Recent field, household, and community-level

advertisement
Understanding the existing
agricultural input landscape in
sub-Saharan Africa:
Recent field, household, and community-level
evidence
Megan Sheahan and Christopher B. Barrett
Cornell University
Presented at the International Livestock Research
Institute, Nairobi, Kenya, 10 April 2014
AG RICULTURE
IN AFRICA
T E L L I N G FA C T S
FROM MYTHS
Broader Motivation: LSMS-ISA data
Current challenges
The world in which African farmers operate
has changed:
• High and more volatile food price environment
• Africa is growing and urbanizing quickly
• Production environment changes due to climate
change and soil erosion
Concurrent renewed investment in
agricultural sector
However, knowledge base is grounded in “old
ideas” about African agriculture or
inappropriate data  Salient case studies,
purposively selected samples, agricultural statistics of
unknown quality
Page 2
Broader Motivation: LSMS-ISA data
An opportunity!
• Collecting household survey data with
focus on agriculture in 8 SSA countries
 Burkina Faso, Ethiopia, Malawi, Mali, Niger,
Nigeria, Tanzania, Uganda
• Improving methodologies in data
collection, producing best practice
guidelines and research
• Documenting and disseminating micro
data for policy research
• Building capacity in national
institutions
Page 3
Broader Motivation: LSMS-ISA data
Survey features
• Nationally representative (rural and
urban, and various administrative
levels)
• 4 I’s
Integrated: multi-topic and georeferenced (link with eco-systems)
Individual: gender/plot
Inter-temporal: panels with tracking
Information technology:
concurrent data entry (CAPI, GPS)
• Open data access policy
http://www.worldbank.org/lsms-isa
Page 4
Broader Motivation: LSMS-ISA data
Survey instruments
Household
• Individual-level data on
demographics, education,
health, labor & anthro
• Housing, durable assets
• Food & non-food
consumption
• Income
• Food security
• Non-Farm enterprises
• Subjective welfare
Agriculture
• Plot-level data on (i) Land
Areas, (ii) Labor & non-labor
inputs, (iii) Crop cultivation
& production
• Crop sales & utilization
• Farm implements
• Extension services
• Livestock
• Fisheries
Community
• Demographics
• Services
• Facilities
• Infrastructure
• Governance
• Organizations & groups
• Market prices
Page 5
Broader Motivation: LSMS-ISA data
Data collection schedule for panel rounds
Burkina
2014/15
Ethiopia
2011/12
Malawi
2013/14
2010/11
2013/14
Mali
2014/15
Niger
2011/12
Nigeria
Tanzania
Uganda
2008/09
2009/10
2014/15
2010/11
2012/13
2010/11
2012/13
2010/11
2011/12
2013/14
Page 6
Broader Motivation: LSMS-ISA data
Data collection schedule for panel rounds
Burkina
2014/15
Ethiopia
2011/12
Malawi
2013/14
2010/11
2013/14
Mali
2014/15
Niger
2011/12
Nigeria
Tanzania
Uganda
2008/09
2009/10
2014/15
2010/11
2012/13
2010/11
2012/13
2010/11
2011/12
2013/14
Page 7
Broader Motivation: “Myths and Facts” Project
Project objectives
• Provide a solid, updated, and bottom-up picture of Africa’s agriculture
and farmers’ livelihoods
• Create a harmonized and easy-to-use database of core agricultural
variables for tabulation and regional cross-country benchmarking
• Build a community of practice
– Partnering institutions: World Bank, African Development Bank, Cornell University,
Food and Agriculture Organization, Maastricht School of Management, Trento
University, University of Pretoria, Yale University
– Mentorship program for young African scholars
AG RICULTURE
from US and African institutions
IN AFRICA
Project led by Luc Christiaensen
T E L L I N G FA C T S
FROM MYTHS
lchristiaensen@worldbank.org
Page 8
Broader Motivation: “Myths and Facts” Project
Common wisdoms revisited
1) Use of modern inputs remains
dismally low
2) Land, labor and capital markets
remain largely incomplete
3) Agricultural labor productivity is
low
4) Land is abundant and land markets
are poorly developed
5) Rural entrepreneurs largely operate
in survival mode.
6) Extension services are poor
7) Agroforestry is gaining traction
8) African agriculture is intensifying
9)
10)
11)
12)
13)
14)
15)
Women perform the bulk of
Africa’s agricultural tasks
Seasonality continues to
permeate rural livelihoods
Smallholder market
participation remains limited
Post harvest losses are large
Droughts dominate Africa’s risk
environment
African farmers are increasingly
diversifying their incomes
Agricultural commercialization
and diversification improves
nutritional outcomes
Page 9
Broader Motivation: “Myths and Facts” Project
Common wisdoms revisited
1) Use of modern inputs remains
dismally low
Page 10
Motivation
Why is it important to explore input use?
