JRC53518Institute for Prospective Technological Studies - Update of a Quantitative Tool for Farm...

7/27/2019 JRC53518Institute for Prospective Technological Studies - Update of a Quantitative Tool for Farm Systems Level An

1/94


2/94

Update o aQuantitative Tool orFarm Systems Level

Analysis o AgriculturalPolicies (EU-FARMS)

Author: Alexander Gocht

Editors: Ignacio Prez Domnguez and Adriana

Cristoiu

2010

EUR 24321 EN


3/94

European Commission

Joint Research Centre

Institute or Prospective Technological Studies

Contact information

Address: Edifcio Expo. c/ Inca Garcilaso, 3. E-41092 Seville (Spain)

E-mail: [email protected]

Tel.: +34 954488318

Fax: +34 954488300

http://ipts.jrc.ec.europa.eu

http://www.jrc.ec.europa.eu

Legal Notice

Neither the European Commission nor any person acting on behalf

of the Commission is responsible for the use which might be made of this

publication.

Europe Direct is a service to help you fnd answersto your questions about the European Union

Freephone number (*):00 800 6 7 8 9 10 11

(*) Certain mobile telephone operators do not allow access to 00 800 numbers orthese calls may be billed.

A great deal of additional information on the European Union is

available on the Internet.

It can be accessed through the Europa server

http://europa.eu/

JRC53518

EUR 24321 EN

ISSN 1018-5593

ISBN 978-92-79-15543-7

doi:10.2791/40423

Luxembourg: Publications Ofce o the European Union

European Union, 2010

Reproduction is authorised provided the source is acknowledged

Printed in Spain

The mission o the JRC-IPTS is to provide customer-driven support to the EU policy-making process by

developing science-based responses to policy challenges that have both a socio-economic as well as a

scientifc/technological dimension.


4/94

3

UpdateofaQuantitative

ToolforFarmSystemsLevelA

nalysisofAgriculturalPolicie

s(EU-FARMS)Table o contents

Abbreviations 9

Acknowledgments 11

Executive Summary 13

1 Introduction 17

1.1 General objective o the tender 17

1.2 Organisation o the report 171.3 Background 17

1.4 CAPRI and the arm type layer workow 19

2 Databases used to build up the arm type layer 23

2.1 CAPRI database 23

2.2 The Farm Structure Survey database (EUROFARM) 25

2.3 FADN database 27

2.4 Comparing the degree o consistency between FSS and CAPRI database 27

3 Base year calibration or the arm types 31

3.1 Defnition o arm group dimensions (CAPREG-FSS) 31

3.1.1 Defning the rules or aggregating the FSS arm groups to CAPRI arm types 34

3.1.2 Results o the current selection rule 36

3.2 A comparison between the recovered arm types in CAPRI and their representativeness

in FADN 44

3.3 Consistency o arm types to NUTS II (CAPREG-FARM) 46

3.3.1 Production 46

3.3.2 Yields 493.3.3 Inputs 53

3.3.4 The arm type supply model 57

3.3.5 Selected results or the arm type database 57

4 Trend projection and arm type baseline mode 63

4.1 Introduction 63

4.2 Farm type trend projection 64

4.3 The baseline mode or the arm types 67

5 Conclusion 69


5/94

4

Tableofcontents 6 Manuals Farm Type Model 71

6.1 Data import routine (FSS to CAPRI) 71

6.2 Data import routine (FADN to CAPRI) 72

6.3 Frm2tables 72

7 Reerences 75

8 Annexes 77

8.1 Final request document or FSS aggregation (to Eurostat) 77

8.2 Overview mapping o FSS and CAPRI code or crops 80

8.3 Selected FSS variables or the data request 82

8.4 Calculation o relative shares or calculating type o arming 84

8.5 Comparison on o animal production categories between CAPRI and FSS presented

or the Top 100 NUTS II regions 86


6/94

5




s(EU-FARMS)

List o Tables

Table 1: Data items and their main sources in COCO 25

Table 2: Ofcial data availability in REGIO 26

Table 3: Sampling rate o FSS and overview o the used FSS database per MS 26

Table 4: Comparison between FSS and CAPRI animal production activities on MS level 29

Table 5: Type o arming groups in CAPRI 32

Table 6: ESU groups selected or the data request and the fnal CAPRI arm types 32

Table 7: Overview o the selection rule or the CAPRI arm types based on the FSS arm

group statistics or Denmark 33

Table 8: General overview o arm types selected or the CAPRI layer 36

Table 9: Number o CAPRI arm types without a representation in FADN 44

Table 10: UAA o CAPRI arm types without representation in FADN 45

Table 11: Number o slaughtered pigs (in thousands) or attening without representation in FADN 45Table 12: Farming types and ESU class recovered rom the FSS raw data 48

Table 13: Prior and estimated partial SGMs (P1-P5) or all arm types in Denmark 48

Table 14: Prior and estimated ESU values and UAA o all arm types in Denmark 49

Table 15: Estimates or selected crop activity level in Denmarkk 49

Table 16: Overview o the regional coverage o the current arm type layer and the

representativeness o FADN or the FSS-based CAPRI arm groups evaluated at MS level 50

Table 17: UAA and pig production activities in the Noord-Brabant region 56

Table 18: Farm type nutrient balances in the Noord-Brabant region 57

Table 19: Comparison o the change o rapeseed production between the expected changerom expert data and the recovered change during the trend between 2005 and

2013 (estimated in percent) 66

Table 20: Comparison o the projected change in rapeseed production or NUTS II region in

France 67


7/94

6

Tableofcontents List o Figures

Figure 1: Data sources and relationships in the arm type layer 24

Figure 2: Comparison UAA between FSS and CAPRI or all MS where FSS is available in year 2000 27

Figure 3: Comparison UAA between FSS and CAPRI or all MS where FSS is available in year 2005 28Figure 4: Comparison UAA between FSS and CAPRI or all MS where FSS is available in year 2003 29

Figure 5: Comparison o selected crop activities at NUTS II between the CAPRI database

and aggregated FSS arm type statistics 30

Figure 6: General overview o the distribution on economic size groups across EU-27,

EU-25, EU-15, EU-10 and EU-02 34

Figure 7: Distribution o the number o CAPRI arm types aggregated by type o arming and

compared or EU-15, EU-10 and EU-02 35

Figure 8: Distribution o FADN prior yields and posterior yields or all German arms with

type o arming FT13 and ESU Group three (>100 ESU) (in tons) 51

Figure 9: Distribution o FADN prior and fnal adjusted yields or all arms in Germany with type o

arming FT13 and ESU Group three (greater than 16 and less than 100 ESU) (in tons) 51

Figure 10: Distribution o FADN prior and fnal adjusted yields or all arms in Germany 52

Figure 11: Distribution o FADN prior and fnal adjusted yields or all arms in Germany in the

case o dairy cows (high and low yield) 53

Figure 12: Levels o dairy activities low yield (fg. a) and eed distribution or Denmark and

relevant arm types (in kg eed per head) (fg. b) 55

Figure 13: Income distribution or all cattle activities or the three economic size classes or

specialist dairying arms across the EU-27 in Euro (per heads) 58

Figure 14: Main cropping shares or selected arm types in Denmark 60

Figure 15: Shares o Farm Types or animal categories or all arm types in Denmark 60

Figure 16: The baseline mode or CAPRI overview 63

Figure 17: Hectares o rapeseed (in thousand) or the arm types in NUTS II region Centre

(France) in absolute change (comparison between 2005 and baseline year 2013) 68

Figure 18: Defnition o arm types and arm type aggregates in table.xml and its GUI representation 73

Figure 19: The use o predefned group selections in the case o the arm type layer 74


8/94

7




s(EU-FARMS)

List o Maps

Map 1: Distribution o (a) total UAA, (b) number o holdings (in thousand hectares) and

(c) share on NUTS II UAA in percentage o arm type: specialist cereals, oilseed

and protein crops (FT 13) less than 16 ESU in 66 NUTS II. 37


(c) share on NUTS II UAA in percentage o arm type: specialist cereals, oilseed and

protein crops (FT 13) greater than 16 less than 100 ESU in 101 NUTS II regions. 38


