar Supply Functions in
the EU-15 in € per ton (2002 – 2005) .............................................................................. 44
Table 2.5: Processing Margins for the EU-15, €/ton White Sugar (2003)............................... 46
Table 2.6: Variables of LDC sugar sectors .............................................................................. 50
List of Figures
Figure 2.1: Price Transmission for Wheat (EU-15) ................................................................. 23
Figure 2.2: Price Transmission for Beef with ES..................................................................... 24
Figure 2.3: Price Transmission for Beef with ES and TRQs ................................................... 26
Figure 2.4: Land Supply Curve Determining Land Conversion and Land Prices ................... 29
Figure 2.5: Schematic Overview of the Feed Model ............................................................... 36
Figure 2.6: Cost Data and Estimated Sugar Beet Supply Function for France........................ 43
Figure 2.7: Sugar and Beet Supply Curves for France............................................................. 47
Figure 2.8 Shape of Isoelastic Supply Curves with Different Elasticities ............................... 51
3
1
Introduction
1.1
Model Documentation
This technical model documentation describes the structure of ESIM as well as base data and
parameters used in the current version. A focus is on extensions of the model which were part
of the EU project Improving the ESIM Model for Ex-Post and Ex-Ante Market Analysis
(depiction of bioenergy, sugar, individual member states, delayed market integration for
accession countries and the inclusion of a historical database). The model documentation does
not include a description of the GAMS software, which can be found elsewhere (Brook et al.,
1998). ESIM in its current version is designed such that it can be run in GAMS IDE as well as
in the user interface GSE (Gams Simulation Environment). The usual approach for this
version would be to use GAMS IDE or GTREE for any model development activities like
adjustments of the model structure or updates of base data and behavioural parameters and to
use GSE for the formulation of policy scenarios. The use of ESIM in GSE including the
policy scenario design are described in an earlier version of a complementary user handbook
(Banse, Grethe, and Nolte, 2005). In addition, a recent user manual provides examples for the
application of the current ESIM version (Banse, Grethe and Nolte, 2007). The full
functionality of the GSE software is described in Dol and Bouma (2003).
1.2
Model Type
ESIM is a partial equilibrium multi-country model of agricultural production, consumption of
agricultural products, and some first-stage processing activities. It can be used in a
comparative static as well as a recursive dynamic version. ESIM is programmed in GAMS.
ESIM is a partial model, as only a part of the economy, the agricultural sector, is modeled.,
i.e. macroeconomic variables such as income or exchange rates are exogenous. As a world
model it includes all countries, though in greatly varying degrees of disaggregation. Some
countries are explicitly modelled and others are combined in an aggregate: the so-called rest
of the world (ROW). In its current version ESIM includes 25 EU Members (the Czech
Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia, Slovenia; Malta, Cyprus,
Turkey, and the individual EU-15 member countries, with Belgium and Luxembourg being
summarized as one region), four accession candidates (Bulgaria and Romania, which where
still non-members in the model base period, Turkey and Croatia) and the US as well as the
Western-Balkans. All other countries are aggregated as the ROW. As ESIM is mainly
designed to simulate the development of agricultural markets in the EU and accession
candidate countries policies are only modeled for these countries. I.e. for the US and the
ROW production and consumption takes place at world market prices. Trade is modeled as
net trade for all countries, except for sugar, for which bilateral trade is modeled.
In the comparative static version, adjustments in time are not explicitly covered. There are, for
example, no lagged price responses or price expectations at the supply side. Therefore, all
4
simulation results have to be interpreted as long term equilibrium states. Nonetheless, ESIM
is a projection model as shifters at the supply as well as the demand side (e.g. productivity or
income growth) are accounted for. Projections are made for a period of 15 years (2006-2020)
after the base period. But all of these projections are independent comparative static
equilibria. Alternatively, ESIM can be applied in a recursive dynamic specification with a
lagged price response at the supply side.
ESIM depicts a high variety of policy instruments like specific and ad valorem tariffs, tariff
rate quotas, intervention and threshold prices, export subsidies, product subsidies, direct
payments for keeping land in agricultural use, production quotas and voluntary as well as
obligatory set aside (as obligatory set aside still exists in the model base period). Table 1.1
shows the product coverage and activities modeled in ESIM.
5
Table 1.1: Product Coverage and Activities in ESIM
Product
Crops
Common wheat
Durum wheat
Barley
Corn
Rye
Other grains
Rice
Sugar
Potatoes
Sunflower seed
Soybeans
Rapeseed
Manioc
Fodder
Silage maize
Animal Products
Raw milk
Sheep meat
Beef
Pork
Poultry
Eggs
Processed products
Sunflower oil
Sunflower cake
Soy oil
Soy cake
Rape oil
Rape cake
Palm oil
Cheese
Skim milk powder
Butter
Fat
Protein
Cream
Whole milk powder
Concentrated milk
Acidified milk
Whey
Other dairy products
Milk
Biodiesel
Bioethanol
Other products
Pasture
Voluntary set aside
Other energy
Other protein
Gluten feed
Farm
supply
Processing
supply
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Human
demand
Seed
demand
Feed
demand
Processing
demand
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Source: own compilation.
6
For all 15 crops, 6 animal products, pasture, voluntary set aside and palm oil supply is defined
at farm level. Although, not each product is produced in each country: For example manioc is
only produced in the aggregate rest of the world.
Sugar, which of course is a processed product, is classified as "crop" in ESIM as the
processing activity is not modeled explicitly, i.e. as for other farm products, the area allocated
to the production of sugar is defined as well as the yield per ha. The processing activity is
implicitly covered by the high margin between the farmgate price and the wholesale price.4
Human demand is defined for each of the farm products except rapeseed, fodder, pasture, set
aside, silage maize and raw milk. Some of these products only enter the processing industry,
e.g. rapeseed, and others are only used in feed consumption like fodder or grass from
permanent pasture.
Seed demand is defined for those products for which seed accounts for a significant share of
production.
Processing demand is defined for raw milk, its components fat and protein, oilseeds and
inputs for biofuel production. In the case of sunflower seed, soybeans, and biofuel inputs,
processing demand adds to human and feed demand at the unprocessed level. For the other
processing inputs, processing demand is the only demand component except, in some cases,
seed demand. Feed demand is defined for various farm products as well as for four processing
outputs (three oil cakes and gluten feed) and some products from the processing industry such
as other energy and other protein, for which supply functions are specified. Processing
demand for biofuel inputs is specified for three plant oils for biodiesel production and wheat,
corn and sugar for bioethanol production. Palm oil is the only plant oil which does not
originate from oilseed crushing, but is depicted directly by a supply function in the rest of the
world, as it is not at all produced in the EU.
1.3
Model History
ESIM was initially developed by the Economic Research Service of the USDA and Josling
and Tangermann and programmed in SuperCalc. It was first used in Tangermann and Josling
(1994). The further development of ESIM has taken separate paths. In the ERS ESIM was
further developed mainly with the goal of market forecasts and policy analysis and
programmed in EXCEL (Lillard et al., 1995). In Europe ESIM was further developed in
Tangermann and Münch (1997) and Münch (1995, 2000, 2002). Main developments were an
extended country coverage and the combination with a CGE framework. Nölle (2000) has in
4
This approach is chosen because, in contrast to the oilseed industry, processing inputs (beet, cane) usually are
not traded, and the quantity of the raw product produced is therefore identical with the quantity processed. In
addition, the processing activitiy transforms a single product (beet) into a single output (sugar).
7
his master thesis validated ESIM for the period 1993 to 1997. Since 2001 ESIM is used and
further developed in DG AGRI. In 2005, ESIM was programmed in GAMS for the European
Commission (Banse, Grethe and Nolte, 2005) and several extensions to the original model
were developed such as a new price transmission mechanism (Banse and Grethe, 2006), the
depiction of subsistence production for milk in Central European Countries (Banse and
Grethe, 2005), a more refined depiction of the feed requirements of ruminants allowing for a
better depiction of decoupling effects (Balkhausen, Banse, Grethe and Nolte, 2005), a more
sophisticated sugar market (Grethe, Nolte and Banse, 2008) and a biofuel market module
(Banse and Grethe, 2008). An ex-post validation of the EU supply module in ESIM can be
found in Banse, Grethe and Nolte (2005).
1.4
File Structure
ESIM in GAMS including database and user-interface is organized in GAMS, EXCEL and
GSE files. GAMS files are organized in a tree-like structure which is displayed in Table 1.2.
The model is started from ESIM.GMS. At the beginning of ESIM.GMS it has to be defined
whether a new version or a scenario should be run by uncommenting the respective row. If a
version is RUN, the READING-DATA.GMS file is loaded, which organizes the reading of the
database. Otherwise, only a scenario is run based on a reading of the database at an earlier
stage. Before the database is read GSE is controlled by the file ENVIRONMENT.GMS and the
three files SETS.GMS, ALIAS.GMS and PARAMETERS.GMS which include basic definitions are
included.
In the READING-DATA.GMS file, the LOAD-PARA.GMS and LOAD-PRICE.GMS files are included.
LOAD-PARA.GMS organizes the reading of all behavioural parameters and feed rates from the
EXCEL file PARAMETER.XLS. LOAD-PRICE.GMS organizes the reading of price data from
EXCEL files PRICES.XLS as well as price margins and EU-policy data. Optional the user can
include the file LOAD-ALL.GMS in the READING-DATA.GMS file as well. Here
DATA_2003_VAL.XLS, DATA_2004_VAL.XLS, DATA_2005_VAL.XLS, and DATA-AREA.XLS are
read. Otherwise base data are taken from the gdx file DATA_ESIM_II.GDX. The DATA files
contain base quantities for production, human demand, seed demand, feed demand,
processing demand, net exports and base area. Averages of base years are calculated in the
LOAD-ALL.GMS file. After the inclusion of the LOAD-PARA.GMS and LOAD-PRICE.GMS files all
base data and parameters are stored in a GDX-file GSE_XLS.GDX.
8
Table 1.2: Overview of ESIM File Structure
File Names
ESIM.GMS
Content
 Deciding on scenario or version
ENVIRONMENT.GMS

Control of GSE
SETS.GMS

Definition of static and dynamic sets
ALIAS.GMS

Definition of aliases
MAPPING.GMS

Mapping of energy crops on setaside with conventional crops for energy use
PARAMETERS.GMS

Definition of parameters
READING-DATA.GMS

Writing all base data and parameters to the gdx file GSE_XLS.GDX
LOAD-PARA.GMS

Reading of behavioural parameters and feed rates from PARAMETER.XLS
LOAD-PRICE.GMS

Reading price data from PRICES.XLS
LOAD-ALL.GMS (OPTIONAL)


Reading base data from DATA_2003_VAL.XLS, DATA_2004_VAL.XLS,
DATA_2005_VAL.XLS and DATA-AREA.XLS
Calculation of averages
VARIABLES.GMS

Definition of variables
ASSIGNMENTS.GMS


Assignments of base values to variables
Assignments of some parameter variables
GSE-READING-DATA.GMS

Unloads parameter saved in the gdx file GSE_XLS.GDX
STOCHASTIC-DATA.GMS

Assignment of variances
MACRO-DATA.GMS

Contains values for macroeconomic parameters like exchange rate
CAP-2030.GMS

Contains values for various policy parameters in the base and over the
simulation period
LAND_DATA.GMS

Contains data for the calibration of land supply functions
DAIRY_ELAST.GMS

Contains data for the dairy market
9
Table 1.2: Overview of ESIM File Structure (continued)
Content
File Names
CALIBRATION.GMS

Definition of subsistence milk share
ADJUST-DAIRY.GMS



Reading data for the dairy modul
Checking the balance of the milk market
Calibration of dairy market data
CHECK-CONSIST.GMS


Calibration of world net exports to zero (adjusting supply in ROW)
Calibration of base feed rates
ASSIGN-DYNAMIC-SETS.GMS

Assignment of dynamic sets (which are defined in SETS.GMS)
CALIBRATION-PARA.GMS






Calculation of trade shares
Preliminary definition of upper and lower price bounds
Calculation of policies for the accession candidates
Definition of shadow prices
Assignment of finally valid domestic prices from LOGIT-CALIB.GMS
Calculation of per hectare premia under SFP and SAPS
LOGIT-CALIB.GMS

Generation of a set of finally valid domestic base prices
CALC-PARA.GMS





Reading elasticities for bio-fuel demand and supply
Calculation of seed parameters
Calculation of crushing rates
Calculation of crushing elasticities
Calculation of intercepts for all behavioural equations

Calibration of biofuel parameters
MAKEGSESHOW.GMS

Functionality only for ESIM in GSE: Defining and tagging input parameters
which cannot be edited and writing those to a gdx file SHOWONLY.GDX
FAPRI.GMS

Calibration of shifters in RoW to meet fapri world market price projections
CALIBRATION-DATA.GMS
CALIB-BIOFUEL.GMS
10
Table 1.2: Overview of ESIM File Structure (continued)
Content
File Names




Fixing of variables which are zero in the base at zero
Definition of equations
Solve for the base period
Solve over the complete simulation period
CALIB SETASIDE IN NMS.GMS

Calibration of area allocation of non-food crops for setaside land in NMS
CREATERESULTS.GMS

Writing of all solution values of variables and values of various parameters to
result parameters


