Transcript Does the Cutting back of the Public Sector Improve

17th ICABR Conference
18th -21th June, 2013, Ravello, Italy
Manuela Coromaldi – University of Rome “Niccolò Cusano”
Giacomo Pallante - University of Rome Tor Vergata
Sara Savastano - University of Rome Tor Vergata
Building blocks
• Background and motivations
• Our aim
• Literature
• Data description
• Empirical model
• Estimation results
• Conclusions
Background and motivations
 Evidence from the Green Revolution show a strong impact of
the adoption of improved varieties on agriculture productivity
growth (Evenson and Gollin, 2007) , on food production and
food security (Tillman, 1999).
 High-yield-crop of Asian Green Revolution were bred to work
better with greater applications of fertilizer than traditional
varieties and to work better on irrigated land (Larson et al.
2010).
 Level of adoption in Sub-Saharan Africa has been limited.
Our aim
 This paper aims at testing the factors affecting productivity
and agricultural biodiversity in terms of local and improved
varieties.
 As the Sub-Saharan Africa countries heavily rely on traditional
technologies, we study the potential of local species who are
expected to perform better in marginal production
environments and who can be more resistant to climatic
stress.
 We adapt the Steinfeld approach (2000) to empirically verify
the effect of intensification of crop production on the
opportunity costs for smallholders of conserving local species
when there are factors market and agronomic constraints that
inhibit the efficient and economic use, and availability of
improved variety.
Literature (1)
 Evidences from Sub-Saharan Africa show low level of
adoption of improved technology and inappropriate use of
inputs:
 40% of fertilizer is used on maize (Morris et al., 2007; Heisey and
Norton, 2007)
 Sub-Saharan Africa : the average dose is about 17 kg/ha of nutrients
 Developing countries: 100 kg/ha
 Developed countries: 270 kg/ha on the same crop
 Sub-Saharan Africa failing in adoption of agricultural
technology is due not only to market constraints but also to
low responsiveness of marginal land respect to external
inputs (Sanchez et al., 1997).
Literature (2)
 The use of improved seeds has increased the farmers yield in Asia and Latin
America, but has also let arise diffused concerning about the inter and intra
species genetic diversity erosion (Harlan, 1992).
 The variability is then dramatically reduced, when single lines or F1 hybrids
varieties become dominant to the detriment of diversity-rich landraces
(Lipper and Cooper, 2009).
 When the farmers preserve a wide range of local landraces, they conserve a
genetic portfolio that minimizes a set of private and public risks (Weitzman,
2000).
 To the point where smallholders allow the conservation of local public goods
as the resilience of the local ecosystem to face biotic and abiotic stresses
(Jarvis et al., 2007) or global public goods as the maintenance of a pool of
genetic material and the option value to use it (Bellon, 2009), those farmers
are “custodians” of the agricultural biodiversity conservation (Silveri and
Manzi, 2009).
Literature (3)
Authors and year
yield
Quantity of fertilizer
Soil characteristics
Matsumoto et al.
(2013)
Nkoya et al.
(2010)
maize (local variety)
12.4 kg of base fertilizer plus 10 kg top-dressing fertilizer for
a quarter acre of maize
-1
80 kg N ha , 5 t ha of manure and 100% retention crop
residue
poor soil quality
Glover Amengor
and Tetteh (2008)
tomato
460 and 375 g ha-1 of lindane and propoxur insecticides
ferric acrisols of
Ghana
Muthomi et al.
(2007)
legume
mixing inorganic fertilizer at a rate of 200 kg/ha with 0.7 L/ha
and 2 kg/ha of insecticide and fungicide respectively
rainy and non rainy
area of Kenya
Morris et al.
(2007)
Heisey and
Norton (2007)
maize
17 kg/ha compared to the developing country average of 100
and the industrialized country average of 270 kg/ha
Shankar and
Thirtle (2005)
cotton
1 L/ha of insecticide
South Africa
Jama et al. (1997)
maize
10 kg ha-1 of Phosphorous
alfisols and ultisols
of humid and
subhumid areas
sorghum
50 kg N ha-1
entisols and
vertisols area of
Zimbabwe
250 kg P ha-1 plus urea at 60 kg N ha-1
Kenyan alfisols and
oxisols
15-30 kg ha-1 of Phosphorous (P)
sandy soil in
semiarid regions
100 kg ha-1 of Nitrogen (N)
Sandy soil in
semiarid regions
Nyakatawa
(1996)
Hoekstra and
Corbett (1995)
Bationo and
Mokwunye
(1991)
Mugwira and
Mukurumbira
(1986)
Agricultural intensification index (1)
 The agricultural intensification concept is
defined as the increase of agricultural
production on a fixed portion of land, as
opposite to a raising in production due to
expansion of land (Netting, 1993).
 It involves the substitution of other inputs, as
well technology, for a constant land in order to
let the yielding function rise (Brookefield, 1993).
Agricultural intensification index (2)
 Herzog et al. (2006) developed an intensification index with
the aim to study the effect on biodiversity at the landscape
level (Agricultural Intensity index, AI ).
where yi is a variable of agricultural intensification, n is the
number of intensification variables and yimin and yimax are the
minimum and maximum value of the agricultural
intensification variables in the sample.
Agricultural intensification index (3)
The use of AI holds two interesting features:
1.
The input’s application on per hectares basis is a widely
comparable measure over regions.
2.
It is a relative index. Since it accounts for maxima and
minima administering of inputs into a community, the
differences in soil qualities, climatic conditions and farm’s
practices are endogeneized so that the context-specific
traits are taken into consideration.
Agricultural intensification index (4)

