AGEC 640 – Ag Development & Policy
Measuring Impacts
October 30th, 2014
Today:
Focus on Malawi’s subsidy program
Readings:
Chibwana, C. et al. (2014) “Measuring the Impacts of
Malawi’s Farm Input Subsidy Program.” Forthcoming in
African Journal of Agricultural and Resource Economics.
Fisher, M. and G. Shively (2005) “Can Income Programs Reduce
Tropical Forest Pressure? Income Shocks and Forest Use in
Malawi.” World Development 33(7): 1115–1128.
labor
Price
S=MC
S=MC - σ
fert
maize
fert
other
D
Quantity
maize
What do we know about input subsidies?
• Great for farmers
(therefore popular with politicians)
• Expensive
(therefore unpopular with donors)
• Problematic
(therefore popular with researchers)
Issues surrounding fertilizer subsidies
• Is fertilizer a private good or a public good?
• Does public provision undermine
or “crowd-out” the private sector?
• Do subsidies reach the intended beneficiaries?
• What are their short-run and long-run impacts?
– Fertilizer use
– Crop choice/land allocation
– Unintended impacts (e.g. forests)
A dynamic perspective on what might be
going on and why we should care
At+1 = At + f(At, Xt) – ct
Malawi’s 2009 FISP
• Program goal: maize self-sufficiency
• Targeted 1.7 million farm households
• Inputs:
– Maize fertilizer (NPK): 150,000 mt
– Tobacco fertilizer: 20,000 mt
– Maize seed (OPV + Hybrid): 4,750 mt
– Cotton seed, legume seed, cotton chemicals
and grain storage pesticides
• Program cost: US$221million
– 13.5% of national budget
– 5.5% of GDP
FISP details
• Program delivered via voucher/coupon system
“Typical” voucher
– 100 kg fertilizer for maize (50 + 50)
– 2 kg of improved seed (Hybrid or OPV)
– 5.5% of GDP
• Distribution:
 Ministry of Agriculture
 District Officials
 Local chiefs
 Village heads and village development committees
Stated Targets: FHOH, residents, vulnerable
Research questions:
1. Who benefited from the subsidy program?
2. Did the program boost smallholder’s use of
fertilizer and maize output?
3. Did the participation influence crop choice?
4. [ additionally:
what effect (if any) did the subsidy
have on area expansion and forests? ]
Data
• Household panel covering 2002, 2006, 2009
• Kasungu and Machinga Districts
• Approx. 400 respondents
Results presented here are primarily based on our
2009 survey, with some additional insights drawn
from the 2002 and 2006 surveys.
IV estimation strategy
Sample statistics
Variable
Age
Household size
Land owned per HH (ha)
Female-headed (1=yes)
Net buyer of maize (1=yes)
Education
None
Some primary
Some secondary
Mean
47
6.3
1.6
0.15
0.73
15%
72%
13%
Land shares
Traditional maize
Improved maize
Tobacco
Other crops
Mean
0.38
0.25
0.06
0.31
Yields (kg/acre)
Mean
Traditional maize
1196
Improved maize
1391
(National smallholder average: 1483)
21% of households
in rural Malawi are
female-headed.
13% of all coupon
recipients in sample
were female-headed
Program participation
Category
Male
Female
Poor
Non-poor
Male, non-poor
Male, poor
Female, non-poor
Female, poor
None
Seed
only
Fertilizer
only
Seed &
fertilizer
0.12
0.21
0.18
0.09
0.09
0.16
0.13
0.28
0.06
0.04
0.05
0.06
0.07
0.06
0.04
0.03
0.18
0.25
0.20
0.18
0.17
0.18
0.21
0.28
0.64
0.50
0.58
0.67
0.67
0.60
0.63
0.41
Participation cannot be considered exogenous…
Proportion of subsidy in total fertilizer
For women,
the subsidy
represented about
3/4 of total fertilizer
used.
Male
Non-poor
Female
For the poor,
the subsidy
represented more
than two-thirds of
total fertilizer used.
Male
Poor
Female
0
50
100
150
Quantity of fertilizer used (kg)
Subsidized
Unsubsidized
200
Question 1:
Who received coupons?
• Estimated using probit, multinomial logit and Tobit
• Dependent variables:
– Probit: receipt of any coupon (0/1)
– MNL: type of coupon received – 4 or 7 categories
e.g. no coupon, seed only, fertilizer only, both seed and
fertilizer, etc.
– Tobit: monetary (market) value of subsidized inputs
• Identification variables for 1st stage:
– FHOH, years of residency, village size, asset poverty
– Relevant and satisfy standard overidentification tests
Findings for question 1
•
Female-heads less likely to receive coupons
•
Asset-poor HHs less likely to receive coupons
•
Residency matters
•
Village size does not matter
(cf. Jayne – member of parliament matters!)