Writing all results to a gdx file RESULTS.GDX
Tagging of all result parameters

Calculation of various results based on core results saved in
CREATERESULTS.GMS: aggregation, indices, budgetary implications…
MODEL.GMS
OUTPUT.GMS
RESULT-TRANSFORM.GMS
11
After all base data is read the files
VARIABLES.GMS, ASSIGNMENTS.GMS
and CALIBRATION.GMS are
included in ESIM.GMS. In VARIABLES.GMS all variables are defined. In ASSIGNMENTS.GMS start
values are assigned to variables and prices. Furthermore, the files GSE-READING-DATA.GMS,
STOCHASTIC-DATA.GMS,
MACRO-DATA.GMS,
CAP-2030.GMS,
LAND_DATA.GMS
and
DAIRY_DATA.GMS are included. GSE-READING-DATA.GMS unloads parameter saved in the gdx file
gse_xls.gdx. In STOCHASTIC-DATA.GMS variances are assigned.
MACRO-DATA.GMS
contains

base period exchange rates,

demand shifters population and income growth,

the supply shifter technological progress

growth rates for input prices, and

changes in the real exchange rate.
The file CAP-2030.GMS specifies base and future EU CAP parameters like accession dates, quotas,
tariffs, export subsidies, intervention prices, tariff rate quotas, direct payments and set aside rates.
LAND_DATA.GMS and DAIRY_DATA.GMS contain data on elasticities and prices regarding the land
and dairy market.
In
CALIBRATION.GMS
three files are included: CALIBRATION-DATA.GMS, ASSIGN-DYNAMICSETS.GMS, and CALIBRATION-PARA.GMS. In CALIBRATION-DATA.GMS, the file ADJUST-DAIRY.GMS is
included, which reads dairy parameters like the fat and protein content in delivered milk from the
GDX-file PARA_DAIRY.GDX and checks the balance of the milk market. Furthermore, the file
CHECK-CONSIST.GMS is included which adjusts supply in the country group rest of the world in
order to ensure world net exports to add up to zero and the base feed rates in order to ensure that
total base period feed demand equals the sum of animal supply multiplied with the respective feed
rates. Finally, in CALIBRATION-DATA.GMS, the base levels of subsistence milk for some Central
European countries are defined.
As a next step, the file
ASSIGN-DYNAMIC-SETS.GMS
is included in
CALIBRATION.GMS.
In this file
dynamic sets as declared in SETS.GMS are filled, which cannot be done before base quantity data is
finalized in the CALIBRATION-DATA.GMS. The next file included in CALIBRATION.GMS is
CALIBRATION-PARA.GMS in which trade shares are calculated which are used for the formulation of
the price transmission process. Upper and lower price bounds are defined for each product based
on observed policies for the EU-15. For the accession candidates, policies are calculated
endogenously based on the observed wedges between domestic and international prices.
Furthermore, shadow prices for quota products are defined in
CALIBRATION-PARA.GMS
relative to
the market price level in the base period. The effective producer price for production decisions is
12
set to be the minimum of shadow and market price. Thereafter two files are included:
CALIB.GMS and CALC-PARA.GMS.
LOGIT-
The file LOGIT-CALIB.GMS generates a set of finally valid domestic base prices. In this process it is
ensured, that final base prices deviate only minimally from original base prices (for details see
below, Section 4.3.1). The CALC-PARA.GMS file calculates seed and crushing coefficients as well as
intercepts of behavioural equations from base data. Furthermore the file CALIB-BIOFUELS.GMS is
included, which generates a set of finally valid domestic base prices in biofuel production.
Subsequently, the rate of technical progress and human demand shifters of certain products in the
rest of the world are calibrated in
FAPRI-MODEL.
In
MODEL.GMS,
FAPRI.GMS
to meet market price projections created with the
in order to save on computational capacity, all quantity and price variables which
are zero in the base (like rice production in Latvia) are fixed to zero. What follows are the
elements of the core model: the definition of equations, the model and the solve statement.
Equations are described in detail in Chapter 2 below. The first model solve is to regenerate the
base situation. If results differ from base data fed into the model the model is aborted
automatically and product and country specific differentials are displayed in order to check why
the model is not able to regenerate calibration data.
Afterwards the model is solved in a loop over all years between base period and 2020. At the end
of the MODEL.GMS file the files CALIB SETASIDE IN NMS.GMS and CREATERESULT.GMS are included.
The former implements the set aside requirements for the New Member States starting in 2011. In
the latter file all results of the respective simulation period are written to result parameters before
the loop over simulation periods is closed.
Finally, a file
OUTPUT.GMS
is included in
ESIM.GMS,
which first includes a file
RESULT-
TRANSFORM.GMS
and then writes all outputs to a GDX-file RESULTS.GDX. In RESULTTRANSFORM.GMS various supplementary results are calculated based on the primary results saved
in CREATERESULT.GMS. For example price, quantity and value indices, budgetary effects and stock
changes, and aggregations over various product and country groups.
1.5
Installation
ESIM comes in the form of model files (GAMS Version 22.4) and some data files in EXCEL. For
installation you should create a subdirectory DATA in the folder where you want to save the model
files (this name is indicative. With other names the model won’t run). Care: the directory path
should not include directory names with blanks. For installation extract all GAMS files to the main
directory and the XLS files to the subdirectory DATA. If you have done this, perform the following
steps to check whether installation was successful:
1. Open GAMS-IDE, create a new project "ESIM" in the main directory in which you keep
ESIM.
13
2. In that project, open the ESIM.GMS file.
3. Uncomment the row "*$setglobal GSEACTION VERSION" by deleting the star at the
beginning of the row. Run ESIM.GMS.
4. Now reinstall the star in row "*$setglobal GSEACTION VERSION" and uncomment the
row "*$setglobal GSEACTION SCENARIO" by deleting the star at the beginning of the
row. Run ESIM.GMS.
After each run the process window should show "normal completion". This run currently includes
the accession of Bulgaria and Romania in 2007, and the introduction of the 2003 Reform with SFP
in the EU-15, Slovenia and Malta and SAPS in the remaining eight NMS. To change the scenario
from "accession" to "no accession" the parameter SCENARIOS at the end of file PARAMETERS.GMS
has to be set to 0.
Alternatively, ESIM can be used from the GSE user interface. To do so perform the following
steps:
1. install the GSE software,
2. create a directory ESIM in NAQUIT\MODELS\,
3. copy an empty database (file EMPTY.MDB) to that directory and rename it to ESIM.MDB,
4. copy the ESIM.exe and the ESIM.gse to the Naquit directory,
5. copy the ESIM.bat and the LOAD.bat to the ESIM directory where you have the source
code,
6. open the GSE programme, select new version and run the programme.
For more details and an alternative to this "handmade" approach see the User Manual.
2
Description of Core Model
2.1
Overview
As a first overview of the core model, Table 2.1 provides an overview of all equations in ESIM.
14
Table 2.1: Overview of Equations in ESIMa
Supply equations
Supply (MCP equation 1, ten specifications)
(1) Supply of crops in European countries
SUPPLYone,crops = ALAREAone,crops · YIELDone,crops
(2) Supply of energy crops in European
countries
op
(3) Supply of crops in other countries
SUPPLYrest,crops = ƒ (PPrest,crops, tp_grrest,crops)
(4) Supply of animal products
SUPPLYcc,livest
= ƒ (PIcc,livest, FCIcc,livest, lab_indcc, cap_indcc,
int_indcc, tp_grone,livest, subs_milkcc,
FDEM_MLKcc,”milk”)
(5) Supply of oilseed products
SUPPLYcc,ospro
= Oilsd_ccc,ospro,oilseed · PDEMcc,oilseed
(6) Supply of residual feed
SUPPLYcc,feedres = ƒ (PDcc,feedres, tp_grcc, feedres)
(7) Supply of milk fat and milk protein
SUPPLYone,en_cr = ALAREAone,en_crp · YIELDone,en_crp +
ALAREAone,en_crp on set aside · YIELDone,en_crp
SUPPLYcc,milk_c = PDEMcc,milk_cont ·contentcc,milk_cont
ont
(8) Supply of dairy products
SUPPLYcc,milkpr = Addcomp_dairycc,milkproc MPDEMcc,milkproc +
subs_milk_dcc,milkproc
oc
(9) Supply of biofuels
SUPPLYcc,energ
= ƒ (PIcc,energ, BCIcc,energ, pdem_trcc,energ)
(10) Supply of gluten feed
SUPPLYcc,”gf”
= i_ethanol coef_p_bf i_ethanol PDEM cc, i_ethanol
Demand equations
Human demand (MCP equation 2, one specification)
(11) Human demand
HDEMcc,comm.
= ƒ (PCcc,comm, pop_grcc, inc_grcc, subs_milkcc,
hdem_trcc,comm)
Seed demand (MCP equation 3, two specifications)
(12) Seed demand in European countries
SDEMone,crops
= Seed_cone,crops · ALAREAone,crops
(13) Seed demand in other countries
SDEMrest,crops
= Seed_crest,crops · SUPPLYrest,crops
Processing demand for oilseeds and milk (MCP equation 4, four specifications)
(14) Processing demand for oilseeds
PDEMcc,oilseed
= ƒ (PDcc,oilseed, PDcc,ospro, pdem_trcc,oilseed)
(15) Processing demand for biofuel inputs
PDEMcc,i_biofuel
= PDEM_BF cc, energ,i_biofuel
(16) Processing demand for milk to fat and PDEMcc,milk
protein
= ***
(17) Processing demand for fat and protein PDEMcc,milk_cont = mlkproc MPDEM cc,milk_cont,mlkproc
Processing demand for various dairy products (MCP equation 5, one specification
(18) PDEM for fat and protein in different
dairy products
MPDEMcc,,dairy_co = ***
mp
Feed demand for all feed products except milk (MCP equation 6, one specification)
(19) Feed demand
FDEMcc,feed
= Feed_exogcc,feed + livest FRATEcc,feed,livest ·
SUPPLYcc,livest
Feed demand for milk (MCP equation 7, one specification)
(20) Feed demand for milk
FDEM_MLKcc,”milk”
= feed_milkcc · SUPPLYcc,"milk"
Feed rates for livestock (MCP equation 8, one specification)
(21) Feed rate
FRATEcc,feed,livest
= ƒ (PDcc,feed, tp_frcc,livest)
Total domestic use (MCP equation 9, one specification)
(22) Total use
TUSEcc,comm.
15
= HDEMcc,comm + SDEMcc,comm + PDEMcc,comm
+ FDEMcc,comm.+ FDEM_MLKcc,comm.+
subs_milk_scc,comm
Table 2.1: Overview of Equations in ESIM (continued)
Area and Yield Equations
Yield (MCP equation 10, three specifications)
(23) For non-quota crops
YIELDone,crops
= ƒ (PPone,crops, int_indone, lab_indone,
tp_grone,crops)
(24) For quota crops
YIELDcc,qu
= ƒ (PPcc,crops, PSHcc,qu, int_indcc, lab_indcc,
tp_grcc,crops)
(25) For energy crops on set aside
YIELDcc,sa_crp
= YIELDcc,en_crp
Unrestricted area per product (MCP equation 11, three specifications)
(26) Unrestricted product specific
area
ALAREAone,crops
= ƒ (PIone,crops, LP1one lab_indone, int_indone,
cap_indone)
(27) For sugar
ALAREAone,qu
= ƒ (PIone,qu, PSHone,,qu, LP1one, lab_indone,
int_indone, cap_indone)
(28) For biofuels crops on set aside
land
ALAREAone,sa_crp = ƒ (PPone,sa_crp, LP2one, lab_indone, int_indone,
cap_indone)
Obligatory set aside area (potentially to be scaled) (MCP equation 12, one specifications)
(29) Determination of obligatory set
aside area
OBLSETASone
= setas_eu15 + setas_eu12
Determination of agricultural land use (MCP equations 13-18, 6 specifications)
(30) Total available agricultural area
for non set aside land
Limiteff1ne
= Area_maxone * ch_areaone – marg_landone *
OBLSETASone
(31) Total available area for set aside
crops
Limiteff2one
= ALAREAsetas_crp + marg_landsetas_crp *
OBLSETASsetas_crp
(32) Land supply for non set aside
land
LS1one
= Limiteff1one – bend_ldone / shift_ldone + LP1one
(33) Land supply for set aside land to LS2setas_crp
be used for non-food production
= Limiteff2setas_crp – bend_ld2setas_crp / LP2setas_crp
(34) Market clearing for non set aside LS1one
land
= non set aside crops ALAREAone, non set aside crops
(35) Market clearing for set aside
land
= sa_crp ALAREAone, sa_crp
LS2one
Direct Payment Equations
Direct payments (MCP equation 19, two specifications)
(36) Direct payments for crops per
ton
DIRPAYone,crop
= DIRP_t_ncone,crop /YIELDone,crop •
YIELD0one,crop
(37) Direct payments for livest. per
ton
DIRPAYone,livest
= DIRP_t_ncone,livest
Price equations
Lower bound of Logit-function (MCP equation 20, four specifications)
(38) For NMS prior to accession
P_LOone,it
= PWit/exrateone · (1 + subs_adone,it).
(39) For EU products with interv. price P_LOone,it
= MAX(PWit/exrateone, intprone,it + exstabone,it)
(40) For EU and delayed integration
P_LOdelay_r,delay_c
= delaydelay_r,delay_c PD”EU15”,delay_c exrateeuro/
exrateone ***
(41) For EU products without interv.
Price
P_LOone,it.
= PWit/exrateone
Upper bound of Logit-function (MCP equation 21, three specifications)
(42) For NMS prior to accession
P_UPone,it
= PWit/exrateone (1+tar_adone,it)
(43) For EU products with thresh. price
P_UPone,thresh
= MAX(PWthresh/exrateone, thrprone,thresh)
(44) For EU products with ad valorem or P_UPone,it
specific tariffs
16
= PWit/exrateone · (1+tar_adone,it) + sp_done,it
Second upper bound of logit function because of export subsidy (MCP equation 22, one specification)
(45) Second upper bound
P_UPone,it
= MAX(exp_subone,it, qual_adone,it) +
PWit/exrateone
Price transmission function (MCP equation 23, five specifications)
(46) World market price
PDrow,it
= PWit
(47) PW transmission for EU markets
without export subsidies
PDone,it
= Logistic function
(48) PW transmission for EU markets
with export subsidies
PDone,it
= Logistic function
(49) PW transmission for new members
with delayed integration
PDone,it
= Logistic function
(50) PW transmission for non-members
PDone,it
= Logistic function
Shadow price determination (MCP equation 24, three specifications)
(51) Shadow price livestock
PSHone,livest
= ƒ (quotaone,livest, subs_milkone,ag,
FDEM_MLKone, livest, PIone,livest, FCIone,livest,
lab_indone, cap_indone, int_indone, tp_grone,livest)
(52) Shadow price crops
PSHone,crops
= ƒ (quotaone,crops, PIone,crops, LP1one, lab_indone,
cap_indone, int_indone, tp_grone,crops)
(53) Shadow price voluntary set aside
PSHone,"setaside" = ƒ (quotaone,"setaside", PIone,"setaside", LP1one,
lab_indone, cap_indone, int_indone, tp_grone,crops)
Wholesale/producer price transmission (MCP equation 25, four specifications)
(54) Producer price if margin
PPcc,nq
= PDcc,nq/margin0one,nq
(55) Producer price if no margin
PPcc,nq
= PDcc,nq
(56) Producer price for quota products
PPone,qu
= MIN(PDone,qu/margin0one,qu, PSHone,qu)
(57) Producer price for energy crops on
set aside
PPone,sa_crp
= PPone,en_crp
Consumer price (MCP equation 26, two specifications)
(58) For agri-food commodities
PCcc,comm
= PDcc,comm · + pctaxcc,comm.
(59) For mineral oil
PCcc,"cr_oil"
= p0cc,"cr_oil" · + pctaxcc,"cr_oil"
Determination of producer incentive price (MCP equation 27, four specifications)
(60) For non quota products
PIcc,nq
= PPcc,nq + prod_effcc,nq • DIRPAYcc,nq
(61) For quota products
PIcc,qu
= MIN(PPcc,qu + prod_effcc,nq • DIRPAYcc,qu,
PSHcc,qu)
(62) For energy crops in EU on non setaside
PIcc,energ
= PDcc, energ + pay_biof/energ member SUPPLY
c,energ • exrate"EURO"/exratecc
(63) For energy crops in non EU
PIcc,energ
= PDcc,energ
Net Price for biofuel inputs (MCP equation 28, two specification)
(64) Net Price for biodiesel inputs
NetPDcc,i_biodies = PDcc,i_biodiesel - PDcc,i_biodiesel,byproduct
el
(65) Net Price for bioethanol inputs non- NetPDcc,i_bioeth = PDcc,i_bioethanol - PDcc,i_bioethanol,byproduct
sugar
naol
(66) Net price for bioethanol input sugar
in non-EU
NetPDnon_memb = PDnon_member, “sugar” – PDnon_member, “sugar”, byproduct
(67) Net price for bioethanol input sugar
in EU
NetPDmember,
ers, “sugar”
= PDmember, “sugar” – PDmember, “sugar”, byproduct
“sugar”
Feed cost index (MCP equation 29, one specification)
(68) Feed cost index
FCIcc,livest
= feed FRATEcc,feed,livest · PDcc,feed/ FC_0cc,livest
Biofuel input cost index (MCP equation 30, one specification)
(69) Biofuel input cost index
BCIcc,energ
17
= i_biofuel QUANCES0cc,i_biofuel ·NetPDcc,i_biofuel/
i_biofuel QUANCES0cc,i_biofuel /BCI0
Equations for CES technology in biofuel production
Determination of unscaled relative input quantities (MCP equation 31, two specifications)
(70) Biodiesel
QUANCEScc,energ,i_biodiesel
= CES function
(71) Bioethanol
QUANCEScc,energ,i_bioethanol
= CES function
Scaling of relative input quantities to add up to supply of biofuels (MCP equation 32, one specification)
(72) Scaling of input quantities
QUANCEScc,energ,i_biofuel / = PDEM_BFcc,energ,i_biofuel/convbfenergy,i_biofuel/
SUPPLYcc,energ
i_biofuel
QUANCEScc,energ,i_biofuel
Other equations
Net exports (MCP equation 33, one specification)
(73) Net exports
NETEXPcc,it
= SUPPLYcc,it – TUSEcc,it
Determination of trade shares (MCP equations 34-36)
(74) Share of net exports in domestic
market volume (EU)
(75) Share of net exports in domestic
market volume (delayed integration
region)
TRADESHReu,it
= member NETEXPmember,it/member TUSEmember,it
· 100
TRADESHRdelay_ = delay_c NETEXPdelay_r,delay_c/ delay_c
TUSEdelay_r,delay_c· 100
r,delay_c
(76) Share of net exports in domestic TRADESHRnome
market volume (individual countries) mber,it
= NETEXPnomember,it/MAX(SUPPLYnomember,it,
TUSEnomember,it) · 100
Determination of quality shares (MCP equation 37, one specification)
(77) Share of high quality exports in
domestic market volume
QUALSHReu,it
= QUALQUANT"eu",it/member TUSEmember,it ·
100
Determination of export subsidy shares (MCP equation 38, one specification)
(78) Share of export subsidy limit in
domestic market volume
SUBSHReu,it
= SUBQUANT"eu",it/ member TUSEmember,it · 100
Determination of TRQ shares (MCP equation 39, one specification)
(79) Share of TRQ in domestic
market volume
TRQSHReu,it
