Professional Documents
Culture Documents
RESEARCH INSTITUTE
1
Presentation outline
The presentation has the following sections
• Context of study
• Data and methods
• Storage and associates of storage decisions
• Losses and associates of losses during storage
• Conclusions
4
2. Data and methods
Methods
• Descriptive analyses are used to study
• Proportion of households (HHs) that store grain crops and that
suffered losses during storage
• Quantity/periods of crops stored
• Estimate losses during storage
• Regional differences in storage and losses during storage
• We employ:
• Probit analysis to indicate associates of crop storing decisions
• Tobit analyses to study crop losses during storage
5
2. Data and methods
Data
• Presentation includes results obtained from Feed the Future (FtF)
midline survey data
• Study will include AGP midline survey (8000 HHs, 2012/13 meher)
• The FtF midline survey
• Was conducted in May 2015 (pertains to 2014/15 meher season)
• Includes 8/4, 22/7, 29/13, & 20/13 woreds/zones in Tigray, Amhara,
Oromiya, & SNNP
• 6,691 HHs surveyed; analyses include 5,766 with relevant data (10%,
28.3%, 34.7%, & 27% of HHs in respective regions)
• Caveats of FtF midline survey data
• Data available only on crop losses during storage (PHL broader)
• Households’ own estimate of losses used, which may be biased;
• As a result, estimates on PHL need to be carefully interpreted
6
2. Data and methods…Contd.
• Household total crop area averaged about 1.2 ha
• Area without 2, 2, and 1 zones in Tigray, Amhara & Oromiya about 1 ha
• 19.7% of total sample area covered with teff, maize 15.7%, wheat,
pulses, sorghum, and oilseeds for 12%, 11%, 9.3%, & 7.6%.
• Crop area around 0.4 ha in most crops except sorghum (0.65) and
teff (0.57) while oilseeds (0.9) area is high
• Oilseeds area dominated by Western Tigray, North Gondar, Horo Gudru W.
• Without these zones average oilseeds area 0.36 ha
• HHs cultivate 4 crops and this is similar across regions
• 3.7 in Oromiya, 3.8 in Tigray, and 4.2 in Amhara & SNNP
• 65%-75% of output consumed (slightly higher in sorghum & maize);
• Most oilseeds is sold (70%) & teff is the most commercialized cereal
7
2. Data and methods…Contd.
Table 1. Summary of crop area, output, and crop utilization
8
3.1 Crop storage
Households storing crops
• 40%-50% of cereal growing HHs store crops and proportions are
lower in pulses (36.5%) and oilseeds (23%)
• Have higher average output in all crops relative to non-storing HHs
12
3.2 Associates of crop storage…Contd.
• The decision to store crops positively associated with:
• HH size (higher consumption needs, availability of labor);
• Storage likely in HHs with female, younger, and more educated heads
• Seasonal price gap (opportunity cost of not storing higher) and
• Number of months between harvest & beginning of next season (non-
linear)
• Crop storage negatively associated with
• HH wealth (can afford purchasing food latter/less subsistence),
• Proximity to urban centers (more integrated with market/access to
food market), and
• Long-run average annual humidity in area (higher humidity may lead to
crop rotting and pest infestation).