• Increase in agricultural productivity necessary for agricultural transformation
and poverty reduction
• Expanded use of modern inputs, embodying improved technologies, is often
seen as a prerequisite to increasing agricultural productivity
• Common wisdoms (“stylized facts”):
– African farmers use few modern inputs
– Input provision systems remain poor
• Those stylized facts have helped spur the new government input subsidy
paradigm in SSA, although little cross-country, nationally representative, and
recent evidence exists to support those stylized facts
• Use LSMS-ISA data to describe “input landscape” related to fertilizer, modern
seed varieties, agro-chemicals (pesticides, herbicides), irrigation, mechanized
inputs (animal traction, farm machinery)
Page 11
Structure of Paper
Meticulously assembled data set but simple descriptive methodology
• From where do common conceptions on input use currently come?
– Macro-statistics: FAOStat, World Bank’s World Development Indicators,
CGIAR’s Diffusion and Impact of Improved Varieties in Africa project
– Micro-statistics: Literature review of studies on input use from
household level data with large samples by country and input
• With the newest available round of LSMS-ISA data in each country:
1. Who uses modern inputs and in what amounts?
2. What is the input provisioning situation?
3. What is the main source of variation in binary input use decision?
• 10 most striking and important findings presented here
Page 12
Sample and data considerations
Households that cultivate at least one field in main ag season
Country
Year
Season
# hh
# plots
Ethiopia
2011/12
-
2,852
23,051
Malawi
2010/11
Rainy
10,086
18,598
Niger
2011/12
Rainy
2,208
6,109
Nigeria
2010/11
-
2,939
5,546
Tanzania
2010/11
Long rainy
2,372
4,794
Uganda
2010/11
First
1,934
3,349
Sample includes over 22,000 households and 62,000 plots
across 6 countries
Page 13
(1) Input use is not uniformly low, especially with respect to percentage of
cultivating households using inputs but also rates of use.
Most true of inorganic fertilizer and agro-chemical use
• Near-perfect match with macro-stats on
inorganic fertilizer use rates across four
countries  Ethiopia, Niger, Tanzania, Uganda
• Largest discrepancies in 2 of 3 countries with
fertilizer subsidy programs  Malawi and
Share of cultivating households (%) using
input on fields
100
77
80
56
60
40
33
31
20
3
8
Nigeria
41
17
13 17
11
3
0
Ethiopia
Malawi
Niger
any agro-chemical
Nigeria
Tanzania
Uganda
inorganic fertilizer
• Relatively high shares of households use
inorganic fertilizer, with 3 of 6 countries
> 40 percent
• Where > 30 percent of households use
agro-chemicals, any implications for
human health?
• Uganda has lowest input use prevalence
of 6 included countries
121
Inorganic fertilizer application (nutrients)
120
100
80
64
kg/ha
60
40
56
33
26
2523
20
2 1
6
8 7
12
13
1 2
0
Micro data (LSMS-ISA, 2009-2011)
Macro data (World Bank, 2010)
Page 14
(2) The incidence of irrigation and mechanization are really quite small.
Micro-statistics similar to macro-statistics
10
9
9
Mechanization is proceeding slowly
Water control is limited
•
8
7
7
•
•
6
5
5
4
4
4
1
0
4
3
3
2
4
Traction animal ownership above 20 percent
in all countries except Malawi
1-2 percent of households own a tractor
1/4 of households in Nigeria used a
mechanized input or animal power on their
plots during main ag season
1
1
2
2
0.4
0.2
% of all cultivated land under irrigation by smallholders
% of households with at least some irrigation on farm
Page 15
(3) Huge amount of variation within countries in the prevalence of input
use and intensity.
Example from Ethiopia
• Most input use appears to be driven by certain regions and zones within countries
Page 16
(4) Input use is as high on maize dominated plots as it is on average at the
household level.
Cash crops not driving input use?
Inorganic fertilizer use (kg/ha)
0
50
Ethiopia
45
100
88
135
Malawi
Niger
1
146
5
123
128
Nigeria
15
16
Tanzania
Uganda
150
3
1
maize plots*
Households
*Niger: millet/sorghum/millet/cowpea instead of maize (too few)
• Commercially purchased maize seeds are used by
25-40 percent of maize cultivating households
Page 17
(5) Consistent negative relationships between farm and plot sizes and input
use intensity.