= TRQ"eu",it/ member TUSEmember,it · 100
Preferential sugar imports or pre-fixed sugar imports (MCP equation 40, one specification)
(80) Pref. or pre-fixed sugar imports SUGIMP_EU”suga = ƒ (preftrq, exrateeuro, PD”sugar”, PW”sugar”,
sug_imp_exog”sugar”)
r”
World market clearing (MCP equation 41, one specification)
= cc NETEXPcc,it
(81) World market clearing condition 0
Domestic market clearing (MCP equation 42, one specification)
(82) Domestic market clearing
condition for non tradables
= TUSEcc,nt
SUPPLYcc,nt
Determination of milk prices derived from fat and protein prices (MCP equation 43, one specification)
a
(83) Price for milk derived from fat
PDcc,”milk”
= Addcompcc,”milk” · dairy_comp contentcc, dairy_comp
and protein
· PDcc, dairy_comp
In the MCP formulation, equations often include a set of different specifications. For example, the specification of
the supply equation for animal products in the EU is different from the specification of the EU supply equation for
crops. For better readability, all specifications of equations are numbered consecutively in this table and referred to
as equations throughout the documentation. In addition, equations are displayed throughout this documentation
with the dependent variable at the left hand side – in contrast to the MCP formulation in the GAMS code, in which
equations are written as implicit functions such that a zero appears at the left hand side.
18
Table 2.1 gives an overview of definitional equations which are for the most part spelled out
completely (e.g. equation 1), as well as behavioural equations such as equation (3). The latter are
constant elasticity functions throughout, and are presented in Table 2.1 only in their general form.
In the case of behavioural equations, some parameters (like tp_gr for productivity growth in
equation 3) are cited, whereas intercepts and elasticities are generally omitted for reasons of
readability.
Supply includes farm supply of crops and animal products. In the case of crops in European
countries, supply (1) is defined as product specific area (26-27) multiplied by yield (23-24). For
energy crops, yield on set aside land is added additionally (2). Supply of crops in other countries
(3) is a direct function of own and cross producer incentive prices (PP) and technical progress
(tp_gr). Animal product farm supply (4) is a function of own and cross incentive prices (PI), a
productivity shifter, indices for labour-, capital- and intermediate-costs and a feed cost index (68)
based on feed composition and component prices. For milk, supply includes two additive
exogenous elements "subsistence milk" and "feed milk". Producer incentive prices (PI) are an
aggregation of producer prices (PP) and direct payments per product unit (DIRPAY), as specified
in equation (60).
Processing supply of oilseed products (5) is a linear transformation of processing demand for
oilseeds (14). Supply of residual feed components "other energy" and "other protein" (6), mainly
being residuals of food production, depends on own price and a technical progress shifter.
Supply of biofuels (9) is a function of incentive prices (PI), trends in processing capacities in
biofuel production (pdem_tr) and a biofuel input cost index (69). The latter weights net input
prices with input quantities of the base period. Relative input quantities are determined by CES
functions (70, 71) and net input prices are defined as input prices minus prices of byproducts (6467). Equation (72) ensures that relative input quantities add up to total supply of biofuels. Supply
of gluten feed finally is a linear transformation of processing demand for inputs in bioethanol
production (10).
Demand includes human demand, seed demand, processing demand and feed demand. Human
demand (11) is a function of own and cross (consumer) prices (PC), income, trends in private
consumption and population. Only in the case of milk an additive autonomous demand component
"subsistence milk" is modelled. Consumer prices are defined at wholesale level (PD) and are
adjusted for any product specific taxes (58).
Seed demand in European countries (12) is a fixed quantity per area unit (seed_c) allocated to the
product concerned, while it is a fixed quantity per output unit for other countries (13). Processing
demand for oilseeds (14) is a function of prices of the respective processing input (oilseeds) and
outputs (oils and cakes) as well as trends in processing capacities in biofuel production. For oil,
sugar and cereals as inputs in biofuel processing, demand equals the sum of processing demand of
the respective input for both biodiesel and ethanol (15).
19
Feed demand per animal output unit (21), defined for each of the animal products, is a function of
domestic market prices for feed (PDcc,feed). Feed prices are the domestic wholesale prices (PD)
adjusted for any feed subsidies (?). Total feed demand (19) is defined as the sum over animals of
feed demand per animal unit (FRATE) multiplied by animal production, plus feed demand of
animals not modelled in ESIM as an autonomous feed component feed_exog. Feed demand of
milk is a linear transformation of milk supply (20). Total use (22) is the sum of feed demand,
human demand, processing demand, seed demand and a component of subsistence milk
production in new member states subs_milk_s.
In the European countries, supply of crops is composed of a yield and an area component. Three
different yield functions exist. i) Yield for non-quota crops (23) is dependent on own price, the
price indices of non-agricultural inputs (int_ind, lab_ind) and a productivity shifter. ii) Yield for
quota crops is modelled in a similar manner, but depends in addition on the shadow price of any
particular quota crop (24). iii) The yield of energy crops on set aside land (25) is assumed to be
equal to yield on non-set aside land.
The area allocation process is divided into three specifications as well: i) For non-quota products
it’s a function of own and cross incentive prices (PI), price indices of labour, capital and
intermediates (lab_int, cap_ind, int_ind) as well as the land price (LP1) (26). ii) Area allocation for
sugar (the only quota crop in ESIM at the moment) depends on the same values plus the shadow
price for sugar (27). iii) The area for biofuel crops on set aside land is determined by the producer
price (PP) instead of the incentive price and a different land price (LP2) (28).
Land prices are equilibrium prices of the land market. The supply of non set aside land (32) (LS1)
is a function of the total available agricultural area for non set aside land (Limiteff1/equation 30),
two location parameters and the price for non set aside land. Synchronously the supply for set
aside land (33) is determined by the total available area for set aside land (31), location parameters
and the price for set aside land (LP2). Market clearing on the land market is ensured by equations
(34) and (35). The effective obligatory set aside area consists of the set aside area in the EU15 and
the EU12 (29).
Direct payments for crop products (36) are determined per ton of output and are independent
variables in the producer incentive price equation (60, 61) for the respective products. Also for
animal products (37) direct payments are calculated per ton of output. The amount and
composition (coupled and decoupled parts) of the payments are determined during the calibration
process. Equation (36) provides an adjustment of payments per ton in the base period to a ton in
the actual period caused by changes in the total yield.
The world market price (46) is set equal to the domestic price in the rest of the world, because no
policies are modelled for these countries. Lower und upper price levels (38-44) are defined for
European countries and new member states with a delayed integration, which would occur in clear
net export or net import situations. They are based only on world market prices, border policies
20
and a factor for the delayed integration. Lower and upper price levels enter the Logit functions
(47-50) which determine the domestic price level in between the bounds dependent on the share of
net exports in domestic market volume (74-76). For some products specific quantity trade policies
like export subsidy limits and TRQs apply. For these products the relevant quantities are
transformed into shares in total domestic use (78, 79) and, in the case of export subsidies, the
Logit function is calibrated such that the price falls at the export subsidy limit (45).
In case of accession, in ESIM, domestic prices in the NMS equal EU €-prices. This is not to say
that in reality observed domestic prices would equal, because various reasons other than policies
lead to heterogeneous prices (quality, transportation cost). These price differences are, however,
not depicted in ESIM. As a result, accession leads to price adjustment only to the degree to which
price differentials are explained by border policies in the base period and by the development of
real exchange rates in the NMS currencies against the €. Exceptions are those countries and
products with a delayed market integration.
Shadow prices are not only calculated for quota crops (only sugar in the current version, (52)) and
quota livestock products (only milk in the current version (51)), but also for voluntary set aside
(53). This is because a quantity limit to voluntary set aside (33% of total area for individual farms)
applies in the ESIM base period. If shadow prices are below market prices they are the explaining
variables for production decisions (56). If market prices are below shadow prices the market price
is the relevant variable for production decisions.
The producer price (54, 55) is linked to the market price by an exogenous marketing margin. For
energy crops on set aside land the producer price equals the producer price of energy crops grown
on conventional land (57). The incentive price (PI) for energy crops grown on set aside land in the
EU is a function of the domestic prices (PD) and the biofuel premia (62), in non EU countries the
incentive price is the same like the domestic market price (63).
Net exports are defined as total supply minus total use (73). For tradables, the market clearing
condition requires world net exports to equal zero (81), and for non tradables, domestic markets
must clear (82).
2.2
Price Mechanism
2.2.1 Basic Challenge
In the real world, various reasons including cif/fob spread, differing political price protection
depending on the net trade situation, and domestic transportation cost contribute to situations in
which domestic prices are different in an exporting compared to an importing situation. In bilateral
trade models using the Armington approach, this poses no problem as different price wedges are
typically imposed for imports and exports at the same time and adjustments to price changes are
smooth. For net trade models, however, different price wedges in an importing and an exporting
situation are more difficult to handle. Some modelling approaches abstract from such details and
simply model uniform policy determined price wedges independent from the trade situation (FAO,
21
2001). Other models that take these differences into account must somehow deal with the problem
of products that are not exported at the export-based price, because it is too low, and are not
imported at the import-based price, because it is too high. In such a case, the equilibrium price lies
somewhere between the export-based and the import-based price. One option applied in some
spread sheet models is to generate a stepwise price adjustment in situations were the net trade
situation is close to zero (Münch, 2002). Another option is to repeatedly solve the model to
reconcile net trade position and the resulting domestic price (Grethe, 2004).
Surry (1992) applies the logistic functional form in an econometric model in order to depict EU
price formation within the price band between intervention price and threshold price dependent on
the net trade situation. This concept is also applied in equilibrium simulation models, e.g. the
MISS model (Guyomard et al., 1993), WATSIM (von Lampe, 1999), GTAP (van Meijl and van
Tongeren, 2002) and a CGE of the French economy (Gohin et al., 2002).
2.2.2 Approach Chosen in ESIM5 (This Section refers to a Former Base Period; The Basic
Mechanisms have not Changed since, but the Actual Price and Policy Data)
While the above applications use the Logistic functional form only for products for which EU
prices are fixed institutionally, domestic price formation in ESIM is based on the Logistic
functional form for all products. Furthermore, the approach is extended to cover the application of
export subsidies (ES) for WTO limited quantities and imports within tariff rate quotas (TRQs).
Price transmission between domestic and import prices is determined by a Logistic functional
form and thus allows for a smooth course between higher import based prices to lower export
based prices. Figure 2.1 depicts an example of a price transmission function for wheat in the EU in
the ESIM base period (2000-2002). The horizontal axis depicts the share of net exports in total
domestic use (in %), and the vertical axis depicts the wheat price in €/t.
5
The following sections are copied from Banse and Grethe (2006).
22
Figure 2.1: Price Transmission for Wheat (EU-15)
200
Price (€/t)
160
120
80
40
0
-15
-12.5
-10
-7.5
-5
-2.5
0
2.5
5
7.5
10
12.5
15
Net Trade Share (%)
calibrated Logit-Function
Figure 2.1 shows, that in a situation in which the share of net exports in total domestic use is larger
than 5%, i.e. the EU is in a very clear net export situation, the domestic price is at the lower price
bound PLO, which is the maximum of the intervention price level and the world market price.
which for wheat is the maximum of the world market price and the EU threshold price for cereals,
which is 155% of the intervention price. In a situation where the net export share is in between –5
and 5% of domestic use, the domestic price is in between upper and lower bound, subject to the
course of the Logistic function, which is differentiable throughout.
This may depict reality quite well: The closer the net trade situation comes to zero, the less the
domestic price is determined by the respective import or export price alone, and the more it is
subject to domestic price formation, but also to the effects of import and export prices at the same
time, as considerable intra-industry trade may be hidden behind a net export situation of zero. The
precise course of the Logistic function, however, may be subject to discussion – and the functional
form allows for adjustments. The specification of the Logistic function depicted in Graph 2.1 is
TRADSHR
NX
(2.1) PD   PUP  PLO     e
 PUP , with TRADSHR 
 100 , with α and β = 1.
TRADSHR
TUSE
1   e
But it may be empirically evident that the domestic price is more closely linked to international
prices even in case of relatively low net trade shares. In such a case one may like to model the
price transmission function steeper. This can be done by setting α different from 1. Also one may
find empirically that the trade share at which the domestic price is predominantly determined by
international prices is asymmetric. Such asymmetric course of the Logistic function can be
modelled by adjusting parameter β.
23
But more complex policy determined price transmission mechanisms may apply. For example, an
ES may allow keeping the domestic price significantly above international level, but only up to a
certain limit, the WTO bound quantity limit. After that limit the domestic price would fall to the
world market level for a comparable quality. On the other hand, TRQs may allow for considerable
imports, without causing the price to move in the direction of the upper price bound consisting of
threshold price or world market price and tariff, because imports occur at tariff levels below the
Most-Favoured-Nation (MFN) level, potentially without any tariffs at al. Both these situations can
be depicted by the Logistic function model of price transmission. Figure 2.2 shows the price
transmission function applied in ESIM for beef, taking into account ES commitments of the EU.
Figure 2.2: Price Transmission for Beef with ES
4000
3500
3000
Price (€/t)
2500
2000
1500
1000
500
0
-15
-12.5
-10
-7.5
-5
-2.5
0
2.5
5
7.5
10
12.5
15
Net Trade Share (%)
Logit-Function (w/o subsidies)
Logit-Function (incl. exp. subs.)
Final Logit
The final price transmission function consists of two separate Logistic function specifications by
defining the maximum of those two separate functions as the relevant price transmission function.
The upper bound is valid in a clear net import situation. If net trade comes close to zero the price
begins to fall and reaches the upper bound of the second Logistic function (PUP_2) at a net export
level of about 0%, and begins to fall further at an export level of 9.7%, which results from
transforming the WTO bound quantities into an export share equivalent. At about 15% export
share the price transmission function reaches its lower bound, which is the world market price.
The final Logistic function takes the maximum value out of both individual Logistic functions.
Eq. (2.2) shows the specification of the Logistic function establishing the middle level of the price
transmission function, which applies in the range of net exports which are assumed to take place
while receiving ES.
24
*
NX  subsquant
  eTRADSHR
(2.2) PD*   P

