Transcript Lambarraa-serra-gil-aeea07.ppt (863Kb)

De la Economía Agraria a la Economía Rural y Agroalimentaria
TECHNICAL EFFICIENCY AND PRODUCTIVITY
ANALYSIS OF SPANISH CITRUS FARMS
Fatima Lambarraa, Teresa Serra and José M. Gil
Centre de Recerca en Economia i Desenvolupament Agroalimentaris, CREDA-UPC-IRTA
PRESENTATION OUTLINE
I. THE RELEVANCE OF THE CITRUS SECTOR
II. OBJECTIVE
III. METHODOLOGY OF ANALYSIS
IV. EMPIRICAL APPLICATION
V. EMPIRICAL RESULTS
VI. CONCLUDING REMARKS
I. RELEVANCE OF CITRUS SECTOR AND MOTIVATION
The leading citrus fruit producing countries are Brazil, the
Mediterranean countries, the United States and China. Within
the Mediterranean area, Spain is the first producer with more
than 5.9 million tons (57% of EU production and 6% of the
worldwide production).
Oranges are the most relevant citrus fruit produced in Spain
(48% of EU production and 5% of worldwide production).
I. RELEVANCE OF CITRUS SECTOR AND MOTIVATION
Citrus fruits are the first fruit crop in international trade in
terms of value with more than 10 millions tons in 2004.
The EU is the main destination as well as the main supply
region, with almost half of the world imports and more than
40% of world exports.
The Mediterranean region plays a prominent role as a world
fresh citrus exporter (60% of global fresh citrus fruits
exports). Spain is the first exporting country with almost 25%
of total exports in the world (FAO 2005).
II. OBJECTIVE
 The objective of this study is to analyze the Technical
Efficiency (TE) for a sample of Spanish farms specialized in
citrus production and …
 To decompose Total Factor Productivity (TFP) growth into
its various components (using a primal approach).
III. METHODOLOGY OF ANALYSIS
Stochastic frontier model
To measure the TE with which farms are operating, we use the
stochastic frontier methodology (SFM) (Aigner, Lovell and Schmidt
1997, Meusen and Van den Broeck 1977 ) :
We consider the general stochastic production frontier function in the
presence of panel data:
yit = f(xit β,t)e
where v it ~
N (0,  v2 )
& u it ~ N  (0,  u2 )
vit -uit
III. METHODOLOGY OF ANALYSIS
Stochastic frontier model
u it captures the effects of statistical noise outside the firm's control
u it is a one-sided, non-negative component associated with outputoriented technical inefficiencies.
We adopt the temporal pattern of technical inefficiency by Battese and
Coelli (1992):
.


uit  exp    t  T  ui
Where ξ captures the temporal variation of individual output-oriented technical
efficiency ratings, and t  1, 2, ..., T 
III. METHODOLOGY OF ANALYSIS
Decomposition of TFP growth
We measure productivity change and determine its various sources
following the primal approach by Kumbhakar and Lovell (2000):
εk °
 εk
 °
TFP = TΔ+(ε - 1)( )x k +  ( )- S k  xk +TEΔ
ε

k
k  ε
°
Where:
Technical change:
T 
f ( x, t ;  )
t
III. METHODOLOGY OF ANALYSIS
k 
SC  (  1) ( ) X k

k
Scale component:
Where:
ε = ε(x,t; β)=  εk (x,t; β)
&
εk = εk (x,t; β)=
k
Allocative inefficiency:
Where:
 εk
 °
AE =  (
) - S k  xk
ε

k 
E =  wk xk
&
Sk =
k
Technical efficiency change:
u
TE  
t
wk xk
E
xk  f(x,t; β) xk 
f(x,t; β)
IV. EMPIRICAL APPLICATION
We use farm-level data taken from the FADN (Farm Accounting
Data Network) for the period 1995-2002.
We also use regionally aggregated derived from the Spanish
Ministry of Agriculture (MAPA) & Eurostat (MAPA provided land
prices & Eurostat provided other input and output price indices).
We chose our sample of farms from the specialist citrus fruits
group. Specifically, we concentrate on those farms whose orange
sales represent more than 70% of citrus sales.
Our sample is composed by 859 observations.
We use Frontier 4.1 to estimate SFM & SAS 9.1 to decompose
TFP Change.
IV. EMPIRICAL APPLICATION
The production frontier function is approximated by the quasiTranslog functional form (Fan, 1991):
yit = β0 e
βt t
K
 βk + βkt t  vit -uit
x
e
 kit
k=1
Xit : (1x4) vector of inputs
X1: Fertilizers and pesticides,
X2: Other variable crop-specific inputs,
X3: Hectares occupied by olive groves,
X4: Labor input measured in labor hours per year.
V. EMPIRICAL RESULTS
Table 1. Maximum Likelihood Estimates of the Production Frontier Function
Production function
Parameter
Estimate
Standard Error
α0
0.6038
(0.0395)*
αK
0.3242
(0.0738)*
αL
0.1841
(0.0370)*
αF
0.2201
(0.0362)*
αO
0.1040
(0.0279)*
αKT
-0.0586
(0.0610)
αLT
0.1304
(0.0386)*
αFT
0.0609
(0.0348)*
αOT
-0.0060
(0.0312)
αT
-0.2628
(0.0460)*
αTT
-0.2330
(0.0330)*
3.2963
(0.5244)*