We estimate the AI index for each households by using the
following three variable of agricultural intensification:
1. Chemical fertilizers (kg/ha): sum of Nitrate, Phosphate,
Potash and Mixed.
2. Pesticides
(Kg/ha):
sum
of
Miticides/Acaricides,
Fungicides,
Herbicides and growth regulators.
3. Years of fallow (number of years).
Insecticides,
Rodenticides,
Data description (1)
 Data are taken from the Uganda National Panel
Survey 2009/10 (UNPS).
 UNPS is carried out annually over a twelve-month
period on a nationally representative sample of
households.
 The survey includes 3,123 households that were
distributed over 322 enumeration areas.
Data description (2)
Traditional Variety
Improved Variety
Net Crop Income/ac
4733.074
4914.85
Area Operated
2.243***
3.144***
48.198***
45.703***
5.007***
5.692***
0.996
0.993
0.296***
0.197***
17.369
17.481
72.663*
77.117*
22.046***
15.332***
0.392
0.483
0.519***
0.693***
270.771***
322.134***
0.096*
13.938*
0.165
0.224
Class of lenght of fallow
9.233***
8.959***
Intens. ind. with fallow,chem,pest
30.861**
30.060**
Age of household head
Avg adult yrs of education
Household has access to road
Head is female
N. of HH using HYV in the district
Dummy soil good quality
Dummy slope steep
Area Rented in irrespective of the
season and the use
Household hired labor input
Person Days of family labor
Sum of N,P,Po,Mix per ha
Sum of ins,mit,fun,rod,nem per ha
* significant at 10%; ** significant at 5%; *** significant at 1%
Conceptual framework (1)
• We implement the conceptual framework developed
by Steinfeld (2000) and adapted by Narloch et al.
(2011);
• the aim is to verify as the gross profit of farmers from
the use of traditional or improved variety changes
under the increase of the intensification degree of
the farming system.
Conceptual framework (2)
Gross
profit
Improved gross
profits
Traditional gross
profits
Degree of
intensification
0
200
400
600
800
The empirical results
0
20
40
60
intens. ind. with fallow,chem,pest
improved
improved fit
80
traditional
traditional fit
17
Estimation strategy (1)
• To control for selection bias in the assessment agriculture
productivity in the adoption of improved variety, we make use of
Heckman's two-step estimation (Heckman, 1978).
• In the first stage, we compute a Probit regression in order to
estimate the probability that a given farmers will adopt a new
technology.
• This regression is used to estimate the Inverse Mills Ratio (IMR) for
each farmers, and this will be used as an instrument in the second
regression where we will analyze the determinant farmers efficiency
through an ordinary OLS
Estimation strategy (2)
• Following Maddala (1983) Amemiya (1985) and Johnston and
DiNardo (1997), we will use different instruments to control for
identification problems
 The first stage Heckman procedure, namely the adoption equation
is a standard Probit regression of the form:
(1)
Pr ob(Y  1 | X )   (Z )
where Y indicates adoption (Y=1 if farmers adopted the improved
technology, 0 if they did not adopted). Z is a vector of explanatory
variables such as household characteristics, asset’s endowment,
credit access, and dummy for organic inputs used, fertilizers,
geographical dummies).
Estimation strategy (3)
 In the second stage, we correct for self-selection by incorporating a
transformation of these predicted farmers probabilities as an
additional explanatory variable.
 The agriculture productivity equation may be specified as
AP  X  u
*
(2)
where AP* denotes an underlying level of technical efficiency,
which is not observed if the farmers did not adopted.
 The conditional expectation of efficiency if the farmer adopted is:


E AP | X ,Y  1  X  Eu | X ,Y  1
*
(3)
Estimation strategy (4)
 Under the assumption that the error terms are jointly normal, we
have:


E AP* | X ,Y  1  X  u (Z )
(4)
 where ρ is the correlation between unobserved determinants of the
probability of adopting the technology ε and the unobserved
determinants of farmers technical efficiency u;
 and λ is the Inverse Mills Ratio.
Table 1: Two-step estimates of the adoption equation
Variable
#1
Intens. ind. with fall~t
Log operated area
Log age of hh
Log age-squared of hh
Log avg adult yrs of e~n
Log distance to road (~)
Household has access t~d
head is female
Dummy soil good quality
Dummy slope steep
Log area rented in
Log h hired ag labor i~t
Log pers-days family l~r
Number of HH using HYV~i
Constant
OLS
0.0257*
-1.0683***
0.7987
-0.1600
0.0766
0.0988
1.4577*
0.0207
0.0057**
0.0074***
-0.0309
0.4387*
0.6788***
-0.0178
0.2971
PROBIT
0.2769***
5.2589*
-0.7339*
0.1347
-0.0091
-0.3202
-0.2110*
0.0007
0.0001
0.0566***
-10.6277*
dimprov
Intens. ind. with fall~t
Log operated area
Log age of hh
Log age-squared of hh
Log avg adult yrs of e~n
Household has access t~d
head is female
Dummy soil good quality
Dummy slope steep
Log area rented in
Log h hired ag labor i~t
Log pers-days family l~r
Number of HH using HYV~i
Log distance to road (~)
Constant
HECKMAN
0.0506**
-1.1029***
6.5042
-1.0004
0.1916
2.3625
-0.1032
0.0084**
0.0049
0.0523
0.4909
0.7373***
-11.3548
-0.0032
0.2017**
2.8182
-0.4162
0.0611
-0.3494
-0.2117*
0.0000
-0.0000
0.0991
0.5140***
0.1885***
0.0567***
-0.0216
-6.8310
Statistics
N
rho
sigma
lambda
1226
1226
1226
-0.1475
2.4492
-0.3613
Conclusions
 Low level of technology adoption in Uganda, but increasing probability if





neighbor farmers’ adopt.
The perception of technology profitability increases with the number of
households using improved seeds in the district.
Positive impact of Intensification Index on productivity but level of input use
is still very low among adopters. Need to increase extension services for
maximizing use of scarce seeds and agriculture productivity.
Evidence of the presence of an inverse farm size productivity relationship,
but larger farmers have higher probability to adopt.
Adoption does not lead to substantial difference in agriculture productivity.
Extensions: Estimate Heckman Model with 2 selection equations:
 Probability of adopting HYV
 Probability that once adoption has taken place, fertilizer use will follow.