P

 PUP , with TRADSHR* 
 100 ,

UP _ 2
LO
*
TUSE
TRADSHR
1   e
with α = 1, and β = 3.
What now, if TRQs come into play, which are widely applied by the EU, among others for beef
from ACP countries as well as minimum and current access TRQs for beef which are bound in the
WTO? Such TRQs can take the form of predetermined imports, which may result from the fact
that the TRQ is clearly binding as well as a situation where the quantity which can be delivered to
the EU market is rather fixed at a level below the TRQ due to restricted production potential in the
exporting country. If further the assumption is that they enter the market at zero or low tariffs and
thus do not drive the domestic market price up to the level of the world market price plus MFN
tariff, their influence on the price transmission function can simply be modelled as a leftward shift:
Although considerable imports take place, the domestic price still is at the lower price bound.
Figure 2.3 shows the resulting price transmission function for beef applied in ESIM.
25
Figure 2.3: Price Transmission for Beef with ES and TRQs
4000
3500
3000
Price (€/t)
2500
2000
1500
1000
500
0
-15
-12.5
-10
-7.5
-5
-2.5
0
2.5
5
7.5
10
12.5
Net Trade Share (%)
Logit-Function (with TRQ, w/o ES)
Final Logit (with ES and TRQ)
The complete price transmission-function is shifted leftward compared to Figure 2.2 above.
Without ES, the domestic price would remain at the lower (world market) bound up to an
import level of about 4%. This is because imports would occur at the world market price
without any tariff until the TRQ is filled. Alternatively, a preferential tariff with a reduced rate
could be modelled.
This is because imports would occur at the world market price without any tariff until the
TRQ is filled. Alternatively, a preferential tariff with a reduced rate could be modelled.
2.2.3 Determination of the Upper and Lower Bounds of the Price Transmission
Functions
Three commodity groups are distinguished in ESIM with respect to the way in which upper
and lower bounds of their respective price transmission functions are determined. For the first
group the upper as well as the lower bound consist of the world market price plus any tariff in
case of the upper bound or any ES in case of the lower bound. If no tariffs or ES exist, the
Logistic function is a horizontal price line without any step. For a second commodity group
the lower price bound is the maximum of the intervention price and the world market price. If
the intervention price is above world market price the EU is willing to buy each quantity –
resulting in a situation where the intervention price is thus the absolute price minimum. ES
are not modelled for these products, they are derived from model results as the difference
26
15
between intervention price and world market price multiplied by supply minus demand, up to
the ES limit of the EU. If market surplus exceeds the WTO bound it is modeled as going into
stocks. The upper price bound for products in the set FLOOR is the world market price plus
tariff, which would be the domestic price in a clear net import situation. For a third
commodity group the lower bound is modeled as for the second group, but the upper bound is
defined relative to the lower bound. In the current ESIM version this group only includes
some cereals and the level of the upper bound is 155% of the intervention price.
2.2.4 Changes in TRQ and Export Subsidy Levels
For the base situation the observed amount of preferential imports under TRQs and applied
ES are included in the calibration of the price transmission functions. Therefore, the observed
domestic market price is a function of the given level of net exports including TRQs and ES.
If the level of ES and imports under TRQs remains unchanged over the projected period of
time, no additional impact on domestic market prices is expected. However, if the limit of ES
or the level of imports under TRQs alters, changes in domestic prices are expected.
For changes in ES the second Logistic function will shift to the left and domestic prices will
begin to fall to PLO at lower export shares (see Graph 2.2). An expansion of TRQs, on the
other hand, will have two effects: first, a leftward shift of the price transmission function (see
Graph 2.3), and second an (exogenous) increase of supply of the commodity concerned which
affect market-clearing. Consequently, net-exports will increase, i.e. the domestic price will
move towards the lower export price PLO. The second effect is covered technically by adding
any change of imports under TRQs to the right hand side of the net trade equation in the EU,
i.e. net exports = supply +  TRQ – domestic use. The additional quantity  TRQ is deducted
on the right hand side of the net export equation for the rest of the world.
2.2.5 The Modelling of Delayed Integration of Selected Regional Agricultural Markets
into the Single Market
The 2004 enlargement brought about a significant extension of the EU territory. Increasingly,
transport costs counteract the virtues of the single markets, i.e. the preference of delivering
EU produce from surplus to deficit regions. High transport costs due to long distances as well
as due to insufficient infrastructure significantly constrain the integration of a number of
regional markets and play an increasing role, e.g., in the situation of land locked new Member
States. Therefore, price prospects for some agricultural commodities appear different in trend
and in level to those found in the majority of EU countries. Further enlargement as well as
changes of import and export policies might see these characteristics reinforcing.
ESIM provides for a modification of price formation which currently takes place at the EU
level. It is enabled to analyse the impact of delayed integration of individual markets in
individual current and future new Member States which are decoupled for a specified time
27
period from the single market in respect to price formation. In the previous version of ESIM,
EU-accession was modelled as an instantaneous introduction of a single EU price in the
respective acceding country. Each new member state’s trade share was captured to calculate
the trade share of the enlarged EU. Therefore, if the EU prior to accession was a net-exporter
while the acceding new member states were net-importers, the trade share of the enlarged EU
became smaller or even negative. As a consequence, the single market price in the enlarged
EU increased.
The extended version of ESIM includes an option which simulates a gradual path of
integration as it could happen when infrastructure is gradually improved. The gradual path of
integration is modelled as a combination of individual country based price formation
equations (logistic functions) and price formation which is based on the single market without
country-specific price formation equations. It is therefore possible for the ESIM user to select
the speed of integration into the single market for selected new member states and products
individually.
The price transmission mechanism is applied to the EU including New Member States and
accession candidates. Before accession, this price mechanism describes the transmission of
prices in all candidate countries. With accession to the EU, all policy measures are applied by
the new members. In an older version of ESIM, the integration to the Single European Market
was modelled as an immediate introduction of the ‘single market price mechanism’ which
describes a common price for all EU Member States.
Now it is possible to postpone this introduction of the ‘single’ EU price for some selected
countries or regions for selected products. These countries will follow for a defined period of
time their national price formation mechanism however with a full introduction of all policy
measures, e.g. intervention prices, direct payments, tariffs and subsidies. Therefore, the
intervention price will serve as the lower bound also in these selected countries. If world
prices are higher than intervention prices, world prices determine domestic market prices in
the EU. For the selected countries which are excluded from the price formation of the single
European market, domestic market price for cereal will be lower than for the other EU
member states, if world market prices are higher than intervention prices.
2.3
The Land Market
The basic idea of the land market module in ESIM relies on a land supply curve, which
specifies the relation between land supply and the price for land in each region. Thereby, a
distinction between rental and purchase prices is not relevant. Land supply to the agricultural
sector can be influenced by urbanisation, which is a very common situation in EU member
states, and by conversion of non-agricultural land into land that can be used for agricultural
purposes (VAN MEJIL et al., 2006). The latter case occurs in some of the NMS only (see
28
below). In addition, also political measures, e.g. obligatory set-aside restrictions, have a direct
influence on land supply. Figure 2.4 illustrates the land supply curve.
Figure 2.4: Land Supply Curve Determining Land Conversion and Land Prices
Land price
Land supply
D2’
P2’
P2
D2
D1’
D1
P1’
P1
Agricultural land
Q1
Q1’
Q2 Q2’
ω
Limiteff
Source: Own composition, following VAN MEIJL ET AL. (2006).
The design of the land supply curve is based on the idea that the most productive land is first
taken into production. Additionally, the physical limit of taking additional land into
agricultural production is considered.
If the difference between land, which is used for agricultural land, and the overall endowment
of land resources, which could potentially be used by farming activities, is large, an
increasing demand for agricultural land would lead to a conversion of non-agricultural land
into agricultural land (VAN MEIJL et al., 2006). Such a situation is depicted at the flat part of
the land supply curve in Figure 2.4. Here, an increase in demand for land by farmers, which
could, for example, result from an increase in the overall price level for agricultural products,
shifts the land demand function from D1 to the D1’. Though a significant amount of land is
additionally taken into production, which is represented by the shift from Q1 to Q1’, land
prices increase only moderately from P1 to P1’. Thereby, the increase in land prices reflects
the costs of bringing additional land into production. In contrast, if overall land endowments
are scarce and there is not much room left to bring additional land into production, an increase
in demand for agricultural land leads to significant increases in land prices, while agricultural
area is extended only slightly. Such a situation is depicted on the steeper part of the land
29
supply curve. While land prices increase considerably from P2 to P2’, agricultural land is
extended from Q2 to Q2’ only.
The mathematical specification of the land supply function as well as the interactions between
land supply and land demand are not identical to the specifications made by VAN MEIJL et al.
(2006). This can be traced back to fact that the model structure of the CGE model, which is
the basis of the approach chosen by VAN MEIJL et al. (2006), is different from the structure of
a partial equilibrium model like ESIM. Accordingly, mathematical specifications regarding
the inclusion of the land market in ESIM rely partly on own considerations, while the basic
idea regarding the features of the land market function are taken from VAN MEIJL et al.
(2006).
In ESIM, the shape of the land supply (LS) curve is modelled according to the following
equation6:
LS
cc

