martha negash & johan swinnen center for economic performance and institutions (licos), kuleuven
TRANSCRIPT
Biofuels and Food Security: Micro-
evidence from Ethiopia
Martha Negash & Johan Swinnen
Center for Economic Performance and Institutions (LICOS),
KULeuven
1. MotivationImpact of biofuel expansion
views:
- worsen food insecurity (von Braun, 2008; Mitchel, 2008)
on the contrary:
- high food prices - not always bad
- biofuels stimulates economic growth & reduce poverty (case-Mozambique) (Arndt et al, 2010)
- reduce the incidence of poverty & support food self-sufficiency goals (Huang, et al. 2012)
‘food vs fuel debate’
- weak land governance & property rights – risk to vulnerable
hhs (Cotula et al 2010) “Fueling exclusion” -> conflict
Foreign land investment:
- investment brings inefficiently utilized/under-utilized land
- emp’t & income effect
- cheaper energy source to remote rural areas (quite an
issue ‘energy poor countries’)
‘land grab vs land investment’
Other concern:
Evidence in current literature:
- based on aggregate economic wide simulations
or qualitative studies
- largely focused on developed economies
- impact analysis on poor smallholder context -
limited
Research questions:
1- identify factors associated with biofuel crop adoption
decisions?
2- how participation decision influences food security status?
Survey– privately organized castor (biofuel feedstock crop)
outgrowers in Ethiopia
Hunger index
Ethiopia
modern energy (extremely poor)
food (alarming hunger)
unutilized/underutilized land low potential areas
good case to study
Energy poverty index
Source: IFPRI, 2010Source: Nussbaumera et al., 2011
Castor outgrower scheme in Ethiopia
Advantages-can be preserved on the field relatively for longer periods - allows piecemeal collection of seeds
-good for soil fertility
-contract farmers may record higher productivity in food crops through
– higher input use - spillover effects - crop management practices
Disadv.- Invasive species - castor has no other use in the area – (bargaining power of farmers ??)- default is mainly from redirecting input use for other crops
Supply chain
Raw seed export
Company -> via supervisors -> input loan & seed -> farmers
Farmers-> village centers-> via supervisors -> company -> export-> China processors
2. DataSampling frame all villages in range of
1100– 2000 m.a.s.l. covered by the program included in our sampling frame
Sample size- 24 villages randomly
selected- total of 478 household
- 30% participants
Participant/Adopters a household that
allocated piece of land for castor & entered contractual agreement w/t the company
Source: FEWS, 2010
Most biofuel projects are located in dry & low land areas of the country
Ade Dewa Mundeja
Anka Duguna
Degaga Lenda
Fango SoreSura Koyo
Tura Sedbo
Olaba
Mayo Kote
Hanaze
TulichaSortoBade Weyde
Bola GofaSezgaUba PizgoZenga ZelgoSuka
Tsela Tsamba
Lotte Zadha Solle
Gurade
Bala
Zaba
.1.2
.3.4
.5.6
Ad
op
tion
ra
te
0 20 40 60 80 100Distance to near by town
-better access
-better infras
-dairy supply to
town
- poor acce
ss;
- poor infra
s (tel.,
electric)
- no alternativ
e cash
crop
Sampled villages & castor bean adoption
-distant villa
ges
-alternative cash crop
– fruits
& ginger
Village level observation
- dissemination of the castor crop into inaccessible & remote places
- widespread adoption rate (20-33%) in three years of promotion
- unlike low rate of new crop or fertilizer adoption rates in developing countries
- villages with limited alternative cash crop markets show higher adoption incidence
No food gap Less than one month
One to three months
More than three months
0
10
20
30
40
50
60
Non participants
Participants
0.2
.4.6
.81
Cum
ilativ
e fr
act
ion
of h
ous
ehol
ds
5 6 7 8 9 10Log of total food consumption (kcal energy equivalent)
Participants Non-participants
Figure : Food gap (number of months)
******
Figure: Per capita food consumption
Descriptive (outcome variables) (1/2)
%
measured by number of food shortage months – decline in value improvement in welfare
total consumption in energy equivalent (kcal/person/day) – increase in value ->improvement in welfare
Participants Non-participants |t/chi-stat|
Household wealth variables
Owned land size (in ha) 0.93 0.72 3.54***
Own land per capita 0.15 0.13 1.00
Farm tools count (Number) 4.20 3.84 1.48
Proportion of active labour 0.49 0.51 0.99
Access related variables
Formal Media (TV/radio/NP) main info. source (1=yes) 0.27 0.18 1.73***
Fertilizer use (kg/ha) 33 24 9.0***
Borrowed cash money during the year (1=yes) 0.42 0.36 1.14
Distance from extension center (Minutes) 27.53 27.80 0.10
Contact with extension agent (Number of visits) 12.63 11.08 0.98
Household characteristics
Gender of the HH head (1=female) 0.06 0.14 2.95***
HH head attended school (1=yes) 0.60 0.50 1.67*
Family size 6.87 6.10 2.98***
Descriptive (explanatory variables) (2/2)
* p<.1; ** p<.05; *** p<.01
3. Estimation
Effect of castor contract participation on income
- represent– participation as a regime indicator variable
(1)
Regime 1:
Regime 0:
(2)
(3)
If cov (ui , ℇ1i ) and/or cov (ui , ℇ2i ) are statistically significant,switching is endogenous, self-selection - on obs. or unobser. or both).