(c) share on NUTS II UAA in percentage o arm type: specialist cereals, oilseed

and protein crops (FT 13) greater 100 ESU in 70 NUTS II regions. 39

Map 4: Average arm size hectares and arm type: specialist cereals, oilseed and protein

crops (FT 13) greater than 100 ESU in 70 NUTS II regions. 40


(c) share on NUTS II UAA in percentage o arm type: specialist dairy (FT 41) less

than 16 ESU (FT41L16) in 42 NUTS II regions. 41


(c) share on NUTS II UAA in percentage o arm type: specialist dairy (FT 41)

greater than 16 less than 100 ESU (FT41GT16L100) in 120 NUTS II regions. 42


(c) share on NUTS II UAA in percentage o arm type: Specialist dairy (FT 41)

greater than 100 ESU (FT41GT100) in 77 NUTS II regions. 43

Map 8: Surplus at soil level in kg nitrogen per hectare or The Netherlands at NUTS II level

in the base year 56

Map 9: Premium per cattle activity or all dairy arm types less than 16 ESU 59Map 10: Supply o bee meet o Specialist cattle-rearing and attening (FT 42) and Cattle-

dairying, rearing and attening combined (FT 43) distributed across the EU-27 and

the dierent economic size classes (a) less than 16 ESU; b) more than 16 and less

100 ESU; c) more than 100 ESU (in thousand tons) 59


9/94

8

Tableofcontents


10/94

9




s(EU-FARMS)

Abbreviations

CAP Common Agricultural Policy

CAPMOD CAPRI simulation tool

CAPRI Common Agricultural Policy Regionalised Impact Modelling System (www.capri-model.org)

CAPRI-RD CAPRI - The Rural Development Dimension

CAPRI-GUI CAPRI - Graphical User Interace

CAPREG CAPRI regionalisation tool

CAPTRD CAPRI trend tool

CMO Common Market Organisation

COCO Complete and Consistent database

DG AGRI Directorate-General or Agriculture and Rural Development

DG ENV Directorate-General EnvironmentEAA Economic Accounts or Agriculture

ESIM European Simulation Model

ESU Economic Size Unit

EU European Union

EUROSTAT Statistical Oce o the European Communities

EU-02 Bulgaria and Romania

EU-10 Member States that joined the European Union on 1 May 2004

EU-15 Member States o the European Union beore 1 May 2004

EU-25 Member States o the European Union beore 1 January 2007

EU-27 European Union ater the enlargement on 1 January 2007FADN Farm Accountancy Data Network

FAO Food and Agriculture Organisation, United Nations

FSS Farm Structure Survey

GAMS General Algebraic Modelling System

GTAP Global Trade Analysis Project

GUI Graphical User Interace

HPD Highest Posterior Density estimator

IPTS Institute or Prospective Technological Studies

LSU Livestock Standard Unit

MS Member State(s)NMS New Member State(s)

NPK Nitrogen, Phosphate and Potassium

NUTS Nomenclature o Territorial Units or Statistics

REGIO Abbreviation or the regional domain at EUROSTAT

TSGM Total Standard Gross Margins

UAA Utilised Agricultural Area

SGM Standard Gross Margin
http://www.capri-model.org/http://dict.leo.org/ende?lp=ende&p=wlqAU.&search=Phosphathttp://dict.leo.org/ende?lp=ende&p=wlqAU.&search=Phosphathttp://www.capri-model.org/


11/94

10


12/94

11




s(EU-FARMS)Acknowledgments

The study could not have been accomplished without the help o several people. The author would

like to thank Sergio Gomez y Paloma or having supported the idea o building up a EU-FARM layer

module in the CAPRI model to contribute to policy impact assessment at arm level, Adriana Cristoiu or

directing the study and Christine Mller (DG AGRI) or having supported the process to obtain access

to the Farm Structure Survey and Farm Accounting Data Network. In addition, the author would like to

acknowledge the work o Pol Marquer rom EUROSTAT who extracted dierent data selections or the

new arm type layer and supported the whole data selection process with his knowledge and experience.

Furthermore the author would like to acknowledge the work o the external reviewer Erling Andersen rom

the Danish Centre or Forest, Landscape and Planning or contributing with ruitul remarks to the nal

report. Furthermore, the author wishes to thank Ignacio Prez Domnguez or his eedback on the moretechnical parts o the report, the organisation o a specic training session or Commission sta at IPTS and

the nal editorial work and to Maria Espinosa and Pavel Ciaian or their comments on the drat versions o

the report.


13/94

12


14/94

13




s(EU-FARMS)Executive Summary

Project objectives

The main objective o the study was to

develop and rene the arm type component o

the agricultural sector model CAPRI (Common

Agricultural Policy Regionalised Impact System) in

order to obtain an operational system able to assess

policy induced changes or dierent arming types

across the Member States (MS) o the European

Union (EU) beore 1 January 2007 (EU-25).

According to the contract terms, the tender

is divided into ve separate tasks. The frst task

is to update and complete the current arm

module database o CAPRI. For that the most

recent available databases should be utilised to

cover at least the EU-25. This updated arm type

database should include detailed inormation

on arm structure or all Nomenclature o

Territorial Units or Statistics (NUTS) II regionsacross the EU-25. Task two aims to develop

a typology and identiy important associated

typical arm types or each NUTS II region. Task

three comprises the incorporation o the arm

types into the existing EU-wide agricultural

sector modelling ramework CAPRI. Task

our discusses the current advantages and

shortcomings o the developed module and

proposes urther developments needed to rene

it. Task fve enhances the graphical exploitationtool or users, enabling the deployment o arm

and spatial typologies or presentation and

interpretation o results.

Hence, the general objective was to

achieve a tool able to analyse policy impacts

at arm level, embedded into the CAPRI

modelling system and thereore take advantage

o common eatures rom CAPRI, such as the

link to a global agricultural sector model and aspatial, global multi-commodity market model

or agricultural products.

Results

There have been ve main achievements

o the project explicitly worth mentioning. The

rst our achievements comprise the regional

coverage and technical implementation and the

last concerns the methodological enhancement

developed during this tender. The frst

achievement is the expansion o the arm type

layer towards the entire EU ater the enlargementon 1 January 2007 (EU-27) using a Farm Structure

Survey (FSS) dataset and replacing the Farm

Accountancy Data Network (FADN) data source

or deriving the arm structure across the EU-27.

The second achievement is that the arm layer

in CAPRI is now attained by breaking down

NUTS II regional agricultural statistics into up

to ten typical arm types using FSS data rom the

Statistical Oce o the European Communities

(EUROSTAT). The thirdoutcome o the tender isthat the complete integration o the arm layer in

the CAPRI modelling structure now allows the

arm type level to be integrated as a standard

level in CAPRI or policy driven scenario analysis.

A ourth methodological achievement is that the

developed estimation approach to integrate FSS

data into the top-down approach o CAPRI keeps

the type o arming and Economic Size Unit

(ESU) group unchanged. The fth achievement

comprises the work invested into the CAPRI -Graphical User Interace (CAPRI - GUI) to manage

the arm layer and the generation o the baseline.

Regarding achievement one: A major

objective o the current tender was to extend

the arm typology or CAPRI towards all New

Member States (NMS). The new releases o

FADN data, obtained or this tender, included

the EU-25. However, due to the decision that

the FADN approach should be replaced bythe more appropriate databases rom the Farm

Structural Survey (FSS) it was possible to go


15/94

14

Execu

tiveSummary beyond the coverage o the EU-25 and include

Romania and Bulgaria in the current arm type

version on CAPRI.

Regarding achievements two and three: There

are two important aspects o linking the FSS data

with the regional agricultural statistics. The rst

one includes the denition and population o the

typology. For the tender at hand, the general arm

typology is dened by arm size (three classes)

measured in ESU and by arm specialisation (14

dierent specialisations) according to Standard

Gross Margin (SGM) denitions. For each region,

a subset o this typology (region specic typology)

is dened by selecting the nine most importanttypes by specialisation and size class, using

cropped area and livestock units as weights.