Limiteff  bend _ ld / shift _ ld  LP
cc
cc
cc
cc

where
Limiteffcc
is the effective total amount of land, which can potentially be used for
agricultural production,
bend_ldcc
is a parameter determining the bend of the land supply curve,
shift_ldcc
is a parameter determining the bending of the land supply curve,
LPcc
is the land price.
Limiteff is endogenous to the model and is determined as follows:
Limiteff
cc

area _ max  ch _ area
cc
cc
 marg_ land
 Oblsetas
cc
where
area_maxcc
is the total amount of land, which can potentially be used for agricultural
production, which includes currently used agricultural land, fallow land
plus obligatory set-aside area,
ch_areacc
is the rate, by which area_max changes each year over the whole
simulation period due to urbanisation or conversion of land into
potentially usable agricultural area,
6
There are two land supply functions: one for non set aside land and one for set aside land used for non-food
production.
30
Oblsetascc
is the obligatory set-aside area, which depends on the set-aside area in the
EU15 and in the EU12.
The following paragraph describes, which mathematical specifications, mechanisms, and
equations ensure the equilibrium on the land market by the adjustment of land prices.
Assume again an increase in the overall price level for agricultural products. As mentioned
above, this leads to an increase in overall demand for agricultural area given the area
allocation function as specified in previous ESIM versions:
 ln area _ int cc,cr, j    cc,cr, j  ln PIcc, j    cc,cr  ln cap _ ind cc
j
Alarea cc,cr  exp 
   cc,cr  ln lab _ ind cc  cc,cr  ln int_ ind cc




where
area_intcc,cr,j
is the area intercept,
εcc,cr,j
is the elasticity of area allocation with respect to prices,
λcc,cr
is the elasticity of area allocation with respect to capital costs,
μcc,cr
is the elasticity of area allocation with respect to wages,
ςcc,cr
is the elasticity of area allocation with respect to costs of intermediates.
However, in order to ensure that total land demand does not exceed total land supply, and in
order to meet the overall equilibrium condition on the land market, which requires that
LS
cc

 Alarea
cr
cc,cr
,
the land price increases. This, in turn, compensates the stimulating effect of increasing
product prices on area demand to some extent, since the land price has been introduced as one
argument into the land allocation function, which is now re-specified as:
 ln area _ int cc,cr, j     cc,cr, j  ln PIcc, j    cc,cr  ln cap _ ind cc
j
Alarea Cr,C  exp 
   cc,cr  ln lab _ ind cc  cc,cr  ln int_ ind cc  cc  ln LPcc




where
σ,cc
is the area allocation elasticity with respect to the land price.
In ESIM direct payments are part of at the incentive price PI, which consist of the producer
price and the direct payment multiplied by a parameter of the effectiveness of the payments.
The amount and composition (coupled and decoupled parts) of the direct payments are
determined already during the calibration process.
31
PIcc,comm  PPcc,comm  prod _ eff cc,comm * DIRPAYcc,comm 7
Even though in reality direct payments are connected to land, in ESIM it has the same effect
whether the payments are linked to the incentive price or the land price. Both variables are
part of the area allocation function and the elasticities of area allocation with respect to
incentive prices are set proportionally to the elasticities of area allocation with respect to land
prices. Furthermore, products with high marginal revenues have low shares of area costs,
which means that they react less on land prices than products with low marginal revenues in
either case, when the payments are inserted at the incentive or at the land price. So, similar
impacts to area demand are received, no matter where the payments are introduced.
Due to limited data availability it was not possible to estimate area allocation elasticities.
Therefore and in contrast to all other input elasticities the area allocation elasticity with
respect to the land price is the same for all kinds of area uses. We applied a rather low
elasticity value which reflects a rather ‘sticky’ behaviour for agricultural land markets.
However, it should be mentioned that this assumption has consequences for the functioning of
the land markets and needs further research. The approach of determining this elasticity,
however, corresponds to the determination of the labour, capital, and intermediate
elasticities.8
In order to calibrate the parameters of the land supply function, shift_ld has been assumed to
be proportional to the total land supply in each country. The parameter bend_ld has been
calibrated in such a way that it reproduces the base data for land prices, total land supply, and
effective total amount of land, which can potentially be used for agricultural production. Most
land prices have been taken from EUROPEAN COMMISSION (2007) and LATRUFFE and LE
MOUEL (2006), who provided a comprehensive overview of the land markets in various
member states of the EU. Due to a lack in data availability, however, some land prices rely on
own estimations. Table 2.2 shows the levels of land prices assumed for the analysis in this
work. Though ESIM does not distinguish between rental and purchase prices it shall be
mentioned that these figures represent rental rates.
It is striking that for the time of the base period, i.e. prior to EU enlargement and
implementation of the MTR reform, rental prices in the EU-12 are far below average prices in
EU-15 members. However, in both groups of member states significant differences among
countries exist. Romania and Bulgaria are expected to have the lowest land prices within the
7
In the model code it is distinguished between non quota products, quota products and energy crops. The
incentive price for energy crops is determined by a biofuel premia and not by direct payments.
8
It should be mentioned that the approach of land modeling is similar for all member states. Possible market
imperfections which might limit the functioning of land markets are only reflected in the parameters of the
functional form.
32
group of EU-12 Member States. Land prices in Hungary, Poland, and the Czech Republic are
three to four times higher.
Table 2.2: Land Rental Prices in EU Member States in the Base Period (in Euro)
Austria
200
Latvia
10
1
Belgium/Luxembourg
185
Romania
10
1
Denmark
290
Slovenia
25
Finland
150
Lithuania
12.5
1
France
123
Bulgaria
10
Germany
200
Poland
40
Greece
455
Hungary
45
1
Ireland
200
Czech Republic
30
Italy
377
Slovakia
25
1
Netherlands
370
Estonia
12
Portugal1
Spain
Sweden
United Kingdom
300
400
140
199
1
own estimation
Source: EUROPEAN COMMISSION (2007), LATRUFFE and LE MOUEL (2006), SWINNEN and VRANKEN
(2008) and own estimation.
Among the group of EU-15 members the highest prices for agricultural land exist in the
Netherlands (370 Euro) as well as in the Mediterranean countries (300 Euro to 455 Euro)
except France (123 Euro), where the prices are the lowest. The high level of rental prices in
the latter countries might occur due to the existence of irrigation systems on a number of
agricultural areas, which have been installed by landowners and are provided for tenants.
Another or additional explanation might be that the costs of irrigation are included in the
rental prices. The reason for the high level of land prices in the Netherlands may be the large
number of fruit and vegetable producing farms as well as the overall scarcity of land.
In case of the underlying modelling approach it is assumed that potentially available land,
expressed by the asymptote limiteff, is the sum of the agricultural land currently used for
production purposes and fallow land, which is neither obligatory nor voluntary set-aside land.
The asymptote area_max additionally includes obligatory set-aside area. Data for fallow land
are obtained by EUROSTAT (2008) and FAO (2007). Table 2.3 illustrates the amount of total
land demand, i.e. the amount of land that is actually used for agricultural production, and the
amount of land that could potentially be used for agricultural production (limiteff). In
addition, the rate, by which area_max is assumed to change year by year, is shown.
33
Table 2.3: Total Land Demand, Potentially Available Land, and Annual Change in Area
Total land
demand
Limiteff
Gap Land demand /
Limiteff (in % of
Limiteff)
Change in
area_max
3140.8
1406.0
2660.9
2077.0
26671.1
16390.4
1956.5
3820.7
11769.5
1735.1
1020.9
16621.2
2784.5
10626.0
3253.5
1469.4
2690.7
2202.1
28089.8
17450.2
2063.8
3826.3
11967.7
1759.2
1097.1
17318.4
3045.5
11088.9
3.6
4.5
1.1
6.0
5.3
6.5
5.5
0.1
1.7
1.4
7.5
4.2
9.4
4.4
-0.10
0.00
-0.20
0.30
-0.40
-0.50
0.00
0.00
0.00
0.00
-1.60
-1.00
0.30
-0.10
102680.5
107322.6
4.5
-0.37
Latvia
Romania
Slovenia
Lithuania
Bulgaria
Poland
Hungary
Czech Republic
Slovakia
Estonia
1133.0
13096.6
478.4
2292.3
4398.9
15154.7
5437.6
4113.8
2212.9
722.3
1147.5
14107.6
491.7
2531.9
4946.3
16547.7
5653.1
4191.0
2283.1
785.6
1.3
7.7
2.8
10.5
12.4
9.2
4.0
1.9
3.2
8.8
-0.40
0.00
0.00
0.00
-0.50
-0.50
0.50
0.20
0.00
0.00
EU-12
49040.5
52685.7
7.4
-0.14
Austria
Belgium/Luxembourg
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
United Kingdom
EU-15
Source: EUROSTAT (2008), and FAO (2007).
The table illustrates that in all countries of the enlarged EU additional land reserves exist,
which could be converted into agricultural land. In the EU-15, approximately 4.6 mill. ha, and
in the NMS, about 3.6 mill. ha, could additionally be taken into production. In both groups of
countries unused land amounts to approximately 5.5% of the maximum available agricultural
area. These figures indicate lower land reserves than published in VAN MEIJL et al. (2006),
who assume that more than 10 % of the maximum available area is not used for production
purposes. The calculation of the land reserve for this study is based on recently published data
by EUROSTAT which explains this difference. Among individual, countries, however,
34
significant differences exist. With respect to EU-15 members the highest shares of unused
land reserves are observed in the Scandinavian countries, Sweden and Finland (9.4% and
6.0%, respectively). In the member states of high population density, i.e. most of all the
Netherlands and Belgium/Luxembourg, unused land reserves amount to less than 5 % of the
maximum available area. According to recent time series obtained from EUROSTAT (2008)
total available area in most EU-15 members is assumed to remain constant over time or to
decrease by up to 1.6 % per year. An increase in total available agricultural area is assumed
for Sweden and Finland only.
The situation in the NMS corresponds much to the situation in their Western partner
countries. The share of unused land reserves in most countries lies between 1.3% in Latvia
and 12.4% in Bulgaria. The average decrease in maximum available agricultural area is
somewhat lower for total EU-12 on average than on EU-15 average (-0.14 % compared to 0.37 %). Within the group of EU-12 total area is assumed to decrease in Latvia, Bulgaria, and
Poland, while it increases in Hungary and the Czech Republic.
2.4
Feed Model
Except from the inclusion of roughages in the new ESIM version and an additive intercept in
the aggregate feed demand function, the feed model is unchanged compared to the ESIM
SuperCalc version. An overview of the feed model in ESIM is shown in Figure 2.5.
35
Figure 2.5: Schematic Overview of the Feed Model
A n im a l p r o d u c t p r ic e s
( 3 ) A n im a l
s u p p ly
(4 ) F e e d d e m a n d =
(1 ) • (3 )
(2 ) F e e d C o st In de x
(1 )
F e e d co m p o n e nt
d e m a n d / a n im a l
p r o d u c t u n it
(F R A T E )
C o m p o n e n t p r ic e s
Demand for feed components (including roughages) per unit of animal product (FRATE) is
dependent on wholesale prices of feed components (1).9 This relationship is shown in
equation (2.3):
(2.3) FRATE cc,feed,livest = frat_int cc,feed,livest •  PFfeed1
elastfd cc,feed,feed1,livest
•tp_fr cc,livest .
feed1
Own and cross price elasticities of feed demand (elastfdcc,feed,feed1,livest) as well as intercepts
(frat_intcc,feed,livest) are external parameters, the latter being calibrated from base data.
Elasticities, being chosen based on literature and plausibility considerations (see Section 3.4)
are composed such that non-positivity of the own price effect and homogeneity of the feed
demand function in all feed prices (including fodder and pasture for ruminants) hold globally
and symmetry of cross price effects holds locally. The parameter tp_frcc,livest is a model
exogenous technical progress parameter which can take a value below one in order to depict a
higher efficiency in feed use.
From the resulting feed composition and component prices a feed cost index (FCI) is
calculated according to (2.4):
(2.4) FCIcc,livest = feed (PFcc,feed · FRATEcc,feed,livest). FC_0cc,livest, with FC_0 representing feed
cost per animal output unit in the model base period.
9
Actually, the variable PF is the wholesale price adjusted for any feed subsidies (for example for SMP) which
may apply. This probably represents the situation well for poultry and egg production, which is organised
industrially and mainly relies on purchased compound feed. For red meat and milk production, however, it
may be more appropriate to use farm gate prices of feed components as explaining variables, as a large share
of animal feed is produced on farm.
36
The feed cost index and effective farm gate prices for animal products determine animal
supply (3) as defined above. Total feed demand in a country (4) is the product of (1) and (3),
plus a model exogenous additive intercept which represents feed demand of animals not
covered in ESIM.
Due to this modelling approach an increasing feed price for any feed component results in
reduced demand for this component due to two effects. First, the substitution effect, which
results in other components substituting for the more expensive one according to (2.3) and
secondly, the output effect which results, via an increasing FCI, in lower animal production
and therefore lower feed demand.
2.5
Processing Models
2.5.1 Oilseed Processing
The oilseed processing module is not changed compared to ESIM in SuperCalc. Processing
demand for inputs is defined as
(2.5) PDEM cc,oilseed = cr_int cc,oilseed •
 PD
elast_cr oilseed, ospro
ospro
elast_cr
• PD oilseed oilseed,oilseed .
ospro
Explanatory variables are wholesale prices for the processing input (the respective oilseed)
and processing outputs (meals and cakes, contained in the subset "ospro". The intercept
(cr_int) as well as the elasticities of processing demand with respect to input and output prices
(elast_cr) are exogenous parameters, the former being calibrated according to base data. (2.5)
is restricted to be homogenous of degree zero in all input and output prices (price elasticities
with respect to inputs other than oilseeds are taken into account in imposing the homogeneity
condition, but these prices are no explanatory variables in the model, see Section 3.6).
Processing supply is defined as processing demand multiplied by the respective extraction
factor:
(2.6)
SUPPLYcc,ospro = PDEMcc,oilseed • oilsd_ccc,ospro,oilseed.
2.5.2 Dairy Processing
***
2.6
The Bioenergy Market
2.6.1 Overview
The production of agricultural products for biofuel production (oilseeds/plant oils for
biodiesel; wheat, corn and sugar for bioethanol), as well as the processing of these products
and the production of biofuels have been depicted in ESIM. In addition, human demand for
biofuels is modelled. Several biofuel policies are depicted, but the focus is on shifting overall
37
biofuel demand to reach certain political quantitative targets such as a given share in total
transport fuel demand.
2.6.2 Supply of Biofuel Inputs
Supply activities for biofuel crops in ESIM are modelled similar to other crops. For European
Countries, crop supply functions are separated into two parts: a capacity (area) and an
intensity (yield) part. The supply of biofuel crops (sunseed, rapeseed, soybeans, corn, wheat,
sugar) in the EU is modeled by one isoelastic yield function and two isoelastic area allocation
functions for each biofuel crop: on none-set-aside area, area is a function of input prices,
direct payments, output prices for all other crops and the special energy crop premium. The
second area allocation function is for biofuel crops produced on set-aside area, which is a
function of input prices, direct payments, and output prices only for those crops used for
biofuel production, which may alternatively be grown on set aside area.
In the ROW and the US supply of biofuel crops is modeled by isoelastic supply functions
which do not differentiate between a yield and an area component.
Oilseeds are no direct inputs into the biofuel production activity, but are first crushed and
yield plant oils and oilcake. Processing demand for oilseeds is defined as
(2.7) PDEM cc,oilseed = cr_int cc,oilseed •
 PD
elast_cr oilseed, ospro
ospro
elast_cr
• PD oilseed oilseed,oilseed • pdem_tr .
ospro
Explanatory variables are wholesale prices for the processing input (the respective oilseed)
and processing outputs (meals and cakes), contained in the subset "ospro". The intercept
(cr_int) as well as the elasticities of processing demand with respect to input and output prices
(elast_cr) are exogenous parameters, the former being calibrated according to base data. (2.7)
is restricted to be homogenous of degree zero in all input and output prices (price elasticities
with respect to inputs other than oilseeds are taken into account in imposing the homogeneity
condition, but these prices are no explanatory variables in the model)
Processing supply of oilseed products (oilcake, oils) is defined as processing demand
multiplied by the respective extraction factor:
(2.8) SUPPLYcc,ospro = PDEMcc,oilseed • oilsd_ccc,ospro,oilseed.
Supply of palm oil is a direct function of own and cross domestic prices and technical
progress. Palm oil is only produced in the ROW and the supply of palm oil is modelled
without consideration of by-products such as palm kernel oil, palm kernel meal, tree stem and
skin.
38
2.6.3 Production of Biofuels and Biofuel Byproducts
The production of biofuels is modeled as an isoelastic function of the respective biofuel price,
and the weighted net prices of the respective inputs:
(2.9)
elastsp
SUPPLYcc,energ = sup_int cc,energ • PIcc,energenrg,energ • BCI
elast_en_inpenerg
• pdem_trenerg
Net prices are defined as market prices minus the related feed output price, which is for gluten
feed in case of corn and wheat, multiplied by the technical extraction factor which describes
how much gluten feed results from the processing of cereals to bioethanol.
The production of gluten feed is defined as the sum over cereals used in biofuel processing
multiplied by the respective extraction factors.
The shares of feedstocks in bioethanol and biodiesel production are determined by a CES
function based on net energy crop prices:
(2.10) QUANCEScc,energ,i_biofuel =
biof_CES_el
biof_CES_el
cc,energ
1
•
/NetPDcc,energ,i_biofuel
biof_CES_int biof_CES_shrcc,energ,i_biofuel