Identification – assume error terms are jointly distributed
IV –improves identification – eligibility & past adoption history (farmers choice)
Participation decision position
Treatment effect Regime 1
(Participate) Regime 2
(Not participate) Participant (a) E(𝑦1𝑖ȁ�𝑑𝑖,𝑥1𝑖 = 1ሻ 𝛽1𝑋1𝑖 + 𝐸ሺ𝜀1𝑖ȁ�𝑑𝑖 = 1ሻ
=
𝛽1𝑋1𝑖 +ቆ𝛿𝜀1𝑢𝛿𝑢2 ቇቆ
𝜙ሺ𝑧Ƹ𝑖ሻΦሺ𝑧Ƹ𝑖ሻቇ
(c) E(𝑦2𝑖ȁ�𝑑𝑖,𝑥2𝑖 = 1ሻ 𝛽2𝑋1𝑖+ ሺ𝜀1𝑖ȁ�𝑑𝑖 = 1ሻ =
𝛽2𝑋1𝑖 +ቆ𝛿𝜀2𝑢𝛿𝑢2 ቇቆ
𝜙ሺ𝑧Ƹ𝑖ሻΦሺ𝑧Ƹ𝑖ሻቇ
(a)-(c) = TT
Non-participant (b) E(𝑦1𝑖ȁ�𝑑𝑖,𝑥1𝑖 = 0ሻ 𝛽1𝑋2𝑖 + 𝐸ሺ𝜀2𝑖ȁ�𝑑𝑖 = 0ሻ
=
𝛽1𝑋2𝑖 −൬𝛿𝜀1𝑢𝛿𝑢2 ൰ቀ 𝜙ሺ𝑧Ƹ𝑖ሻ1−Φሺ𝑧Ƹ𝑖ሻቁ
(d) E(𝑦2𝑖ȁ�𝑑𝑖,𝑥2𝑖 = 0ሻ 𝛽2𝑋2𝑖 + 𝐸ሺ𝜀2𝑖ȁ�𝑑𝑖 = 0ሻ =
𝛽2𝑋2𝑖 − ൬𝛿𝜀2𝑢𝛿𝑢2 ൰ቀ 𝜙ሺ𝑧Ƹ𝑖ሻ1−Φሺ𝑧Ƹ𝑖ሻቁ
(b)-(d)=TU
‣ can substitute historical comparative data –but useful in the absence of such data
Source: Verbeek, 2012; Di Falco, et al. 2011; AJAE
Endogenous Switching Regression Model
allows estimation of heterogeneous effect of covariates
using the information contained in the distribution functions of the error terms & their covariance, allows predicting counterfactual effects
4. Results Question 1
First stage: selection to participationVariabel Marginal effects
Per capita owned land size (ha) 1.60*** Per capita owned land size squared -2.26** Pr of maize before planting made (in birr) -0.12** Gender of the head (1=Female) -0.14* Household head attended school (1=yes) 0.08 Log of number of social contact and friends -0.05** Media (1= main info source) 0.10** Pre program asset indicator 0.09** Farmers choice indicator (eligibility*past adoption) 0.05* Log of distance from extension center -0.02 Log of number of gov. extension visits 0.01 Family member with non agri inc source (1=yes) -0.05 District dummies yes Other controls (age, age squared, labour size,
enset, livestock, plot distance) yes N 476
distance from the village center
gov. extension service
(---) Maize price Female
(+++) Land Media Asset
(non-significant)
Food gap estimation Participant Non-
participants
Land per capita (ha) -2.799** -0.221
Log of agricultural income per capita -0.063 -0.074*
Household attended schooling (yes=1) -0.03 -0.140**
Family size -0.053*** -0.014
At least one member works off-farm (yes=1) -0.109* -0.113**
Family in polygamy (yes=1) 0.412*** 0.177
Own livestock (TLU) per capita -0.092 -0.165**
Borrowed cash during the year (yes=1) 0.212*** 0.100*
District dummy Yes Yes
Other control
Sigma (δ)-1.09*** -0.77***
ρ-0.22* 0.40**
N476
Likelihood ratio test of independent equations ( X2)
2.98*
differentiated significance & magnitude of coefficients
e.g. family size & livestock coefficients have different signs
opposite sign of ρ – suggest rational sorting into participation
Question 2
Treatment effect
Sub-sample
Decisions stage Treatment Effect
To participateNot to
participateLog of food gap (months)
Households who participated (a) 0.84 (c) 1.20 (treated) -0.37***
Households who did not participate (b) 1.04 (d) 0.98 (untreated) 0.06***
Log per capita annual food consumption (kcal/capita/day)
Households who participated (a) 7.86 (c) 7.59 (treated) 0.27***
Households who did not participate (b) 7.23 (d) 7.41 (untreated) -0.18***
Participants reduction in food gap, 37%, (-11 days) increase in consumption, 27%
Non-participants do not benefit, rather food gap would increase, 6% (+2 days) reduction in consumption, 18%
5. Policy implications
(Question 1) Determinants of adoption: assets are key factors for adoption
adoption of biofuel declines with price of food crop
physical distance showed no significance unlike most studies
Policy implication:
privately organized technology transfer –may efficiently surpass physical barriers
(Question 2) Effect of participation:
impact is heterogeneous
participants are better-off producing castor than if they had not
non-participants would have been worse-off if they had participated
Policy implication:grant farmers more choice
explore castor’s potential contribution to narrow food gap /smooth consumption/
Thanks!