Such a method highlights the most important

arm types in each region in the analysis, whereas

the remaining types are aggregated into a residual

category. Restricted representation o the FADN

could be overcome by using FSS data which

successully upgraded the redistributing aggregation

method o the old version o the arm typeapproach. The computational demanding approach

or redistributing the aggregation weights is now

removed and the similar regional concepts in CAPRI

and FSS (based on NUTS II) allowed a seamless

integration o the arm type selection approach

into the CAPRI core environment. A possible

reclassication o the economic size and arm

typology based on the FSS aggregation is, thereore,

now possible using the standard CAPRI-GUI.

As the production programme o the

arm types is derived rom the individual arm

observations at the regional level, the diversity

o the European arm inside a certain size

and specialisation class across regions is well

presented. Equally, the approach rerains rom

subjective denition o arm programmes. The

routine is sel adjusting in the case o an update

to a newer sample o FSS data.

Regarding achievement our: The second

important aspect o the link between FSS

and regional CAPRI statistics is a consistent

aggregation rom typical arms (arm groups) to

NUTS II regions and rom there to higher regional

levels such as MS, the EU or global agricultural

markets under the condition that the type o

arming and ESU groups remain unchanged.

The respective algorithm simultaneously adjusts

the production programme to ensure a typical

arm sample that is ully consistent with regional

statistics regarding production quantities, activity

levels and input use. In that process, the type o

arming and the ESU class remained untouched

and the majority o the changes are concentrated

in the residual arm type category. The resulting

dataset o arm types inside the NUTS II regionsallows or mutually compatible policy impact

analysis across scales, both or economic, social

and environmental indicators. For example, it

is possible to track down changes in total gas

emissions or arm income rom the EU level to

regions and specic arm types.

Regarding achievement ve: Interpreting

baseline and base year results at dierent scales,

relating to dierent markets and various incomes,environmental and other indicators require a user-

riendly CAPRI-GUI. As pointed out above, the

existing CAPRI-GUI was extended and improved

considerably. A number o tables are now available

(structure o the holdings) or displaying results at the

arm type level. In addition, results can be displayed

in charts as is demonstrated in this report, oering a

number o dierent styles. Similarly, mapping tools

were adjusted to allow the demonstration o results

in arm structures. Furthermore, it was necessaryto develop routines or data conversion and xml

descriptions. Also, the denition o 1,941 arm types

required an additional programme. Furthermore, a

selection o predened groups was implemented

to quickly navigate through the selected arm type

results in the CAPRI-GUI.

Outlook

The ollowing possible extensions are

worth mentioning. First o all, an update o the


16/94

15




s(EU-FARMS)

aggregation weights to indicate structural change

would clearly improve the understanding o the

baseline arm type results. Here, results o the

FP6 project SEAMLESS can help to integrate such

an approach in the uture.

Secondly, certain data rom the FADN sample

have not as yet been ully exploited. The input

coecients used in CAPRI are in part estimated

in FADN, but dierences in arm size and

specialisation are so ar not taken into account.

Here, changes in the data preparation step could

improve the parameterisation o the system to

capture dierences in technical and economic

eciency between arm types and regions.

Thirdly, so ar the arm type models are

solved independently, and it is assumed

that odder and organic manure cannot be

exchanged between arms in the very same

region. Further research is necessary to develop

appropriate modules and routines to account

or such eects. The FP7 project CAPRI The

Rural Development Dimension (CAPRI-RD)

will ocus on those issues to urther improve

the arm type layer.


17/94


18/94

17




s(EU-FARMS)1 Introduction

1.1 General objective o the tender

This report was prepared or the tender

Update o a quantitative tool or arm systems

level analysis o agricultural policies launched

by the Institute o Prospective Technological

Studies (IPTS) with the contract number

150910-2007-F1SC-DE. The main objectives o

this study are:

To update and complete the current farm

module database using the most recent data

to cover as a minimum the Member States

(MS) o the EU beore 1 January 2007 (EU-

25); (chapter 2)

To obtain the associated typical regional

arm structures and veriy the consistency o

the obtained arm structures using external

statistical data sources; (chapter 3)

To incorporate the upgraded and extended

arm module into the existing European

Union (EU) wide agricultural sector modelling

ramework CAPRI (Common Agricultural

Policy Regionalised Impact Modelling System)

and document its incorporation; (chapter 4)

To discuss the current advantages and

shortcomings o the module developed, andpropose urther renements (chapter 5)

To enhance the exploitation tools enabling

the deployment o arm and spatial typologies

or presentation and interpretation o results.

(chapter 6)

1.2 Organisation o the report

The report o this tender describes the

development and implementation o the arm

type layer, the assumptions underlying its use

and outlines the workfow. The report starts with

the description o the new data source Farm

Structure Survey (FSS). In chapter 2 the utilised

databases are summarised and compared i they

are considered important or the implementation

in CAPRI. Chapter 3 discusses the data processing

used to derive and estimate a complete and

consistent arm type base year or CAPRI. Changes

to the CAPRI programming core or the tenderare explicitly discussed. However or the basic

methodological approaches o CAPRI, we reer the

reader to the standard documentation o CAPRI in

Britz and Witzke (2008). Chapter 4 illustrates and

discusses the methods and assumption applied

to the arm type base year data in order to derive

a scenario comparison point or counteractual

scenario analysis. To be able to appreciate the

changes and assumptions developed or the arm

type approach, the general CAPRI approach isbriefy outlined. The problems and assumptions

are briefy discussed and an example or a selected

region underlines the concept. The tender report

nishes by critically evaluating the results and

suggestions on uture research directions. Chapter

6 concludes by summarizing programming tools

developed during the project.

1.3 Background

The Common Agricultural Policy

Regionalised Impact Modelling System (CAPRI) is

a pan-European analytical instrument or policy

support, developed under several Framework

programmes over more than a decade. It has now

been installed and is used within the European

Commissions Joint Research Centre - Institute

or Prospective Technological Studies (IPTS). The

CAPRI arm type layer, a component o the CAPRImodel system, was developed in order to support

the policy assessment at the arm level.


19/94

18

1Introduction

The main aim o introducing arm types into

the CAPRI model (i.e. breaking down regional

models into a number o typical arm models)

was to improve the analysis o agricultural

policies by considering structural variables such

as arm size, crop mix, stocking density and

yields. The arm type layer can be understood as

a structural break-down o the arming sectors

o the Nomenclature o Territorial Units or

Statistics (NUTS) II regions. The implementation

o the arm type layer into the regional layer o

CAPRI means a great improvement with respect

to the reduction o the aggregation bias o the

model. Income eects as well as environmental

and social impacts can be better understood andcompared i arm typology and size classes are

used to model the armers response to prices and

policy incentives/restrictions. Furthermore, the

impact o policy measures on the specialisation

and economic importance o a arm group can be

traced within its associated region.

The attachment o a arm type layer approach

to CAPRI had already been attempted by several

previous research projects. The rst attempt wasconducted in a study or Directorate-General

Environment (DG ENV) in 2000 - 2001 (Helming

et al., 2003). A second attempt was made during

the rst implementation o the CAPRI model

at the IPTS, when a arm typology and a arm

module were developed (Adenuer et al., 2006b).

In both cases, the arm module enhanced CAPRIs

ability to carry out more in-depth analyses o

the implementation o the Common Agricultural

Policy (CAP).

However, the arm type layer has increased

not only the complexity o the CAPRI model,

but also the size o output gures and indicators.

Incomplete coverage, MS o the EU beore 1

May 2004 (EU-15) plus three New Member

States (NMS), absence o spatial typology or

interpreting results and the non-synchronised

development o the arm type layer and the core

CAPRI model (which was substantially upgradedin the meantime) hindered the usability o the

arm layer or policy analyses.