(1-biof_CES_el
)
•
cc,energ
biof_CES_shr
•
Net

PDcc,energ,i_biofuel
i_biofuel


cc,energ
 (1-biof_CES_el )
cc,energ




where QUANCES are the unscaled inputs which are determined according to relative prices,
biof_CES_el are the CES elasticities of substitution among inputs and biof_CES_shr are
calibrated parameters of the CES function. In addition, equation (2.11) scales the quantities
such that they add up, after technical conversion, to the total quantity of biofuel production:
PDEM _ BFenerg,i _ biofuel
(2.11)
convbfccenerg,i _ biofuel
SUPPLYenerg

QUANCESenerg,i _ biofuel

QUANCESenerg,i _ biofuel
i _ biofuel
2.6.4 Demand for Biofuels
Human demand for biofuels is a function of the respective biofuel price, the price of crude oil,
and the tax rates on biofuels and on mineral oil. In practice, when running a biofuel scenario,
human demand functions are shifted such as to reach a certain share in total transportation
fuels.
2.6.5 Biofuel Policies
Other policies depicted include the special premium of 45 €/ha (non-set-aside only), which is
modeled as a subsidy for the production of biofuels, assuming that it accrues to a large part to
biofuel producers, as it results in lower prices of biofuel inputs. EU targets with respect to the
39
share of biofuels in total transport fuels as set out in the EU Biofuel Directive are depicted as
shifters in the human demand functions and in the oilseed crushing and biofuel production
activities.
Finally, changes in the compulsory set-aside rate affect the production of crops for biofuel
production. Generally, a reduction in the obligatory set-aside area increases the total
agricultural area used for crop production. This increase, however, is less than 100% of setaside reduction in order to reflect the comparatively low productivity of set-aside area.
2.6.6 Data
Price information is generally obtained from EUROSTAT. For energy crops, producer and
market prices are identical to those applying if these products are used for food or feed
purposes. Palm oil and ethanol prices are obtained from the FAPRI outlook database.
Quantity data for first generation biofuels is based on data published in F.O. Licht Interactive
Data and World Ethanol and Biofuels Report.
2.7
The Sugar Market
2.7.1 Introduction
In November 2005 the Council of Ministers concluded about the reform of the sugar CMO.
The medium term adjustment of sugar production in the EU Member States depends on the
level of production costs and the adjustment possibilities of farmers. In the 2004 ESIM
version, the EU sugar sector is depicted at the aggregation level of the EU-15 and the 10 new
member states which acceded in 2004 as well as Romania, Bulgaria and Turkey. Yield and
area allocation functions are isoelastic. Quotas and resulting shadow prices in case of binding
quotas are depicted. No differentiation between A- and B-quota exists. C-sugar production is
not depicted explicitly, but the quota is set such that it includes C-sugar production.
Preferential imports are determined exogenously.
Sugar production is now formulated at the level of individual EU member states. This is to
improve accuracy of depicting supply response in the EU in case of significant price changes,
as envisaged in the new Common Market Organization (CMO). Supply response is expected
to differ heavily in case of decreasing EU prices among member states because of different
production costs and thus shadow prices for sugar.
In addition, in formulating EU supply response the assumption of an isoelastic supply
response was dropped. It is expected that sugar production will fall dramatically in some EU
member states with the envisaged price reductions – in some countries production may cease
completely. Therefore, the former isoelastic supply specification of ESIM is inappropriate for
sugar: any realistic change in price, say 30-50%, cannot drive production close to zero in
member states with significant supply in the base period. Therefore, a different functional
40
form with an intercept at the price axis is chosen for the area allocation function of sugar in
the updated ESIM version.
Unfortunately, no historical variation of price quantity combinations exist which could have
served as a basis for econometric analysis to establish supply functions at member state
element. Therefore, the shape of supply functions is now based on FADN production cost
data and their level is based on evidence on shadow prices from the literature.
Finally, in contrast to the 2004 ESIM version where preferential EU imports are set
exogenously, the quantity of preferential EU imports is now specified as a function of the
difference between the EU market price and the world market price. Preferential export
supply functions are based on a literature review and plausibility considerations. The
motivation for this adjustment is evidence from empirical studies, expecting that preferential
imports will vary heavily depending on the EU price level. As it is now intended to use ESIM
for the simulation of heavy changes in sugar market policies, this must be reflected in the
model.
It was not considered necessary to implement new policy instruments in ESIM for depicting
the reform of the CMO for sugar. Private storage aid is not modeled. Instead, the reference
price is modeled like the former intervention price, i.e. as a minimum domestic price level.
Any excess of EU supply and preferential imports over EU demand after full phasing out of
export subsidies will thus go into domestic stocks, as no export possibilities will exist
anymore. Also quota trade among member states (up to 1 Mill. t to be bought by today’s Csugar producers) is not modeled, but enters the model as an exogenous change of member
state specific quota levels. Finally, restructuring aid as it will be paid over a period of four
years is not modeled. Behavioral supply parameters in ESIM reflect medium to long term
adjustments. ESIM is therefore not able to depict the short term adjustments in the sugar
sector which will be heavily impacted by the design of the restructuring funds. Therefore, a
depiction of this short term instrument was not intended.
This documentation is structured as follows. After this introduction, the establishment of
country-specific supply curves for sugar for all EU-27 countries is described in Chapter 2.
Chapter 3 then describes the generation of a preferential export supply function. Finally,
Chapter 4 describes how the changes are implemented in the GAMS model code.
2.7.2 Individual Sugar Supply Curves for EU-27 Member States
On the basis of FADN production cost data and a literature review, shadow prices of sugar
production in EU member states, i.e. the price below which sugar production will start to fall,
the price level at which individual member states will cease sugar production, and the price
responsiveness of sugar production in the range between both were established.
41
2.7.2.1
Shape of the Sugar Beet Supply Function
FADN data provides information about total cost at the level of farm groups, which means it
is not possible to directly derive product specific costs such as for sugar beet. Various
approaches exist to attribute the total costs of a farm group to sugar production in that farm
group. The most common of those is the ARACOST program (European Commission, 1999),
which was also used to calculate the sugar production cost data for 2004 provided for this
project by DG AGRI. In ARACOST, specific costs, farming overheads, depreciation and
external factor costs are attributed to each crop according to its share in total market revenue
(in the terminology of FADN these are referred to as “output”) of the farm or its cropping
branch. Besides its obvious lack of accuracy this approach involves a number of drawbacks
which finally lead to the decision not to apply it for this project. First, this approach leads to a
unique profit rate for all crop products, which leaves no space for a quota rent. Second,
FADN revenue data does not include area payments. Therefore, ARACOST would tend to
overestimate the costs for sugar, just as the revenue share of Grandes Cultures will tend to be
underestimated by that approach. Finally, opportunity costs of internal factors, i.e. land, labor
and capital which are in the property of the farm owner, are not accounted for.
A second approach which is applied by the German Federal Research Center for Agriculture
(FAL) uses detailed information on regional input coefficients for all relevant agricultural
activities in a GAMS algorithm to estimate production costs for different crops. The results of
this approach, however, only exist for a very limited number of EU Member States. For this
project, the FAL approach was not applicable as information on input coefficients was not
available and the amount of work to collect such information would have exceeded the scope
of this project.
Third, a modified ARACOST approach was tried, in which remuneration for farm owned
factors was assumed at the level of the national average cost for external land and labor and a
flat rate interest rate of 2.5% was applied. This approach, applied with FADN Data for 2002
and 2003 provided by LEI, lead, however, to costs exceeding the revenue by large
percentages for virtually all farm groups in the sample.
As all those approaches involved unacceptable disadvantages, finally a pragmatic approach
has been chosen: FADN data is only used to provide information about the distribution of
profitability and thus efficiency within the sector. For each farm group, all costs are added
and divided by total revenues including area payments and other subsidies. To solve the
problem of accounting for opportunity cost of factors of production owned by the farm, in a
first step, factor prices are calculated as described above for the modified ARACOST
42
program10. In the following step, these factor prices are scaled such that at a member state and
sectoral level all costs, including opportunity cost for own factors, are equal to total revenues.
This assumption corresponds to a long term equilibrium in which farmers on average make no
profit except the remuneration of their family owned factors.
Except for Sweden and Finland, this approach led to reasonable results for factor prices in all
EU-15 countries.11 The ratio between total costs and total revenues is then calculated for those
farms producing sugar for the years 2002 and 2003 and data is merged in one data set for both
years12 and the production is scaled such that the total of all farms in the sample adds up to
one. Based on this data set, the following supply curve is estimated for each member state:
Supply  MAX 0,    * price
i
i