Question 2:
Did coupons increase fertilizer use?
• Estimation approach: IV /Tobit
• Dependent variable: fertilizer/ha
• Explanatory variables:
–
–
–
–
demographics, farm size, location
fertilizer-maize price ratio
idiosyncratic shocks
“instrumented” variable for receipt of coupon
• (either from MNL or direct 2SLS in the case of value)
Question 2:
Did coupons increase fertilizer use?
Model 1
Observed
coupon
Model 2
Instrumented
coupon
Model 3
Instrumented w/
lagged fertilizer
Seed coupon only
-4.214
-60.81
-50.54
100kg fertilizer
135.5
189.6*
161.82
Question 2:
Did coupons increase fertilizer use?
Coupon value
(100 Mk)
Model 1
Observed
coupon
Model 2
Instrumented
coupon
Model 3
Instrumented w/
lagged fertilizer
1.21*
0.97*
0.50*
Findings for question 2:
•
positively correlated with coupon receipt
•
Intensity falls with farm size
•
positive correlation with use of improved maize
•
net buyers of maize used less fertilizer
•
adding 2002 and 2006 fertilizer intensity reduces
point estimate for coupon by approximately 50%
1500 1750
Maize yield response to fertilizer
D
C
500
1000
B
A
0
100
Marked points on the graph
correspond to the following
fertilizer-yield combinations:
A [114 kg/ha, 1302 kg/ha]
B [165 kg/ha, 1245 kg/ha]
C [136 kg/ha, 1373 kg/ha]
D [175 kg/ha, 1477 kg/ha]
200
Fertilization rate (kg/ha)
Traditional maize
300
Improved maize
Question 3:
Did the FISP influence land allocation?
• Estimation approach: 2SLS, SUR
• Dependent variable: land shares
–
–
–
–
traditional maize
hybrid maize
tobacco
other crops
 pik 
   j1H i   j 2i   j 3Wi   j 4 Si   j 5 Fi   j 6Ci   ij
Lij   j    jk 
k 2
 p1k 
3
Findings for question 3
• Farmer response to price signals is weak
• Maize and “other crops” act as substitutes
• Maize and tobacco are complements
• Results robust to inclusion of 2006 fertilizer use
• Coupon for maize seed and fertilizer led to:
– 16-22% more land to maize
– 3-8% more land to tobacco
– 20-26% less land to “other crops
Overall impact on maize, revisited
What is the total maize output gain associated with
coupon receipt?
change in yield
x ( 1 + change in maize area )
≈ 456 kg on average
 about half of the gain is from seed, half from fertilizer 
But... maize area comes at the expense of other crops
displaced, and the value of the output of these other crops
constitute about 50% of the average gain.
Overall impact on maize, revisited
Question 4:
Any unintended consequences?
• Approach: IV/Tobit
• Dependent variable: area of forest cleared
– mean: 0.16 acres/household (including 0s)
– 14% of sample reported forest clearing activity
• Explanatory variables:
farm size, agricultural prices,
forest access and control,
shocks, coupon receipt variables
Findings for Question 4
• Farm size has a negative correlation with forest clearing
• Rates of clearing higher on private and communal forests
• Rates lower in presence of a Forest User Group
(more likely to protect what is increasingly scarce?)
• Forest clearing not driven by higher agricultural output prices,
at least in the short run
• Less forest clearing among recipients of maize coupons
(no forest left to clear?)
• Derived demand for forest resources for tobacco
(land less directly affected than timber for drying sheds)
Findings from Fisher and Shively
Findings from Fisher and Shively
Average value of SPS ≈ 450 Mk,
so ≈ 1062 fewer hours, or drop of more than 50%.
Conclusions
• The FISP appears to have had a positive impact on
fertilizer use and maize output in the survey year
(450 kg gross (≈ half this net) = +25% boost)
• Accessing seed and fertilizer is much better than
accessing fertilizer only
• Program seems to have increased land allocation to
maize & tobacco at expense of other crops
• Did program help reduce forest pressure?
– Probably (as with the SPS in 1999-2000)
but…there were some negative effects through tobacco
Implications
• Who should be subsidized?
• Targeting at the poor could be improved
• Subsidies should be part of a comprehensive strategy
to improve agricultural productivity
• Too much emphasis on fertilizer?
Package = 2kg maize seed and 100kg fertilizer
• Half of observed gains came from improved seed
• Improved seed delivery systems needed
• Reinforce research and extension