The present tender was launched by IPTS

to address these shortcomings and to bring the

existent arm type layer in line with the most

recent developments o the CAPRI model. The

research aims to complete the database o the arm

module and to achieve coverage o the EU ater

the enlargement on 1 January 2007 (EU-27) o the

arm typology so that the whole modelling system

becomes operational. Besides the harmonisation

o the arm type routines with the CAPRI core

model, this research ocuses on removing several

o the methodological drawbacks o the ormer

arm type model, mainly related to the sole use

o Farm Accountancy Data Network (FADN) as

the basis. Specicities o the FADN data, togetherwith the CAPRI rule that all regional levels rom

MS top level down to NUTS I, NUTS II, and arm

types, have to be consistent. Production has to

add up to the regional aggregate in some cases

this resulted in a large deviation between the

CAPRI arm structure and the arm structure itsel,

both observed in FADN and in reality.

The ollowing problems are in the ocus o

the current project:

FADN only partially represents agricultural

sector production. In particular, highly

commercial arms are not included in the

FADN database. A comparison is given in

section 3.2;

Furthermore,theeconomicsizethresholdsin

FADN are higher than the thresholds applied

in the databases CAPRI relies on - namely theStatistical Oce o the European Communities

Economic Accounts or Agriculture

(EUROSTAT EAA), the regional domain

at EUROSTAT (REGIO) and slaughtering

statistics. Consequently, in the previous

versions o the arm module the production o

arm types was scaled upwards;

CAPRI and all its statistics refer to

administrative NUTS regions instead oFADN regions. For this, in the old approach

a so-called re-sampling routine o FADN


20/94

19




s(EU-FARMS)

data was developed, responsible or linking

the FADN arm accounts to the NUTS II

regions. Despite the eorts made, that

methodological approach did not result in

an adequate distribution o the arm types

across the EU.

Obviously, to overcome the imperections

o the previous methodological approach, there

is a need or additional inormation to the FADN

database. Most importantly, any new data source

has to be available or all CAPRI regions, both at

the time o the project and in the uture. The arm

structure database (FSS) collected by the individual

MS and held by EUROSTAT ts these requirements.The nal decision to apply or FSS data was made

during the kick-o meeting and was supported

by representatives o Directorate-General or

Agriculture and Rural Development (DG AGRI)

and o IPTS. FSS comprises an intermediate survey

o the arm structure in the EU-27 on the basis o

three years, and a complete survey on the basis o

ten years. However, due to Council Regulations

(CE) No. 322/97 on statistical condentiality (OJ

No L 52/1), and (EURATOM, EEC) No. 1588/90on the transmission o data subject to statistical

condentiality to EUROSTAT (OJ No L 151/

1), which stipulate detailed rules or receiving,

processing and disseminating condential

data, instead o providing individual FSS data,

EUROSTAT oered and agreed to run specic

regional aggregations or the EUFARMS study1. The

contractor provided EUROSTAT with a document

where the data request was described and an

initial test aggregation was proposed. To decide thenal aggregation, it was necessary to run three test

aggregations at EUROSTAT. The aim was to nd an

aggregation or the region (some regions in CAPRI

are aggregated), economic size and arm typology

items that include sucient arms per arm group

to ensure at least a minimum representation o the

sector. In order not to aect the reliability o the

arm groups obtained, the general rounding-o

1 The FSS aggregation was obtained without a ormalagreement. The data are public domain and can beobtained rom EUROSTAT on request.

o the tenth digit o each delivered variable and

leaving out cells relevant or more than 80 per

cent o an aggregated group had to be considered

when deciding the aggregation levels or region,

economic size and type.

The nal aggregation holds the aggregation

error in an acceptable range. The data preparation

process nished at the end o May 2008. The

structure o the arm types in CAPRI is now based

solely on the FSS records, and problems regarding

the representativeness o FADN have to a large

extent been overcome.

In spite o the success o deploying the FSS,the FADN is still a critical source o inormation

or the arm type layer, since the FSS includes the

production structure o a arm but no inormation

on economic production coecients (i.e. prices,

yields or input use). This inormation still has to

be obtained rom FADN.

1.4 CAPRI and the arm type layer

worklow

To achieve a CAPRI baseline2 at the regional

level3, several dierent programming modules4

have to be carried out beore the baseline

database is operational. The ollowing our

programming modules can be distinguished,

whereas the given chapters in brackets point to

the CAPRI documentation (Britz and Witzke,

2008) where they are discussed more thoroughly.

2 A baseline is used as comparison points or counteractualanalysis. The baseline may be interpreted as a projection intime covering the most probable uture development or theEuropean agricultural sector under the status-quo policyand including all uture changes already oreseen in thecurrent legislation.

3 Sometimes the arm type layer is also called the regionallevel in CAPRI. However, a arm type has no closed

regional area associated to it.4 The CAPRI model consists currently o our programmingmodules, namely COCO, CAPREG, CAPTRD andCAPMOD. All modules are built up on the previousprogramming module results.


21/94


22/94


23/94

22


24/94

23




s(EU-FARMS)2 Databases used to build up the arm type layer

To obtain a reliable and operational arm type

layer it is necessary to create wherever possible

sustainable links to well-established statistical data.

As already explained, or this purpose the FSS data

rom EUROSTAT was evaluated and then actually

used. Building up the arm layer also requires

developing robust algorithms that can be applied

across regions and arm types, so that automatic

updates o the dierent pieces o the CAPRI database

can be perormed. In order to build up the arm typedatabase the ollowing points have to be ensured:

Consistency at MS and NUTS II level, by

taking the CAPRI database as given, and

making production, activity levels5, eeding

and ertilizer distribution compatible at the

arm type level;

Agoodmatchofthe farm type production

structure with the arm statistics rom FSS;

A good match between FADN output

coecients and the FSS-based arm types

in CAPRI.

The CAPRI database is treated as a xed and

given data source with which the FSS and FADN

inormation is made consistent. In other words the

CAPRI database gures cannot be changed but

only broken down using the inormation rom FSS.Regarding the link between the CAPRI database

and FSS, holding the same regional coverage,

dierences exist or the ollowing reasons:

CAPRIconsidersathreeyearaverageforex-

post calibration (e.g. 2001 - 2003), whereas

FSS is available or dierent MS at dierent

years (2003 - 2005);

5 Activity levels comprise acreages and herd sizes in CAPRI.

The CAPRI database is already consistent

(e.g. closed market balances), complete

(i.e. data gaps have been lled in by means

o econometric routines) and has been

harmonised over time regarding product/

activity classications (e.g. aggregation o

the cheese or wheat marketed commodities).

Hence, production statistics in CAPRI can

dier slightly compared to the original statistics

and thereore deviation between NUTS II threeyear averages and FSS data exists;

CAPRIuses slaughtering statistics insteadof

average herd sizes as in FSS;

The mapping of codes from FSS to CAPRI

can create dierences due to overlapping

denitions o production activities as

depicted in Annex 8.2.

However, we have to stress that the FSS arm

type production statistics provided by EUROSTAT

or the project are more suitable than FADN to

derive the arm structure in CAPRI. This is justied

by a comparison between FSS and FADN on the

background o CAPRI in chapter 2.4. Figure 1

summarizes the databases and their relation to

the arm type database in CAPRI.

2.1 CAPRI database

The standard CAPRI database (arm types not

included) can be understood as a database which

is the outcome o a two step procedure briefy

explained in this section (or a detailed discussion

see chapter 2.5 and 2.6 in Britz and Witzke

(2008). In general, the database is built up on

original EUROSTAT statistics or MS and NUTS II

regions. Furthermore, the regionalised data mustalso be consistent with the national level. The rst

step comprehends the development o the MS


25/94

24

2Dat

abasesusedtobuildupthef

armtypelayer

database. Here, CAPRI integrates rom the EAA

market and arm balances (with valued outputand input use), crop areas and herd sizes, and

an engineering animal fow model or attening

and dairy animal activities. The CAPRI model is,

as ar as possible, ed by pan-European statistical

sources which are mostly centralised and

regularly updated. Farm and market balances,

economic indicators, acreages, herd sizes and

national input-output coecients were initially

drawn almost entirely rom EUROSTAT. In order

to use this inormation directly in the model, theCAPRI and CAPSIM teams developed COCO at

MS level rom EUROSTAT data (Britz et al., 2002).

The main sources used to build up the national

database are shown in the ollowing table.