 ,

with “price” being the cost ratio. The cost curves built from the data sets for 2002, 2003, and
the total data set for 2002/2003 as well as the estimated supply curve are, as an example,
shown for France in Figure 2.6.
Figure 2.6: Cost Data and Estimated Sugar Beet Supply Function for France
Source: Own calculations.
10
Due to distorted FADN data for the Netherlands, the land leasing rate is set at € 350 per hectare in this
member state (OFFERMANN, 2006).
11
For Sweden and Finland, already the farm expenditures exceed the revenues on average in the agricultural
sector. To fulfil the zero profit condition, the prices for family owned factors thus would have to be negative.
Member States which acceded to the EU in 2004 and 2007 were not included in the dataset.
12
As only ratios between costs and revenues and not absolute values are calculated, no comparability issues
due to inflation arise.
43
To obtain information about the distribution of efficiency among farms in Sweden and
Finland as well as in the seven out of ten member states which acceded in 2004 which
produce sugar13, cost data which has been provided by DG AGRI based on the ARACOST
program is applied. This is justified by the fact that for France, the curve generated from this
data shows only minor differences to the curve generated from the FADN data for 2002 and
2003 with the approach described above. For Romania, Bulgaria and Turkey, no FADN Data
is available. The structure of the agricultural sector is rather fragmented, and the elasticity of
supply is small. Therefore, the distribution of efficiency for Slovenia is chosen also assumed
to be for these countries, which leads to a supply curve which has in the observed point an
own price elasticity of 0.92 for sugar beet supply.
2.7.2.2
2.7.2.2.1
Level of the Sugar Beet Supply Function
Shadow Prices for Sugar
Having estimated the shape of the cost curves based on FADN data, they are scaled such that
they meet the quantity (A-, B- and C-sugar) - shadow price combination that is observed or
extracted from literature, respectively. As a source for the shadow price for sugar
COMMISSION OF THE EUROPEAN COMMUNITIES (2005) has been used as the most recent data
which is mostly in line with other sources (COMMISSION OF THE EUROPEAN COMMUNITIES,
2003; COUNCIL OF THE EUROPEAN UNION, 2004; ADENAEUER, 2005) and the literature cited
there. For the new member states which joined in 2004 and also for Romania, Bulgaria and
Turkey, shadow prices which were/are domestic prices in these countries before their
accession have been taken from the ESIM 2004 database. Table 2.4 shows shadow prices for
white sugar in the EU 15 member states that are applied in the generation of supply curves,
resulting supply elasticities for sugar beet and the intercept on the price axis, at which sugar
production in a country ceases completely.
Table 2.4: Shadow Prices, Intercepts and Supply Elasticities for Sugar Supply Functions
in the EU-15 in € per ton (2002 – 2005)
Member State
Shadow Price Intercept
Supply
Elasticity a
Austria
400
285
1.6
France
400
289
2.0
Denmark
450
298
2.1
Belgium and Luxemburg
400
298
1.4
Germany
400
298
1.8
Sweden
400
242
1.2
13
All except for Malta, Cyprus and Estonia
44
Ireland
550
422
1.8
United Kingdom
400
260
1.8
Spain
550
357
1.2
Netherlands
400
290
1.4
Finland
550
307
1.1
Italy
550
346
0.7
Greece
550
373
1.1
Portugal
550
359
1.1
a
Source: Commission of the European Communities (2005). As mentioned in the text, the elasticities presented
in this table are those for beet supply, not for white sugar supply.
2.7.2.2.2
The Processing Margin between Sugar Beet and White Sugar
To derive costs for the production of sugar beet from those for white sugar, processing costs
have to be deducted. Data on processing costs (including transportation of beets and corrected
for the value of molasses) of European sugar plants is rarely available. Some studies reporting
rough estimates are WITZKE AND KUHN (2003) and NEI (2000). Both studies present figures
that are below the spread of € 265.2 per ton of white sugar, which is applied by the European
Commission as the spread between the intervention price and the beet costs which is used to
calculate the minimum beet price for farmers (as reported in NEI, 2000):
Processing costs
Value of molasses
Transport of beets
Spread
243.6 €/t
-22.5 €/t
+44.1 €/t
265.2 €/t
The most recent estimate of processing costs per ton of white sugar, for which the term
processing margin is used in this study, is provided by WITZKE AND KUHN (2003), at a level
of € 175. Corrected for the value of molasses and the transport costs of beet, this figure
increases to € 196.6 per ton. COMMISSION OF THE EUROPEAN COMMUNITIES (2005) presents
differences in processing costs among the EU-15 member states. Correcting the average
processing margin for these differences national processing margins are calculated and
presented in Table 2.5. For the new member states, no information about processing costs was
available. Therefore the average of the old member states is applied.14
14
Using a higher margin than average as might appear realistic would have resulted in extremely low costs of
sugar beet production in some countries, e.g. in Poland.
45
Table 2.5: Processing Margins for the EU-15, €/ton White Sugar (2003)
Member State
Processing Margin
Austria
159
France
128
Denmark
147
Belgium and Lux.
215
Germany
172
Sweden
134
Ireland
265
United Kingdom
85
Spain
221
Netherlands
190
Finland
190
Italy
333
Greece
271
Portugal
246
Sources: NEI (2000), Commission of the European Communities (2005), WITZKE and KUHN (2003), own
Calculations.
Assuming constant costs for processing, the processing margins have to be deducted from the
shadow price for white sugar to obtain the shadow price for sugar beet expressed in white
sugar equivalents (WSE). After scaling the estimated sugar beet cost curves to meet this
shadow price at the observed quantity, there is one step left to obtain a final supply curve for
sugar as a variable of the internal market price in the EU, which is to incorporate the
processing margin.This is done by simply shifting the supply curve by the processing margin
to meet the price for white sugar instead of that for sugar beet. As an example the supply
curve for white sugar and the supply curve for sugar beet as a function of the sugar price and
the price for sugar beet in WSE are shown for France in Figure 2.7.
46
Figure 2.7: Sugar and Beet Supply Curves for France
Source: Own graph.
In the ESIM 2004 version, the processing margin was accounted for in the model by a relative
wholesale margin of 136% of the producer price for beet in white sugar equivalents. By using
a relative margin, one assumes a supply elasticity of processing services which is equal to the
elasticity of supply of sugar beet. By using an absolute margin, one assumes an elasticity of
scale of unity for sugar factories. For the purpose of this study, an elasticity of scale of unity
seems the more realistic choice. By applying a constant processing margin, however, one
assumes implicitly that the distribution of quota rents is fixed. Alternatively, one could have
used a processing function depending on the processing margin in absolute terms.
2.7.3 The Export Supply Function for Preferential EU Sugar Imports
2.7.3.1
Introduction
The EU applies a number of preferential import schemes for sugar mostly from developing
countries. The most important in terms of trade volume is currently the sugar protocol which
is attached to the treaty of Cotonou with some ACP countries which are entitled to deliver in
total 1.3 million tons of sugar (expressed in WSE) duty free to the EU market. Similar
agreements are in place with some Balkan countries and India with a total quota of 193
thousand tons and 10 thousand tons, respectively. To utilize fully the refining capacities of the
EU sugar sector an annual quota of roughly 200-300 thousand tons for so-called special
47
preferential sugar (SPS) has been opened which is filled by ACP countries and India. After
accession of Finland in 1995 the EU had to comply with Finland’s current access
commitments in the (World Trade Organization) WTO, which meant that TRQs for roughly
80 thousand tons (WSE) of raw sugar originating mainly in Cuba and Brazil were opened15.
The most recent development in terms of preferential access for developing countries to the
EU market is the Everything But Arms initiative, which extends the EU’s Generalized System
of Preferences (GSP) for the group of Least Developed Countries to duty and quota free
access for all products (but arms) originating in these countries to the EU market. For sugar,
as well as for rice and bananas, a phase-in period is decided upon. A zero-duty tariff rate
quota (TRQ) is opened and annually increased for raw sugar, until from July 2009 on, the
access for sugar, be it raw or white is unrestricted. The total quota for imports under SPS is
fixed annually on the basis of the supply needs of the refineries in the EU. Therefore, one can
assume that the imports under SPS will fall at the same pace as those under EBA will increase
until they are completely phased-in.
In the 2004 version of ESIM, preferential imports are a parameter, which means the same
quantity will always be delivered to the EU market, no matter how prices in the EU develop.
After the 2006 reform of the CMO for sugar this is not anymore realistic, as many of the
suppliers will probably decrease or cease their shipments of sugar to the EU if prices undercut
a certain level in future. In the new ESIM version it is, therefore, decided to let the quantity of
imports vary with the price in the EU. Another important issue is to account for the
abolishment of any restrictions on LDC sugar imports after 2009. For that purpose it was not
only necessary to make imports price responsive, but also to introduce a different export
supply function after 2009. Finally, technical progress had to be accounted for. With technical
progress in place, it is realistic to assume that at a given price or given margin between EU
and world market price, the export supply will increase each year.
2.7.3.2
Choice of Explaining Variable
The price difference between EU price and world market price is chosen as the explaining
variable for export supply rather than the absolute EU price and this section justifies this
choice. For those preferential exporters not being competitive at world market conditions
which cover most of today's preferential supply to the EU, the EU market price clearly is the
relevant variable. If it falls below the production and transport costs of the marginal supplier
in that country, that supplier will drop out of the market. The world market price differential
as it shall be used in the new version of ESIM is the relevant variable only for those countries
which are competitive at world market conditions. Those countries cover only a small share
15
Commonly referred to as CXL quota.
48
of today's preferential exports to the EU, but with future prices in the EU declining, this share
will increase. For those countries the choice of the explaining variable will make much of a
difference whereas the difference will probably be minor for those countries which are not
competitive at world market prices.
It is not the pure difference of EU and world market price which is applied as the explaining
variable of the export supply function, but a correction is made for transport and other
transaction costs involved in supplying the EU market. The supplier in question will compare
ex-factory or f.o.b. prices faced in different markets. Most of these markets will
geographically be closer to the supplier than Europe, implying lower transportation costs.
Additionally, the supplier may face transaction costs to supply the EU market, e.g. those
related to proof of origin or to “swapping” domestic sugar with imported sugar from third
countries. In total this difference is assumed to be at 20 real 2002 €.
2.7.3.3
2.7.3.3.1
Estimating Price Responsiveness of Preferential Suppliers to the EU Market
ACP Countries and India
The ACP sugar protocol16 entitles currently 21 ACP countries (2 of which have a quota of
zero) to deliver individual quota-restricted amounts of sugar to the EU market. Together with
a quota open for India this amounts to roughly 1.3 million tons WSE. For each of the
countries and based on shadow prices taken from Isermeyer et al. (2005) and plausibility
considerations an export supply curve to the EU market is constructed. The Export supply
elasticity is assumed to be unity for all countries. An annual technical progress component17 is
assumed to be 2%. I.e. the export quantity to the EU market at a given real difference between
the EU and the world market price minus a transport and transaction cost differential, of each
country increases by 2% annually.
2.7.3.3.2
LDCs
The assumptions made on export supply of LDCs to the EU sugar market are mainly based on
a study of LMC (2004). The study forecasts export supply of LDCs to the EU sugar market
under different reform scenarios for the sugar CMO. The reference scenario in this study is a
market equilibrium in the EU without any production restriction in place, and all current
preferential trading partners (except for Cuba and Brazil) being allowed to export freely to the
16
Commission Regulation 2006/180.
17
This is not equal to the rate of technical progress for total production of a country, not all of which is
exported to the EU. The technical progress component in the export supply function is, therefore, assumed to
be higher than the rate of technical progress which is 1.6% in ESIM for the production of sugar in the rest of
the world (Banse et al., 2005).
49
EU market. The price the study assumes for this scenario in 2015 is 400 € per ton, which is
quite close to the reference price level finally decided upon in the reform of the CMO sugar in
2006. The results of this scenario are, therefore, used to build an export supply function in
ESIM. The LMC study assumes that three years before the final opening of the EU market,
i.e. in 2006 the sugar sectors of those countries which are on a sustainable basis able to
deliver the EU market at the assumed price18 will start to increase sugar production capacities.
Table 2.6 illustrates which variables19 that were used to calculate the export supply of the
LDCs in question could be extracted from the LMC study. The remaining variables, which are
shaded in grey in the table had to be calculated by using assumptions about technical
progress, capacity increases and the development of domestic and regional marketed
quantities.
Table 2.6: Variables of LDC sugar sectors
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 later
Production
LMC
LMC
Exports to EU
LMC
Residual
Source: Own compilation.
The calculated export supply levels correspond to an EU price of sugar of 400 real 2015 € per
ton WSE and a world market price for sugar of 150 real 2015 €. Applying the above
mentioned approach of using the differential between the EU and the world market price
minus a fixed amount for transaction costs and transport cost differential as the explaining
variable of the export supply function, one ends up with roughly 225 real 2015 € as the
relevant price differential variable. As ESIM works with 2002 real € this is deflated assuming
an average inflation rate of 2.11%20, resulting in a price differential to which the export
supply function is calibrated of roughly € 175 in 2002. As investments which are usually
reflected in a supply elasticity are already accounted for during the estimation of the intercept
parameter of the supply function, the export supply elasticity is fixed at the rather small value
of 0.3. To illustrate how different elasticities of export supply affect the level of exports two
18
These are Ethiopia, Malawi, Sudan and Mozambique. Unlike most other sources, LMC does not expect
Zambia to be a major supplier under EBA, which they attribute to the landlockedness and the resulting high
transportation costs.
19
Out of the variables in the table, the exports to the EU are eventually the only relevant one for building an
export supply curve.
20
In the ESIM 2004 version the base period is 2000-2002. The average Euro zone inflation rate applied there
and extracted from IMF (2003) has been applied for the calculation here.
50
curves with the same level of quantity and shadow price are shown in Figure 2.8. The curve
with the high elasticity, which corresponds to the ACP countries which are mainly high cost
producers, falls quickly even with minor price decreases, whereas the curve with the lower
elasticity does not react so strongly. Both curves, however, fall eventually to zero when the
price differential approaches zero.
Figure 2.8 Shape of Isoelastic Supply Curves with Different Elasticities
Source: Own graph.
2.7.3.3.3
CXL, Balkan and SPS
The supply under the CXL quota of 80 thousand tons is not assumed to be price responsive.
The supply under the quota for the Balkans countries of 193 thousand tons is calibrated at a
shadow price of € 400 per ton WSE with an export supply elasticity of unity. The supply
under the quotas for SPS is also not assumed to be price responsive. Following the reasoning
above, the amount is reduced annually by the increase in the EBA quota and abolished after
2009.
2.7.3.4
Estimation of the Annual Aggregate Preferential Export Supply Curve
Having built export supply curves for each individual LDC and ACP country for each
individual year, and an aggregate one for the Balkans countries for each year from 2003 to
2030, price differentials are inserted in each function, which is additionally restricted by the
individual quota. By adding the values up over all countries one obtains the aggregate
preferential export supply for any price in the respective year. In the range between zero,
which corresponds to a situation where the EU price is at world market level plus transaction
and transport cost differential, and € 460 which corresponds to a situation of an internal EU
51
price of € 730 and a world market price of € 250, an isoelastic export supply curve is
estimated for each year from 2003 to 2030.
2.7.4 Implementation of Changes in the Model Code
2.7.4.1
Individual Sugar Supply Curves for EU-27 member states
The supply curves for sugar generated based on the approach explained in Section 2 first had
the form

 MAX 0,     * price  .
 
i
i 

The supply functions for other products in the model, however, are split into yield and area
(1)
Supply
functions. Both are not only functions of the own price but also of prices of factors of
production and cross products (the area allocation function only). They come along in the
following shape:
(2)
Supply  Area *Yield
(3)
Yield   * PP *  Pind 
(4)
i
i
i
yi
i
i
i
yif
f
f
Area   *  PI *  Pind 
ij
i
i
j
j
f
f
if
,
with
PPi = producer own price
PPj = producer prices of all crop products (including i)
PIj = incentive price of all crop products
Pindf = price index for factor f
εij = own and cross price elasticities of area allocation
εif = elasticity of area allocation with respect to factor prices.
εyi = own price elasticity of yield
εyif = elasticity of yield with respect to factor prices.
In order to transform (1) into a set of separate yield and area allocation functions, (3) is used
as the yield equation also in the new ESIM version. The supply functions (1) for sugar are
divided by a term of the form
(5)
 * PP
i
i
yi
,
which is calibrated to meet the yield level of the base period. The supply function has now
been converted in an area allocation function:
52
(6)
Area
i


