close

Вход

Забыли?

вход по аккаунту

код для вставкиСкачать
Input Requirement Estimations
Estimating Input Requirements using
FADN Data
Stephan Pfaffenzeller, NUI, Galway
Objective
• To Estimate Intput Requirement
Coefficients for Each CAPRI
Activity for insertion into the
CAPRI database.
Data
•
•
•
•
•
•
•
•
FADN DATA from 1990-1999
AT - Austria - 2451 farms
BL - Belgium, 2601 farms
DE - Germany, 15110 farms --> price data from ‘91-’99
DK - Denmark, 6625 farms
EL - Greece, 11877 farms --> price data from ‘95-’99
FI - Finland, 1324 farms
IR - Ireland, 3409 farms --> no price data prior to 1995
Data
•
•
•
•
•
•
•
•
FADN DATA from 1990-1999
IT - Italy, 57264 farms
PT - Portugal, 6379 farms
SE - Sweden, 1191 farms
UK - United Kingdom, 6668 farms
ES - Spain, 22609 farms
NL - Netherlands
FR - France
Variables
•
•
•
•
I: CAPRI INPUTS
L: CAPRI LEVELS (hect./head)
Y: CAPRI YIELDS (tonnes p/h etc.)
O: CAPRI OUTPUTS (tonnes, litres etc.)
• 11 Input Variables
• 53 Level Variables and 53 Output/Yield Variables
in Full Model
Model (for Any One Input)
Basic specification
I it 

t
t 

L
 L X Lit  u i  eit
• i is the farm i.d., t the year, L are Relevant
Level Values
Model (for Any One Input)
Alternative specification I
I it 

t
t 

L
 L X Lit 

O
 O X Oit  u i  eit
• i is the farm i.d., t the year, L are Relevant
Level Values and O are Output Values
Model (for Any One Input)
Alternative specification II
I it 

t
t 

L
 L X Lit 

Y
 Y X Yit  u i  eit
• i is the farm i.d., t the year, L are Relevant
Level Values and Y are Yield Values
Choice of Estimator
• SUR eliminated (not feasible – problems
last summer with SUR estimation)
• Choice between OLS, FE and RE for each
model
• Preliminary Tests revealed FE was far
superior: decided to be consistent
Results
• Generally good (pending final evaluation of
specifications)
• Problems with Negative coefficient
estimates in some cases
• Fitted values need to be reconciled with
EAA
Negative Coefficients
• These problems are mainly with the
OTHER categories of activities/inputs
(OFRU, OCRO, OANI etc.).
• The chosen solution is to use the average
coefficient from the relevant activity
category (e.g. the Fruit Average if OFRU is
negative)
Selected Results
Level Coefficient Estimates
L estimates for selected Crop activity levels:
NL
SE
EL
PT
swhe
584.13*
-62.00
336.17* 315.43*
barl
501.75
69.57
372.24* 161.80*
oats
433.46
40.99
247.51* 91.43*
rye
1623.70*
-37.47
272.94* 69.41*
Level Coefficient Estimates
L estimates for selected animal activity
levels:
NL
dcow
SE
EL
PT
1229.45* 1520.40* 493.30* 1118.36*
hens
6.74*
4.44*
7.71*
4.91*
shgm
93.00
6.20
21.02*
40.96*
1/--страниц
Пожаловаться на содержимое документа