In the second step, the regional database

at NUTS II level takes the previously calculated

national data as given (or the purpose o data

consistency) and includes the allocation o inputs

across activities and regions as well as consistentacreages, herd sizes and yields at the regional

level based on the statistics o the REGIO domain

o EUROSTAT. The input allocation step allows

the calculation o regional and activity speciceconomic indicators such as revenues, costs

and gross margins per hectare or head. The

regionalisation step introduces supply-oriented

CAP instruments like premiums and quotas (Britz

and Witzke, 2008, chapter 2.7).

The REGIO domain o EUROSTAT was

judged to be the only harmonised data source

available or regionalised agricultural data in

the EU. REGIO is one o several parts o the newCronos database and is itsel broken down into

domains, one o which covers agricultural and

orestry statistics.

The ollowing tables are available in the

agricultural and orestry domain and are used or

the regionalisation in CAPRI.

1. Land use;

2. Crop production: harvested areas, productionand yields;

3. Animal production: livestock numbers;

Figure 1: Data sources and relationships in the arm type layer


26/94


27/94

26

2Dat

abasesusedtobuildupthef

armtypelayer Table 2: Ofcial data availability in REGIO

Source: Eurostat (http://epp.eurostat.cec.eu.int)

Table 3: Sampling rate o FSS and overview o the used FSS database per MS

Source: EUROSTAT, Unit responsible or EUROFARM


28/94

27




s(EU-FARMS)

Slovakia, Romania and Hungary. This decision

was made based on the recommendation o

EUROSTAT with respect to a possible improved

data quality.6 In Annex 8.3, the table summarizes

the variables obtained rom FSS.7

2.3 FADN database

Beside the FSS data, the FADN database

(collected since the Council Regulation 79/65

in 1965) is required to complete the arm type

module in CAPRI. The ormer version o the arm

type layer included the FADN version or years

2001 and 2002 (and 2004 or the three selectedNMS, Poland, Hungary and the Czech Republic).

6 The consequences o using the latest surveys instead oa ull survey available or some NMS was not quantied,because there was no time in the study to compare thedierent alternatives. Thereore, we had to rely on theexperiences o the Unit experts on FSS in EUROSTAT andtheir recommendations.

7 The data or all NUTS II regions in CAPRI was provided ina text ormat (CSV). The les were converted into a GAMSreadable ormat (GDX) using the developed FSStoGDXjava programme (see Chapter 6.1).

At this time, the network represented almost

4,000,000 arms in the EU-15. The data or the

three mentioned NMS were taken directly rom

the Liaison Agency responsible or the collection

o the FADN data at the MS level. At that time

collection was in a test phase and it is likely that

the quality has improved considerably due to the

test procedures applied by the FADN Unit at DG

AGRI. FADN data or the accounting years 2003

and 2004, including NMS, are exploited in this

study. For Bulgaria and Romania (EU-02), where

no FADN data are available, arm types are derived

solely based on FSS data. Production coecients

such as yields or input use are assumed to be the

same or all arm types as at NUTS II level.

2.4 Comparing the degree o consistencybetween FSS and CAPRI database

We should consider that the CAPRI

database has already been made consistent

(e.g. closed market balances), complete

(i.e. data gaps have been illed in by means

Figure 2: Comparison UAA between FSS and CAPRI or all MS where FSS is available in year 2000


29/94


30/94


31/94


32/94


33/94

32

3Baseyearcalibrationforthefarmtypes

Table 5 shows the arm type dimensions or

the current version o the arm type layer.

There are our FSS arm groups which are

aggregated beore selection as CAPRI arm types:

(a) General Field Cropping (FT14) and Mixed

Cropping (FT60), resulting in FT14_60, and (b)

Specialised Cattle Rearing and Fattening (FT42),

and Cattle Dairying, Rearing and Fattening

(FT43), resulting in FT_42_43. This aggregation

was done to reduce the impact o the Regulation

o Condentiality applied by EUROSTAT. Hence

the arm type dimension in CAPRI reers to a

maximum o 14 dierent specialisations.

In addition, each specialisation o the CAPRI

arm types is broken down into three size classes:

small (less than 16 ESU), medium (more than 16

but less than 100 ESU) and large (more than 100

ESU) arm types. Although we had obtained our

Table 5: Type o arming groups in CAPRI

Table 6: ESU groups selected or the data request and the fnal CAPRI arm types


34/94


35/94

34


classes rom EUROSTAT (see Table 6, column

one), we had to urther reduce the possible

number o arm types per NUTS II region in order

to keep the residual arm type (representing the

remaining unselected arm groups rom FSS) in

an acceptable range. Thereore we aggregate the

ESU group less than 2 ESU together with greater

than 2 and less than 16 ESU arm group.

The set names or the CAPRI arm types

are dened using the last three letters o the

region name. The arm type classication can

thus distinguish 42 possible arm types (14

specialisations and three size classes). The

resulting regionalised typology enables theaggregation or comparison o arm types o the

same specialisation across regions.

3.1.1 Defning the rules or aggregating the FSS

arm groups to CAPRI arm types

The rule applied to select the most relevant

arm types rom the FSS groups is deined as

ollows:

For each FSS farm group (region, type

and ESU), the importance is calculated by

assigning a weight o one per UAA hectare

and by assigning a weight o one per

Livestock Standard Unit (LSU)9. The weights

are added and ranked to obtain the CAPRI

arm types. The average LSU per hectare is

used to represent the animal production,whereas the UAA is used to represent

the land use. This weighting schema was

9 This indicator was already used or the previous studies(Adenuer et al. 2006a). Alternative weighting approachesare considered in Adenuer et al. (2006b). Nevertheless, ingeneral all arm groups rom FSS are considered as supplymodel in CAPRI, because the residual group is treated likea normal supply model.

Figure 6: General overview o the distribution on economic size groups across EU-27, EU-25, EU-15, EU-10 and EU-02


36/94


37/94

36


3.1.2 Results o the current selection rule

In Figure 6 the number o arm types and

their share o the economic size classes across EU

aggregates are presented. As expected, the share

o the size class group with less than 16 ESU is

substantial in the EU-27 and particularly in the

MS that joined the EU on 1 May 2004 (EU-10)/EU-02. Also, the polarity (small and/or big arms)

o the arm structures in the EU-10 is evident.

Clearly, the construction o CAPRI arm types rom

the FSS data draws an authentic picture regarding

the distribution o size classes across Europe.

In Figure 7 is the distribution o CAPRI arm

types according to specialisation presented.10

10 Figure 2 and Figure 3 are not weighted numbers, becausethe gure should demonstrate the distribution o theselected arm types based on the selection rule.

Table 8: General overview o arm types selected or the CAPRI layer

A complete picture o the distribution is given

in Table 8. The arm type distribution o CAPRI is

divided into EU MS aggregates, rom the EU-27

to the EU-02, type o arming (specialisation) and

economic size classication. The overall number

o arms is indicated in the last row. Regions

without a value have either no arm with that arm

type observed or the arms were not selected andaggregated in the residual type.

The ollowing maps are intended to illustrate

the regional distribution o the arms selected or

CAPRI. They present or all economic size groups

the distribution o specialist cereals, oilseed and

protein crops (FT13) across the EU-27 obtained

by the selection procedure.


38/94

37




s(EU-FARMS)Map 1: Distribution o (a) total UAA, (b) number o holdings (in thousand hectares) and (c) share

on NUTS II UAA in percentage o arm type: specialist cereals, oilseed and protein crops(FT 13) less than 16 ESU in 66 NUTS II.

UAA in thousand hectares of farm type: Specialist cereals, oilseed

and protein crops (FT 13) - Less than 16 ESU in 66 Nuts II

Number of holdings in thousand of farm type: Specialist cereals, oilseed

and protein crops (FT 13) - Less than 16 ESU in 66 Nuts II

a) b)

Share of UAA on Nuts II UAA in percentage of farm type:

Specialist cereals, oilseed and protein crops (FT 13) - Less than 16 ESU in 66 Nuts II

c)

Map 1 (a,b,c) indicates all regions where

small arms (less than 16 ESU) with specialist

cereals, oilseed and protein crops are selectedas arm types in CAPRI (as representative). Map

1 (a) reveals the overall importance between the

dierent regions, and Map 1 (b) shows the number

o holdings behind the selected arm type. Map 1

(c) presents the share o the arm type on NUTS IIUAA in percentage. The average size across EU-

27 is about 12 hectares UAA per arm.