MAX 0,    * PPi
 i *PP


 yi
i
Multiplied by the unchanged yield function, the base data is met, and the responsiveness of
yield and, hence, production to factor prices is retained. The area function is, so far, a function
of the own price solely. To incorporate responsiveness to prices of cross products and factors,
the old area allocation function is used. The own price and the own price elasticity are
removed from the equation and the intercept α (Φ in (7)) is recalculated such that the
remaining term is one with prices of the base period. This term is than multiplied with the
new area allocation function. As a result, the new area allocation function for sugar in ESIM
is:
(7)
Areai 

MAX 0,    * PPi
 i *PP
 yi

 *
 *  PP *  Pind 
ij
i
j i
j
f
if
f
i
The shadow price equation is also modified accordingly such that at the shadow price, exactly
the quota level is produced.
2.7.4.2
Price Responsive Preferential Export Supply Function for Sugar
In the model, the preferential imports are not a variable of interest themselves. They are
interesting from a point of view of domestic EU price formation, as they determine the net
trade situation, i.e. the position of the logistic function of price formation. The preferential
imports appear, therefore, only in the equation where the share of the preferential import
(formerly TRQ, the name of the equation is, thus, “TRQSHR_EU”) in total domestic use is
calculated. While this was previously a fixed parameter for preferential imports which was
divided by domestic use, it is now the preferential export supply function which is divided by
total use. The export supply function is:
(8)
Exports

 MIN TRQ ,  * MAX 0, PDUK  PW  tc 


where PDUK is the domestic price for sugar in the UK, PW is the world market price for sugar
and tc is the parameter for the transaction and transport cost differential. To represent the EU
price, the price of the UK has been chosen, as currently most of the preferentially imported
sugar enters the EU via the UK. The price level of sugar in the EU is not unique, as
institutional prices are deflated by individual inflation rates
53
2.8
Stochastic elements in the model analysis21
2.8.1 Overview
ESIM in its standard version ESIM is a deterministic equilibrium model. Equilibria result for
each projection year and must therefore be understood as point estimates or estimates of the
average values of dependent variables in case of exogenous variables taking their average
value – independent from their distribution.
This is appropriate in order to simulate “average developments” on agricultural markets for
the future. But for some variables the tails of the distribution of endogenous variables, such as
supply or trade quantities may matter from a policy perspective: for example for keeping
within WTO restrictions of policies regarding public or private storage of commodities.
The most important variable which has a strong stochastic element is the yield of crops, i.e.
the weather impact on yields. Global warming may even increase the variability of crop yields
in the future. Particularly the southern and eastern parts of the EU see already important
variation of crop production. This impacts prices, stocks and other important market variables
such as trade between EU regions. Present policies as well as possible other policies directly
respond to these variations, e.g. in terms of public stocks and budgetary spending.
Therefore stochastic terms were included in the yield functions of ESIM for the EU key crops
wheat, barley and rapeseed. Two options are implemented for running ESIM stochastically: A
Monte Carlo (MC) approach, which is demanding in terms of computational capacity (solving
ESIM 1000 times for the year 2015 drawing from a given distribution of stochastic terms
takes about 6 hours), and an approach based on the Gaussian Quadrature (GQ) which saves on
computational capacity (to approach the results of a large number of MC draws, 86 solves are
required, taking about 50 minutes).
If ESIM is run in this stochastic version, solution variables are expressed as i) value of the
deterministic version, ii) expected value of the stochastic version, iii) variance of the
stochastic version, and iv) standard deviation of the stochastic version.
2.8.2 Methodology
Based on time series data for EU-27 member states as well as the rest of the world linear yield
trends were estimated and de-trended22 time series served as a basis for estimating the
distribution of the stochastic term of the yield equation. Therefore data for the period 1993 to
2005 were taken for all countries. Also the correlation between error terms in yields of the
considered crops as well as member states were tested and incorporated in the estimation of a
21
This section draws on Artavia et al. (2008).
22
Deviations from a linear trend estimated by least-square regressions.
54
variance-covariance-matrix. Based on correlation, countries have been grouped. In order to
save upon computational capacity and allow for a faster generation of results, the stochastic
model formulation is not only based on MC, but also on the Gaussian Quadrature (Arndt,
1996).
2.8.3 Grouping of Countries
The correlation matrix of estimated yield data, taken from FAOSTAT for the time period
1993 to 2005 and crops was inferred in order to group countries with the highest yield
correlations.
Based on the correlations the EU 27 countries were divided into 15 groups with France,
Germany and Poland each representing a separate group as they have been identified as being
the most important producers in the EU of wheat, barley and rapeseed. Other “stand alone”
countries such as Finland, Sweden, Italy, Denmark and Greece were also not grouped as the
correlation with other country data was rather low. Due to its small production quantities
Ireland was grouped with the UK. The Baltic countries build one group as well as Malta with
Cyprus and Bulgaria with Romania. Due to very high correlation Austria was grouped with
the Czech Republic, Slovakia, Slovenia and Hungary. Portugal and Spain were grouped, and
due to neglectable production quantities Belgium and Luxemburg build one group with the
Netherlands.
2.8.4 Generation of Stochastic Terms θ
For each country group yield y was calculated for each crop for the period of 1993 to 2005 by
dividing total production by total area, delivering 43 stochastic variables.23 The de-trended
stochastic variables θ were derived by dividing the observed yield ^y y by the estimated trend
^y
^y y/^y=θ .
In order to obtain a variance –covariance matrix Σ, the software SPSS was used. As Σ is not
diagonal, it is decomposed via the Cholesky LDLt factorisation, for the GQ version
accounting for correlation (ARNDT 1996):
Σ = LDLt
Where L is a lower triangular matrix, D a diagonal Matrix and Lt is upper triangular matrix
(transposed L).
23
As Malta and Cyprus do not produce rapeseed they were neglected and not included. Hence instead of 45x45
matrix (15 groups á 3 crops), the matrix is only 43x43.
55
2.8.5 Gaussian Quadrature and Numerical Integration
Using Gaussian quadrature points is a convenient method to approximate multivariate
integrals accurately while requiring a very limited number of evaluations of the integrand as
compared to Monte Carlo method (ARNDT 1996).
To obtain the Gaussian quadrature points, we used method developed by Stroud (1957) for
drawing order three quadratures for symmetric distributions of mean zero and standard
deviation one.24 This method permits a systematic sensitivity analysis of the 43 stochastic
variables using only 2n solves of the model.
Let Γk (γk1, γk2, …γk43) be the k quadrature point (k= 1,2,…..2n). Let r= 1,2….[n/2], where
[n/2] denotes the greatest integer not exceeding n/2. One r point is used for two n points, e.g.
for n= 1 and n= 2 r=1.
The points were derived by the following Stroud formulas:
 ( 2r  1)k
 2 r 1  2 cos
n


 ( 2r  1)k
  2 r  2 sin
n





As in this case the last point of n is odd, γk43 = (-1)k , see ARNDT (1996).
The following formula was used to derive the quadrature points which were than fed into
ESIM:
Φ = μ + Γ√D.
2.8.6 Implementation of Quadrature Points in ESIM
For a specified year the model runs 86 times over each of the 43 quadrature points and the
solution variables are saved after each solve and solution variables are expressed as i) value of
the deterministic version, ii) expected value of the stochastic version, iii) variance of the
stochastic version, and iv) standard deviation of the stochastic version.
This feature enables a “risk assessment” of certain CAP instruments, such as the probability
that the EU would exceed/not exceed the WTO commitments on export subsidies, instead of
only knowing whether the EU will do so on average.
24
These assumptions are reflected in the data. For further details see also (STROUD 1957 as cited in ARNDT
1996).
56
3
Behavioural Parameters
4
Base Data for Model Calibration
4.1
Quantities
4.2
Prices and Policies
5
References
Adenaeuer, M. and Heckelei T. (2005), Economic Incentives To Supply Sugar Beets In
Europe. In: Arfini, F. (ed.) (2005), Modeling Agricultural Policies: State of the Art and
New Challenges, Proceedings of the 89th European Seminar of the EAAE, February 3-5,
Parma.
Arndt, C. (1996), An Introduction to Systematic Sensitivity Analysis via Gaussian
Quadrature. GTAP Technical Paper No. 2
Artavia, M., T. Möller and H. Grethe (2008), Including Correlated Stochastic Terms in ESIM.
Draft Final Deliverable to the European Commission, August.
Banse, M., Grethe, H. and S. Nolte (2005a), European Simulation Model (ESIM) in GAMS:
Model Documentation. European Commission, DG AGRI.
Banse, M., Grethe, H. and S. Nolte (2005b), European Simulation Model (ESIM) in GAMS:
User Handbook. European Commission, DG AGRI.
Banse, M., Grethe, H. and S. Nolte (2007), European Simulation Model (ESIM): User
Manual.
Banse, M. and H. Grethe (2008), Effects of the New Biofuel Directive on EU Land Use and
Agricultural Markets. Seminar paper at the EAAE-Seminar Modelling Agricultural and
Rural Development Policies, January 31 – February 1, 2008, Seville, Spain.
Brook et al. (1998), GAMS – A User's Guide.
Commission of the European Communities (2003), Reforming the European Union’s sugar
policy: Summary of Impact Assessment Work. Commission Staff Working Document. COM
(2003)554 final. Brussels, 29 March 2003. SEC(2003) 1022.
Commission of the European Communities (2005), Reforming the European Union’s sugar
policy: Update of Impact Assessment [SEC(2003) 1022]. Commission Staff Working
Document. COM (2005)263 final. Brussels, 22 May.
Council of the European Union (2004), Appendum to the Report Communication from the
Commission to the Council and the European Parliament accomplishing a sustainable
Agricultural model for Europe through the reformed CAP – sugar sector reform. 6
Background Notes. Brussels, 30 September 2004
Dol, W. and F. Bouma (2003), GAMS Simulation Environment.
57
European Commission (1999), ARACOST: A Program for estimating cost of production of
arable crops. Brussels: DG AGRI, March.
European Commission (2007), Agriculture in the European Union - Statistical and economic
information 2005. Table 3.3.9: Rents for agricultural land. Brussels.
EUROSTAT (2008), Eurostat data base on land use.
FAO (2004), FAOSTAT database. http://apps.fao.org/cgi-bin/nph-db.pl?subset=agriculture.
Grethe, H. (2004), Effects of Including Agricultural Products in the Customs Union between
Turkey and the EU. A Partial Equilibrium Analysis for Turkey. CeGE-Schriften, Center for
Globalization and Europeanization of the Economy, Georg-August-Universität Göttingen,
No. 9. Peter Lang Verlag, Frankfurt am Main. Also published at
http://webdoc.sub.gwdg.de/diss/2004/grethe/index.html.
Grethe, H., Nolte, S. and M. Banse (2008), Modelling the Effects of EU Sugar Market
Liberalization on Area Allocation, Production and Trade. Seminar paper at the EAAESeminar Modelling Agricultural and Rural Development Policies, January 31 – February 1,
2008, Seville, Spain.
IMF (2003) International Financial Statistics Yearbook 2002.
IMF (2003), International Financial Statistics Yearbook 2002. Washington D.C.
Isermeyer, F., Gocht, A., Kleinhanß, W., Küpker, B., Offermann, F., Osterburg, B., Riedel, J.
and U. Sommer (2005), Vergleichende Analyse verschiedener Vorschläge zur Reform der
Zuckermarktordung. Landbauforschung Völkenrode Sonderheft 282. FAL, Braunschweig,
April.
Kirchgessner, M. (1982), Tierernährung.
Latruffe, L. and C. Le Mouel (2006), Description of agricultural land market functioning in
partner countries. Working paper, IDEMA project, Rennes and Wye college.
Lillard, P.P. et al. (1995), A Documentation of the Revised European Simulation (ESIM)
Modelling Framework. USDA/ERS. Washington.
LMC International (2004), EU Sugar Reform: The Implications for the Development of LDCs.
Oxford.
Meijl, H. van, Rheenen, T. van, Tabeau, A. and Eickhout, B. (2006), The Impact of Different
Policy Environments on Land Use in Europe. Agriculture, Ecosystems and Environment
114: 21-38.
Menke, K.H. et al.(1987), Tierernährung und Futtermittelheilkunde.
Münch, W. (1995), Possible Implications of an Accession of the Visegrad Countries to the
EU. Can the CAP do without Reform? Paper presented at the Agricultural Economic
Society One-Day Conference. London, 13. Dezember 1995.
58
Münch, W. (2000), Agricultural Implications of CEC–EU accession on Agricultural Markets.
In: Tangermann, S. und M. Banse (eds.), Central and Eastern European Agriculture in an
Expanding European Union. CABI International. Wallingford. pp. 112-132.
Münch, W. (2002), Effects of EU Enlargement to the Central European Countries on
Agricultural Markets. CEGE-Schriften Band 4, Center for Globalization and
Europeanization of the Economy, Georg-August-Universität Göttingen.
NEI (2000), Evaluation of the Common Organisation of the Markets in the Sugar Sector,
NEI, Rotterdam.
Nölle, F. (2000), Die Auswirkungen der McSharry-Reform: Quantitative Analyse im Rahmen
eines Marktmodells. Diploma Thesis. Göttingen.Offermann, F. (2006), Oral Information.
FAL, Braunschweig.
State Institute of Statistics (various issues), Agricultural Structure (Production, Price Value).
Tangermann, S. and Josling, T.E. (1994), Pre-Accession Agricultural Policies for Central
Europe and the European Union. Study commissioned by DG I of the European Commission. Brüssel.
Tangermann, S. and W. Münch (1997), Sugar Markets in Central Europe and Eastward
Enlargment of the European Union. Institut für Agrarökonomie der Universität Göttingen,
Diskussionsbeitrag 9705. Göttingen.
Witzke, H.-P. and A. Kuhn (2003), Assessing Reform Options for the Sugar Common Market
Organisation – Quantitative Analyses with Interlinked Models. Contributed Paper at the
43rd annual Gewisola Conference, 29 September – 1 October 2003, Stuttgart.
World
Bank
(2004),
World
http://devdata.worldbank.org/data-query/.
59
Development
Indicators
Database.