0.9636
(0.0066)*
ξ
0.0270
(0.0080)*
Technical inefficiency model
u2
Note: L refers to labour, K to Land, F to Fertilizers and O to other costs.
* indicate that the parameter is significant at 5%.
V. EMPIRICAL RESULTS
Table 2. Output Elasticities for Spain citrus-Growing Farms, 1995-2003
1995
1996
1997
1998
1999
2000
2001
2002
2003
Land
0.416
0.376
0.352
0.335
0.322
0.311
0.302
0.294
0.287
Labour
-0.02
0.068
0.121
0.159
0.188
0.212
0.232
0.249
0.265
Fertilizers &
pesticides
0.124
0.166
0.191
0.208
0.222
0.233
0.242
0.250
0.257
Other costs
0.113
0.109
0.106
0.105
0.103
0.102
0.101
0.101
0.100
Output Elasticities
V. EMPIRICAL RESULTS
Table 3. Model Specification Tests for citrus farms
Hypothesis
LR teststatistic
Average Production Function, i.e., γ=μ=ξ=0
950.54
32  7.81
Aigner et al., (1977) SPF model with time-invariant
output-oriented technical efficiency, i.e., μ=ξ=0
32.53
22  5.99
Aigner et al., (1977) SPF model with time-varying
output-oriented technical efficiency, i.e., μ=0
22.61
12  3.84
Time-invariant output-oriented technical efficiency, i.e.,
ξ=0
10.18
12  3.84
12.84
52  11.1
12.55
42  9.49
64.17
62  12.6
Constant returns-to-scale, i.e.,
  1
j
j
and

j
 jT  0
Hicks-neutral technical change, i.e.,  jT  0 j
Zero-technical change, i.e., T  TT   jT  0 j
Critical Value
(a=0.05)
V. EMPIRICAL RESULTS
Table 4. Measures of Technical Efficiency for Spanish citrus farms, 1995-2003
TE
95
96
97
98
99
00
01
02
03
<20
4
8
10
7
7
7
5
8
5
20-30
2
7
7
3
5
7
4
2
1
30-40
10
9
8
5
6
1
3
4
4
40-50
3
13
10
9
8
7
8
6
6
50-60
15
12
13
9
7
8
9
9
8
60-70
10
12
10
11
12
10
7
6
6
70-80
16
16
17
16
15
15
14
15
13
80-90
23
28
29
29
32
32
32
32
32
90>
4
6
6
5
7
7
8
7
6
Mean
63%
60%
60%
64%
64%
64%
67%
66%
69%
V. EMPIRICAL RESULTS
Table 5. Decomposition of TFP Growth for Spanish citrus Farms
year
TFP
TEC
SC
AE
TC
1996
0.024222
0.005057
-0.00792
0.027073
0.000012
1997
0.145540
0.002790
0.021737
0.121088
-0.000074
1998
0.092292
0.004003
-0.03599
0.12438
-0.0001005
1999
0.025411
0.001828
0.030043
-0.00632
-0.000139
2000
0.0754524
0.005234
-0.015332
0.085737
-0.000185
2001
-0.031069
0.001128
0.018885
-0.05086
-0.000222
2002
-0.146991
0.002837
0.008711
-0.15828
-0.000259
2003
0.0369498
0.002598
-0.011398
0.046036
-0.000285
1999-2003
average
0.0277259
0.0031843
0.0010918
0.0236067
-0.000157
VI. Concluding Remarks
Our analysis assesses the efficiency and productivity growth of
Spanish farms specialized in citrus produce.
The Spanish citrus sector production function is characterized
by decreasing returns to scale, making the increase of farm size
unattractive.
Results of the Stochastic frontier model suggest an
improvement in efficiency levels during the period studied,
from 53% in 1995 to 69% in 2003. The estimated average
efficiency level is about 64.11% with 38% of farms in the
sample reaching a score greater than 80%.
VI. Concluding Remarks
The TFP growth indicates an increase in average productivity
of about 2.7 % per year during the period of study.
Technical efficiency change, allocative inefficiencies and scale
effects constitute the most relevant components of this growth.