39/94

38

3Baseyearcalibrationforthefarmtypes Map 2: Distribution o (a) total UAA, (b) number o holdings (in thousand hectares) and (c) share

on NUTS II UAA in percentage o arm type: specialist cereals, oilseed and protein crops(FT 13) greater than 16 less than 100 ESU in 101 NUTS II regions.

Distribution of total UAA in thousand hectares

a)

Number of holdings in thousand hectares

b)

Specialist cereals, oilseed and protein crops

(FT 13) - Greater 16 - Less 100 ESU

in 101 Nuts II regions

c)


Specialist cereals, oilseed and protein crops (FT 13) - Greater 16 - less 100 ESU in 101 Nuts II regions

Map 2 (a,b,c) indicates all regions where

medium scale (more than 16 ESU and less than

100 ESU) specialist cereals, oilseed and proteincrops are selected as arm types in CAPRI.

These arm types are particularly important in

France, Spain and the northern part o I taly. The

average arm type size in Europe is about 137

hectares UAA per arm. Map 2 (c) and Map 1(c) indicate the share o the arm type UAA on

NUTS II UAA.


40/94


41/94

40


The average size o holdings greater than

100 ESU and specialised in cereals, oilseed and

protein crops (FT 13) is about 740 hectares o

UAA in Europe, and their regional distribution is

indicated in Map 4.

Map 4 depicts the average UAA per arm

type calculated by the UAA and the number

o holdings indicated in Map 3 (a) and (b). The

Map 4: Average arm size hectares and arm type: specialist cereals, oilseed and protein crops (FT13) greater than 100 ESU in 70 NUTS II regions.

Specialist cereals, oilseed and protein crops

Greater 100 ESU in 70 Nuts II regions

Average farm Size hectares

average size per arm diers across the EU. In

France, the average arm size is 110 hectares

whereas in Eastern Germany the average arm

size ranges rom 690 to 1,200 hectares. The

ollowing three maps present an overview othe distribution and UAA used or all three

economic size classes or specialist dairy

arms across the EU-27 (compare Map 5 with

Map 7).


42/94

41





on NUTS II UAA in percentage o arm type: specialist dairy (FT 41) less than 16 ESU(FT41L16) in 42 NUTS II regions.

UAA in thousand hectares of farm type

a)

Number of holdings in thousand hectares of farm type

b)

Specialist dairyin (FT 41) - Less than 16 ESU

(FT41L16) in 42 Nuts II regions

c)


Specialist dairying (FT 41) - Less than 16 ESU (FT41L16) in 42 Nuts II regions


43/94

42

3Baseyearcalibrationforthefarmtypes Map 6: Distribution o (a) total UAA, (b) number o holdings (in thousand hectares) and (c) share

on NUTS II UAA in percentage o arm type: specialist dairy (FT 41) greater than 16 lessthan 100 ESU (FT41GT16L100) in 120 NUTS II regions.

UAA in thousand hectares of farm type

a)

Number of holdings in thousand hectares of farm type

b)

Specialist dairyin (FT 41) - Greater than 100

ESU (FT41GT100) in 77 Nuts II regions

c)


Specialist dairying (FT 41) - greater 116 - less 100 ESU (FT41GT16L100) in 120 Nuts II regions


44/94

43





on NUTS II UAA in percentage o arm type: Specialist dairy (FT 41) greater than 100ESU (FT41GT100) in 77 NUTS II regions.

UAA in thousand hectares of farm type Number of holdings in thousand hectares of farm type

b)

Specialist dairyin (FT 41) - Greater than 100

ESU (FT41GT100) in 77 Nuts II regions

c)


Specialist dairying (FT 41) - Greater than 100 ESU (FT41GT100) in 77 Nuts II regions

a)


45/94


46/94

45




s(EU-FARMS)

not be ound. For a detailed overview on the

number o FADN records, please see Table 16.

The table shows that or almost all EU-15 MS,

a arm type group exists without representation

in FADN. This particularly holds true or

commercial arm types (F50), because in thiscase many arm types rom FSS were lacking an

FADN counterpart.

Table 10 represents the number o hectares

o UAA (in thousands), which are represented

in the CAPRI arm type layer but do not have a

FADN representation. For Germany this means

that almost 500 thousand hectares are not

presented by FADN. Obviously, a lot o arm

types with an economic size class o less than16 ESU are not characterised in FADN due to

the applied FADN thresholds.

Table 11 presents the number o slaughtered

pigs represented in the arm type layer without a

corresponding FADN arm. Representation means

that at least one arm record exists or the particular

type and the economic size class in FADN.

High values are ound in FT50 + ESC2 and

FT50 + ESC3. This veries the problem that

highly commercial arms are not well represented

in FADN. The comparison could be continued or

many other important production activities.

The comparison o the match between FSS and

COCO and CAPREG data rom CAPRI, respectively,

and the comparison between FADN and FSS,

clearly support the decision to use FSS instead oFADN as the main data source or deriving the arm

type production structure or the arm types.

Table 10: UAA o CAPRI arm types without representation in FADN

Table 11: Number o slaughtered pigs (in thousands) or attening without representation in FADN


47/94

46


3.3 Consistency o arm types to NUTSII (CAPREG-FARM)

The input data or the arm layer consistency

routine include the data provided by the selection

routine CAPRI-FSS (arm type denition and its

FSS production data) and FADN data and CAPRI

data on NUTS II regions. These data are made

consistent at each stage with the CAPRI regional

data. The routine is called CAPRI-FARM and

is technically integrated in CAPREG. To date,

two independent sub-routines relevant or the

arm type consistency step exist. The rst is an

econometric estimation o input allocation based

on FADN data. The routine is currently underdevelopment and not yet integrated. The aim o

this routine is to estimate arm group specic

input allocation or general inputs which are not

observed in FADN. The consistency routine to

make the arm types consistent to the regional

levels in CAPRI includes:

the consistency of level (from annual to

multiannual data FSS and CAPRI database)

theadjustmentsofoutput(grossproduction

and yields rom FADN),

theadjustmentofgeneralinputs,

thefeedcalibrationforthefarmtypes,and

fertilizer use for farm types (nitrogen and

phosphate balances).

In chapter 3.3.4, the arm supply model and

its calibration routine is briefy described. The last

section o this chapter presents some selected

results o the arm type base year (average three

years 2001 - 2003).

The objective o this section is to demonstrate

how the base year at arm levelis constructed. The

NUTS II data are taken as given (or data consistency

purposes). Bearing in mind the goodness o tbetween FSS and CAPRI as depicted in Figure 2

to 4, the aim is to overcome the inconsistencies.

This is done in an estimation ramework using

a Highest Posterior Density (HPD) estimation

approach (Britz and Witzke 2008). The ollowing

points are a potential source o inconsistencies:

Differenttiming of data collection (FSS has

dierent survey years or each MS and is not

an average like the CAPRI base year data);

The differences due to the regulation of

condentiality applied by EUROSTAT to FSS

arm data;

Differences due to inconsistent mapping

between FSS code and CAPRI code (seeAppendix 8.2);

Deviations between EUROSTAT and

estimated times series in the CAPRI

database.

3.3.1 Production

The objective o the estimation is to nd

activity levels or all arm types in a NUTS II regionwhich will minimize the deviation to the raw FSS

activity levels (prior inormation) considering the

ollowing constraints:

Set-asideregulationsforallfarmtypeshave

to be satised (voluntary, obligatory and

maximum rates);

Over allfarmtypes in aregionand forall

activity levels (sot wheat, sugar beet), thelevels have to add up to NUTS II data;

TheUAAofthefarmtypeswillsumuptothe

NUTS II UAA, whereas the deviation rom

UAA is penalised above average.