Example from Nigeria
• Local linear non-parametric regressions of unconditional inorganic fertilizer use rates
• Shape differs by country, especially where ranges in size vary substantially  relatively
flat for Malawi and different pattern for Ethiopia and Uganda
• Negative relationship is even more pronounced at the plot level in all cases except
Ethiopia  important policy implications!
Nigeria – household level
Nigeria – plot level
Local polynomial smooth
kg/ha of inorganic fertilizer applied to field
Local polynomial smooth
200
150
100
50
0
-50
0
200
150
100
50
0
1
2
3
Total hectares of land under cultivation
95% CI
lpoly smooth
kernel = epanechnikov, degree = 1, bandwidth = .34, pwidth = .51
4
0
.5
1
1.5
plot size in hectares
95% CI
lpoly smooth
kernel = epanechnikov, degree = 1, bandwidth = .2, pwidth = .3
Page 18
(6) Little variation in input use when households and plots are split by soil
quality and erosion status, both farmer-perceived and geo-referenced.
Moreover, few farmers consider their plots of ‘poor’ quality
• Regression analysis reveals that ‘average’ and ‘poor’ plots significantly are more likely to
receive inorganic fertilizer treatments than those categorized as ‘good’
• Knowledge gap among farmers? Weak evidence against ‘poor but efficient’ claim?
• Implications for extension programs and the need to invest in simple soil quality tests
Page 19
(7) Surprisingly low correlation between the joint use of commonly ‘paired’
inputs.
Especially apparent when moving from household to field level
• Show correlation between two-way input use in paper
• Can investigate three-way input use (fertilizer, seed, irrigation) for Ethiopia and Niger
• Farmers may use >1 modern input on farm, but appear to be diversifying within farm
rather than reaping output gains by pairing inputs together
• Synergies from pairing inputs still yet to be exploited
Ethiopia – household level
Ethiopia – field level
Page 20
(8) Fertilizer subsidies are not as universal as often believed or reported in
government statistics.
3 of 6 current LSMS-ISA countries have government fertilizer subsidy programs
• Mixed reviews by households on
input market accessibility
changes over time in Malawi,
where fertilizer subsidies are
most pervasive
• > 50 percent of households receive a
government fertilizer subsidy only in
Malawi
• Relatively low fertilizer subsidy
occurrence in Nigeria alongside high
rates of fertilizer use
• Estimates of fertilizer subsidy coverage
in LSMS-ISA data fall short of other
estimates using government data
– Could be issue of LSMS-ISA data collection
timing or sluggish/failed distribution of
planned vouchers by subsidy program
implementers
Percent of households in Malawi
by perception of current input market accessibility
relative to five years ago
45
39
40
35
30
29
30
25
26
22
19
20
21
14
15
10
5
0
More
Less
Fertilizer
About the same
Not applicable
Improved maize seed
Page 21
(9) There is very low incidence of credit use for purchasing modern inputs.
Statistics should capture both informal and formal credit types
• < 1 percent of cultivating
households used credit to
purchase improved seed
varieties, inorganic fertilizer,
and agro-chemicals
– True of all countries except
Ethiopia, where the government
issues input credit
• (1) Country-averaged inputoutput price ratios imply that
fertilizer is a good
investment at aggregate
levels + (2) Wealthier
households more likely to
use fertilizer
– No or under-use may signal cash
flow constraints, which could be
aided by expanding credit
options
• Policy implication  addressing rural financial
market failures may be key for expanding input use
Page 22
(10) Over half of the variation in inorganic fertilizer and agro-chemical use
comes from the country level.
Suggests that policy and institutional environment are very important
• Ultimately interested to learn where most
of the variation in input use comes from
– Biophysical, infrastructure, market, socioeconomic, or policy-specific variables?
• Binary use at household level (avoids bias
from survey design)
– Also reported at field level under a number
of different specifications (fewer field level
characteristics match across surveys)
• R2 decomposition using Shapley-Owen
values
• > 50 percent of variation in fertilizer use can
be explained by country level!
• Suggests that geography, policy, and
institutional environment are important for
ushering a Green Revolution in Africa
Page 23
Conclusions
Main take-away messages
• Input use is not always low  Much more heterogeneity in input use
between and within countries than commonly assumed
- Varies by country, input, crop, and a large number of important covariates
- Micro-level statistics allow us to investigate this variation more fully
• Scope for improvement remains
- Synergies in effectively combining inputs appropriately yet to be exploited
- Policy and institutions seem important for encouraging yield-enhancing input use
• 10 findings presented here are only a small subset of what can be
gleaned about input use from the LSMS-ISA surveys
• Exploiting newly emerging panel data will allow us to provide greater
nuance for guiding more intelligent policy design
Page 24
Thank you!
Contact: mbs282@cornell.edu
Page 25
Download