2. Overview of presentation
1. Motivation
2. Literature Review
3. Sector and Data
4. Theory
5. Empirical Strategy and Results
6. Discussion and Policy Implications
3. Motivation
Relevance
Environmental effects of agriculture
Political will to increase adoption of sustainable
farming techniques
Questionable sustainability of organic agriculture
in Africa?
Our contribution
Link between certification and adoption of agro‐ecological practices
…and between adoption and ROI (not yield)
4. Research Question(s)
Does organic certification increase agro‐ecological
practice use?
How does the intensity of agro‐ecological practice use
influence the ROI?
5. Literature Review
Impact of certification on adoption
“usual impression”: Organic farming in Africa means to stop using chemicals,
but not to adopt alternative soil fertility management practices. Small‐scale
farmers produce ”organically by default”.
Bolwig et al. (2009) and Blackman and Naranjo (2010): some adoption
Impact of organic certification on ROI
Modest positive effect on farmer welfare (poverty,
hh income,…)
Possibly negative effect on yield
Impacts of agro‐ecological practice use
Yield increases? (1999; Pretty et al., 2006; Branca et al., 2011; Onduru et al.,
2002)
Lower production costs? (Dasgupta et al., 2007)
Revenues increases? (Bolwig et al., 2009)
Environmental effects
7. Actors in the Pineapple Sector
Smallholders
Local buyers
MoFA
Development
Agrencies, NGOs
Export
Processors
companies
Large farms
8. Data
Household survey of 386 farmers
January to March 2010
75 villages from 6 districts in Central, Eastern and
Greater Accra region
185 organic (from 9 farmer associations),
201 conventional (from 14 farmer associations)
Stratified random sampling
Districts with high production
Percentage of certified groups in districts
9. Descriptive Statistics
Definition Variable Organic Convent. T‐Stat.
Farmers Farmers
(N=185) (N=201)
Gender of household head (0=female, 1=male) GENDER 0.891 0.982 3.51***
Age of HHH AGE 46.313 42.970 ‐2.82***
Household size (persons living in household) HHSIZE 5.230 5.917 2.35**
Maximal educational level in household EDUC 3.566 3.941 3.40***
Farm size (acre) FSIZE 10.35 18.720 5.02***
Pineapple land (acre) PINLAND 4.014 3.066 ‐2.07**
Share of land owned OWNLAND 0.549 0.204 ‐7.628***
Access to credit (0=no, 1=yes) CREDIT 0.317 0.232 ‐1.78*
Bank account with more than 200 GHS BANK 0.339 0.512 3.21***
Number of durable goods owned WEALTH 4.765 8.481 10.875***
Relation to the local government GOVERN 2.257 1.774 ‐4.27***
Self‐stated openness to innovation and risk RISK 0.152 ‐0.166 ‐3.01***
People that are met regularly MEET 1.175 1.196 0.45
Years of experience in pineapple farming EXPER 11.557 11.595 0.05
Distance to the closest local market (hours) DIST 0.698 0.804 1.59
Assistance or training for farming received ASSIST 0.732 0.708 ‐0.5
11. Theoretical Framework
Basic utility maximization framework
Choice of intensity to maximize the expected net utility
Utility function depends on net returns (Π) which are dependent on the level of
output Q, output prices Po, inputs I and their prices Ii, and farm and household
characteristics Z:
E Π E ,
Π =Π , , ,
with Ae = agro‐ecological intensity
Π =
and probability of adoption for each level:
∗ ∗
Π 1 Π Π
12. Empirical Strategy
1) link between certification and adoption of agro‐ecological
practices
What is the impact of organic certification on the use of agro‐
ecological practices?
2) link between adoption of agro‐ecological practices and ROI
What is the impact of the intensity of agro‐ecological practice use on
the ROI?
14. 2) Impact of use of agro‐ecological
practices on ROI
Generalized propensity score matching (Hirano and Imbens, 2004)
;
We observe (exogenous controls), Ti (continous treatment variable) and
(outcome variable) associated with each treatment level.
We want to measure the average dose response function (DRF) which relates
the unbiased potential outcome to each treatment level :
∀
Assumption: weak unconfoundedness: ⎸ ∀
Generalized propensity score (GPS) : ,
average dose response function (DRF):
, ,
Where , ,
15. 2) Impact of use of agro‐ecological
practices on ROI (ctd.)
Our model:
GPS estimation using normal distribution of the logarithmic treatment
• Kolomogorov‐Smirnov test for normality for logarithm of the
treatment variable (Hirano and Imbens, 2004)
Common support condition: 278 farmers on support
quadratic approximation for average potential outcome at each
treatment level
18. Conclusion
Background: Why is the extent of adoption low?
A common concern is that organic Availability of organic
farmers in developing countries material
remain in an unsustainable low‐ Transport costs
yielding state of ”organic‐by‐default”
production
Agriculture is responsible for
environmental damage
Potential Solution
Results: Use certification or involved market
linkages
Certification has a large
influence Support by buyers
Economic barriers to required agro‐ecological
intensification dependent on practices
intensity