For a detailed discussion on the

implementation o the estimation ramework see

Britz et al. (2007), Britz (1999) and Gocht (2008).

The major drawback o this approach

could be that large deviations between the two


48/94

47




s(EU-FARMS)

data sources (CAPRI NUTS II and FSS) could

lead to deviations in the production structure

which changes the Type o Farming and/or

the Economic Size Class to which a arm type

was attached by denition. As a consequence

o such a deviation and although the arm types

in a region would still be a perect breakdown

o the NUTS II region, the general denition

o a arm type no longer represents the actual

arming structure. To avoid such inconsistency

it would be necessary to dene or extend the

estimation in such a way that each arm type

remains in the original type o arming and

economic size class. Technically, this requires

expansion o the HPD estimation ramework byadding constraints and objective entries which

will ensure that the deviation o the production

levels remain in the original range o the type o

arming and the ESU class.

To achieve such a Type o arming and ESU

consistent approach, it is necessary to dene a

set o rules or each arm group. The set o rules

consists o two dimensions. The rst dimension

(i) is a set o rules which ensures the type oarming and the second dimension (ii) is a set o

rules which keeps the ESU class consistent.

Regarding (i): The type o arming

ollowing the ocial documents (European

Commission, CD 85/377/EEC) is dened

by determining the relative contribution o

dierent enterprises to its Total Standard Gross

Margins (TSGM) (Article 6). To keep the relative

shares o the arm group during estimation,it is necessary to calculate the TSGM o each

arm type and its relative shares o the dierent

enterprises (partial SGM) o the arm type. The

so-called partial SGM (P1-P52) and the TSGM

are dened in Annex 8.4. The partial SGM are

expressed as a raction o the TSGM. Thereore,

the partial SGM measures are used to classiy

the enterprise into its type o arming. The rules

are dened ollowing Annex II Section (B) o

CD 85/377/EEC. These thresholds determine theclass limits and dene the rules or constraints

o the estimation problem.

Regarding (ii): The ESU size dimension o the

arm is captured by the ESU concept and dened in

Chapter IV Article 8 in CD 85/377/EEC and Annex

III. The rules which ensure the economic size class o

a arming group are calculated using the TSGM and

could be introduced similarly to the type o arming

rules as a set o constraints during the estimation

problem. The partial SGM and the TSGM are

calculated as a product o the observed or estimated

levels or animal heads and its corresponding regional

specic SGM. The SGM are provided by the MS.11

When technically implementing the approach

the rules have to be individually linked to each arm

type during the estimation. However, due to theCouncil Regulations (CE) No. 322/97 on statistical

condentiality (OJ No L 52/1) the raw FSS might be

modied and the arm group data o the original

dataset might not be consistent with the type o

arming and size class indicated by EUROSTAT.

Thereore, the type o arming and the ESU class or

each FSS group have to be calculated in order to use

the related rules in the estimation approach. Table

12 presents a comparison or Denmark12 between

identied rules and rules provided by EUROSTATor the raw FSS data. It can be seen that almost all

arming types relevant or the arm type layer could

be recovered rom the FSS data.

The nal estimation problem is thereore

specied as the search or the activity levels

which minimize the deviation between the prior

inormation on activity levels (FSS data), the prior

inormation on TSGM and the shares o the partial

SGMs with respect to:

TypeofFarmingrules,

EconomicSizeClassrules,

Set-asideregulations,

LevelsbeingrequiredtoadduptoNUTSIIdata.

11 The SGM are collected by EUROSTAT rom the MS andare downloadable rom the ocial EUROSTAT webpage.

The special method or grazing stock and odder crops isimplemented in the CAPRI arm type approach (see CD85/377/EEC, Annex I, 5. Treatment o special cases).

12 Denmark has no urther sub-regions in CAPRI, whichprompted its use as an example.


49/94

48


The UAA o the arm types will sum up to the

NUTS II UAA, presenting a comparison between

prior and estimated partial SGMs values or all arm

types in Denmark. Table 13 shows that the deviation

rom the prior partial SGMs shares is minimal.

Table 14 presents a comparison between

the prior and estimated values o the ESU and

the estimated UAA values or all arm types in

Denmark. Here, the deviations are also verysmall, which shows that FSS data rom Denmark

already ts with the CAPRI upper NUTS II data.

Table 15 presents the deviation o crop

groups or the dierent arm types in Denmark.

Two things should be mentioned regarding the

results. Firstly, the deviation or the residual

arm type is larger than or the other arm types.

The reason is the missing rule or the residual

arm type. The deviations or CAPRI arm types

are restricted by constraints and are thereore

less prone to deviations. Secondly, we note

small observations about the cropping pattern,i.e. oils are less robust and the percentage

deviation can be higher.

Table 12: Farming types and ESU class recovered rom the FSS raw data

Table 13: Prior and estimated partial SGMs (P1-P5) or all arm types in Denmark


50/94

49




s(EU-FARMS)

By the Type o arming and ESU class

consistent estimation, the adjustment o the

production levels to the upper regional level

is now achieved in a way which ensures the

consistency o the dened arm typology.

3.3.2 Yields

Yields are derived wherever possible andavailable rom FADN data. Table 16 gives an

overview o the number o arm types in CAPRI,

the number o FADN accounts and the number

o records which could be used or deriving gross

production rom FADN or the arm types. A

separate column indicates this inormation or the

residual arm types. The table also comprehends

a summary about the dierent supply regions,

which currently consist o 27 MS o the EU,

Norway and the Western Balkan countries.

However, not all these regions are broken downinto arm types (column three), because or

Norway and the Western Balkans FSS data do

Table 14: Prior and estimated ESU values and UAA o all arm types in Denmark

Table 15: Estimates or selected crop activity level in Denmarkk


51/94

50


not exist. As or EU-02 no FADN data exist, prior

inormation on their corresponding yields is taken

rom the higher regional aggregate.

Yields coecients or the arm types were

implemented along the ollowing path:

Forall farmtypes allrelatedFADN records

were selected.

TheFADNrecordsweremapped(cross-sets)

into CAPRI code denition.

Afterwards, yields for all crops and milk

yields or dairy cows where calculated rom

the accountancy record or all FADN years

(2004 and 2005).

AnaverageovertheFADNyears2004-2005

was calculated using the gross production

rom FADN as weights.

The last three steps are calculated for allexisting combinations o types o arming

and economic size classes available at

NUTS II regions on MS level, and were

dened on NUTS I level. This was done in

order to select wherever possible identicalarm type and economic size inormation

rom dierent regional levels. I FADN

records were not available on NUTS II

level, inormation was taken rom NUTS I

or MS level. In case there exists a arm type

without representation on either NUTS I/II

or MS, NUTS II data rom CAPRI are taken

as prior inormation.

FADNyieldsfromNUTSIIorfromthehigherregional aggregates are mapped to the arm

types as prior inormation.

For all output coefcients without

representation in FADN, regional yields

are taken rom NUTS II level. Columns (9)

in Table 16 presents the number o arm

types without representation o FADN

records at NUTS II, columns (11) at NUTS

I and column (12) at MS level. For thisarm types yields are taken rom the CAPRI

NUTS II database.

Table 16: Overview o the regional coverage o the current arm type layer and the representativenesso FADN or the FSS-based CAPRI arm groups evaluated at MS level


52/94

51




s(EU-FARMS)

Inalaststepthefinalyieldsareestimated

using a HPD estimation ramework in

which the FADN yields are taken as

prior inormation. The objective o the

estimation is to minimize the deviationbetween the prior inormation and the

posterior estimate (CAPRI arm type

yields), subject to the constraint that

activity levels, times yields over all arm

types equal the regional gross production.

A plausibility test ensures that FADNcoecients are in line with our

expectations. The test is constructed by

Figure 8: Distribution o FADN prior yields and posterior yields or all German arms with type oarming FT13 and ESU Group three (>100 ESU) (in tons)

Note: green line [rectangle] belong to the estimate

Figure 9: Distribution o FADN prior and fnal adjusted yields or all arms in Germany with type oarming FT13 and ESU Group three (greater than 16 and less than 100 ESU) (in tons)


53/94


54/94

53




s(EU-FARMS)

3.3.3 Inputs

A discussion about the input allocation or

arm types has to distinguish between the dierent

input categories, namely:

Mineralfertilizinginputs,

Feed inputs (non-tradable and tradable

eeding stu), and Generalinputsusedbycropsandanimals.