13
3.2 Associates of crop storage…Contd.
Table 3. Average marginal effects of decision to store equation (probit)
Variables Aver. Marg. Eff. SE
HH head female (=1 if yes) 0.082*** 0.015
Age of HH head -0.001** 0.000
Education (literate, primary incomplete) 0.021* 0.012
Education (literate, primary complete) 0.048*** 0.015
HH size 0.011*** 0.002
HH wealth index -0.042*** 0.008
Seasonal producer price gap 0.025*** 0.005
Distance of town near village 0.003*** 0.001
Distance of market near village -0.0003 0.001
Average annual temperature -0.003 0.003
Average annual relative humidity -0.021*** 0.004
Months between harvest and next season 0.189*** 0.038
Months b/n harvest and next season squared -0.008*** 0.002
Zonal and crop dummies Yes
Chi2 1,884
Probability of Chi2 0.00
Number of observations 13,289 14
4.1 Crop losses during storage
• A large proportion (60%-84%) of HHs that stored grain cops did not
suffer losses
• Among HHs that actually suffered losses during storage
• Highest proportion reported losses of only 1-10% of quantity stored
• Next higher proportion reported highest losses (90%-100%) in most crops
10 50
Losses during storage (%)
6 30
4 20
2 10
0 0
Teff Barely Wheat Maize Sorghum Pulses Oil-seeds
kgs Percent
16
4.1 Crop losses…Contd.
• Proportions similar to other SSA countries (Tanzania and Uganda) and
slightly higher than Malawi (Kaminski & Christiaensen, 2014)
• Lower than FAO (2011) on-farm handling and storage PHL estimate (8%)
• Averaged across all HHs (including non-storing) losses are at least half
the proportion when only storing HHs are considered
kgs Percent
4 20
2 10
0 0
Teff Barely Wheat Maize Sorghum Pulses Oil-seeds
17
4.1 Crop losses…Contd.
• Losses mostly increase across months crops were stored
• HHs enquired to assign grades for intensity of damages sustained
• Grades assigned appear consistent with percent of crop lost during storage
18
4.2 Associates of losses during storage
• Given argument made earlier reduced form PHL equation estimated
• Percent of crop lost during storage negatively associated with:
• HH size (higher labor availability/a food need) but education has a wrong sign
• Seasonal price gap (opportunity cost of losses higher)
• Length of period between harvest and beginning of next season
• Temperature during post-harvest season, which reduces moisture in crops
• Percent of crop lost during storage positively associated with:
• Wealth (lower care) and with proximity to urban centers (less subsistence);
• Humidity during post-harvest period
• Most results consistent with what was expected
• Storage and crop loss equations estimated simultaneously using
Heckman’s sample selection method (given losses are unobservable
for those that did not store)
• Results similar (and qualitatively same) to those discussed so far.
19
4.2 Associates of losses…Contd.
Table 6. Tobit model estimates of crop damage equation
Variables (Dep. Var.: percent of crop damaged) Coeff. SE
HH head female (=1 if yes) -2.379 1.957
Age of HH head -0.041 0.060
Education (literate, primary incomplete) 5.833*** 2.174
Education (literate, primary complete) -2.532 2.668
HH size -1.523*** 0.440
HH wealth index 5.230** 2.043
Seasonal producer price gap -3.390** 1.627
Distance of town near village -0.202** 0.081
Distance of market near village -0.054 0.118
Average annual temperature during storage season -1.108** 0.563
Average relative humidity during storage season 1.344*** 0.398
Months between harvest and next season -26.30*** 6.744
Months b/n harvest and next season squared 1.293*** 0.327
Constant -175.7*** 42.05
Zonal and crop dummies Yes
Chi2 1,884
Number of observations 5,629 20
5. Summary
• Less than half HHs that grow grains store crops
• Storing less frequent in more commercialized oilseeds
• Most (76%) HHs store crop for 3 to 6 months
• Storing more frequent in Tigray followed by Amhara
• A large proportion of HHs (over 60%) did not suffer storage losses
• Out of that suffered losses higher proportion reported lower losses
• Losses during storage estimated between 5% & 7% of stored output
• Factors associated positively with storing grain (HH size, seasonal
price gap, and number of months b/n harvest and next season) are
negatively (consistently) associated with losses during storage and
the reverse is true for HH wealth and proximity to urban centers,
• Caveat: Further investigation/purposefully collected data needed to
understand divergence between the proportion of HHs that
consume a large proportion of the output but state not storing crop:
• Out of those that consumed 100%/>=75% only 31%/40% stored crops
21
Comments welcome
Thank you
22