The rst two input categories are discussed

in the next section. In CAPRI, general input costs

are related to the maintenance o materials,

maintenance o buildings, seeding, plant

protection, electricity, heating gas and oil, uels,

lubricants and other input costs.

For these cost positions, in FADN, noinormation on crops or animal activities level

is recorded as only the sum o the various input

categories is available. During this study tests were

conducted (Gocht, 2008) to nd out whether it

is possible to estimate specic input allocations

based on total cost positions. The tight time rame

or the project did not allow the generalisation o

the approach or the current version o the CAPRI

arm type layer. Instead, input levels dependent

on yield are calculated based on the yield level

and others are taken over rom NUTS II leveland are made consistent with the total amount o

input per region.

Feed calibration

The input allocation or eed describes

how many kilograms o certain eed categories

(cereals, rich protein, rich energy, eed based

on dairy products, other eed) or single eeding

stu (odder maize, grass, odder rom arableland, straw, milk or eeding) are used per animal

activity level.

Figure 11: Distribution o FADN prior and fnal adjusted yields or all arms in Germany in the case odairy cows (high and low yield)


55/94

54


The input allocation or eed on arm type

level takes into account the nutrient requirements

o animals, building upon requirement unctions.

The input coecients or eeding stu shall hence

ensure that energy, protein requirements, etc.

cover the nutrient needs o the animals at arm

type level. Ex-post, the coecients should be

in line with the regional odder production and

total eed demand statistics at regional levels

(NUTS I and NUTS II) and on a national level, the

latter stemming rom market balances. The input

coecients or eed together with eed prices

taken rom the national level should lead to

sizeable eeding costs or the activities on the arm

level. Furthermore, the ollowing considerationsand assumptions, taking the heterogeneous

structure o production at arm level into account,

are made:

No restriction for maximum straw feed on

arm level is applied, because there are some

specialised arm types which do not have

animal activities. Otherwise, this constraint

would lead to inconsistency during eed

calibration. However, it is ensured that totaluse over all arm types does not exceed total

production at NUTS II level.

As FADNdoesnot have informationabout

yields or odder production, non-tradable

odder yields are calculated based on the

stocking density, which is calculated based

on the FSS data o each arm type13.

Ex-ante, the sum over all farm types oftradable eeding stu and non-tradable

eeding stu has to be consistent with the

usage o odder at NUTS II region.

Figure 12 (a,b) shows a comparison o the

eed distribution or dairy cows (low yield) or

all milk producing arm types in Denmark. The

upper Figure 12 (a) presents the activity levels o

13 For a detailed description o how the stocking densityinfuences the odder yield distribution we reer the readerto Britz and Witzke (2008, chapter 2.5.3).

dairy cows (low yield), the lower Figure 12 (b)

the associated eed distribution. It can be seen

that the eed distribution clearly depends on

the distribution o the upper regional level, here

Denmark (rst stacked column). All estimated

distributions and animal requirements are

available via the CAPRI-GUI.14

Fertilizer calibration

The applied methodology to achieve a

closed and balanced nutrient fow (mineral

ertilizer use, manure cycle, nutrient losses) or

arm types is similar to the other regional levels

in CAPRI. Hence, or a proound discussionplease reer to the CAPRI documentation

(Britz and Witzke, 2008). In this context we

should ocus on the possible occurrence o

biased estimates o nutrient balance positions

due to the decoupling o animal production

rom land use and the missing interaction o

arm models at regional level. Although the

problem only occurs in NUTS II regions which

are small and where the corresponding arm

type group models cannot be considered asstand-alone models (without any interaction to

the other arm type models) the problem has

to be considered in uture research projects15

by linking those arm types via interregional

markets or nutrients. The section will thereore

present the problem using a NUTS II region,

Noord-Braband in The Netherlands. From the

discussion it also becomes obvious in how

much detail the nutrient balance or the arm

types is treated in CAPRI.

Map 8 presents the nutrient surplus or

nitrogen in kg per hectare at the NUTS II level. In

the NUTS II region Noord-Braband, about 340 kg

nitrogen per hectare is reported.

14 The results can be exploited when exploring the CAPREG-

FARM results on arm type level using the GUI (ContextMenu -> Farm -> eed distribution and eed requirements).15 In the FP-7 project CAPRI-RD (project No.: 226195) a link

between arm types via markets or nutrient and odderwill be developed.


56/94

55




s(EU-FARMS)Figure 12: Levels o dairy activities low yield (fg. a) and eed distribution or Denmark and relevant

arm types (in kg eed per head) (fg. b)


57/94

56


Corresponding to the NUTS II level surplus

Table 17 reveals the arm type distribution o

UAA and produced pigs and sows in the Noord-

Brabant region. It is evident that the arm type

specialised granivores (F50) accounts or a large

share o the pig production in that region.

The corresponding balance or nitrogen isdepicted in Table 18. The last column in particular

reveals a rather heterogeneous distribution o

Map 8: Surplus at soil level in kg nitrogen per hectare or The Netherlands at NUTS II level in thebase year

Table 17: UAA and pig production activities in the Noord-Brabant region

nitrogen losses in soil across the arm types. Such

high losses are probably not observed in reality due

to the regulations which orce armers to spread

manure on less polluted regions. The decoupling

o animal production and land use o certain arm

types is responsible or this overestimation and also

the relatively small NUTS II regions. Such arm type

models cannot be considered as stand-alone modelsbut have to be linked to overcome the problems.


58/94

57




s(EU-FARMS)

3.3.5 Selected results or the arm type

database

The base year database or the arm type layer

includes a variety o indicators and summary

statistics beside coecients or constructing the

supply model. Currently, the database consists

o more than 10 million data cells. For a nal

version o the arm type layer an aggregation has

to be developed to organize the huge amount o

data in a more convenient way. Our proposal is

to dene or each arm type a representation atthe MS and the EU (EU-27, EU-25, EU-15, EU-

10 and EU-02) levels. Furthermore, it is planned

to aggregate the arm type not only with respect

to the region but also with respect to the type

and economic size on Member and EU level.

Consequently, the GUI reporting tool has to be

extended or more arm level specic reporting.

One way could be to add, beside the sector

representation, average arm gures using the

number o holdings per arm group.

The huge amount o arm type data or

the base year oers the possibility o present a

thorough picture over all arm types and regions.

In order to study the arm types in detail we reer

the reader to use the GUI - Exploitation tool.

This section can only present some selected

results and is organised into two sub-sections.

The rst section shows results based on the type

o arming and economic size across Europeanregions. In the second part arm types o a

particular region are presented.

Table 18: Farm type nutrient balances in the Noord-Brabant region

3.3.4 The arm type supply model

The arm type supply module consists o

independent aggregate non-linear programming

models representing activities o all arm types

at the NUTS II level. The programming models

are a kind o hybrid approach, as they combine

a Leontie technology or variable costs covering

a low and high yield variant or the dierent

production activities with a non-linear cost

unction which captures the eects o labour

and capital on armers decisions. The non-linearcost unction allows or perect calibration o

the models and a smooth simulation response

rooted in observed behaviour. The models

capture in high detail, similar to the NUTS

II supply models, the premiums paid under

the CAP, including Nitrogen, Phosphate and

Potassium (NPK) balances and a module with

eeding activities covering nutrient requirements

o animals.

The constraints besides the eed block

are arable land and grassland, set-aside

obligations and milk quotas. The complex

sugar quota regime is captured by a component

maximising expected utility rom stochastic

revenues. Prices are exogenous in the supply

module and provided by the market module.

Grass, silage and manure are assumed to be

non-tradable and receive internal price

JRC53518Institute for Prospective Technological Studies - Update of a Quantitative Tool for Farm...

Documents

Transcript of JRC53518Institute for Prospective Technological Studies - Update of a Quantitative Tool for Farm...