James Stevenson and Paul Viek
Policy Seminar
Managing natural resources for sustainable production systems: A research agenda at the crossroads
Co-organized by CGIAR Independent Science and Partnership Council (ISPC); IFPRI; and CGIAR Research Program on Policies, Institutions and Markets
Feb 28, 2018 - 12:15 pm to 01:45 pm EST
The Economic and Organised Crime Office (EOCO) has been advised by the Office...
Adoption of CGIAR priority sustainable agricultural practices: Synthesis of nine new empirical studies
1. Adoption of CGIAR priority sustainable
agricultural practices:
Synthesis of nine new empirical studies
James Stevenson and Paul Vlek
Policy seminar, IFPIR,Washington DC, Weds 28 February 2018
2. Acknowledgments
Insights from a body of empirical work carried out by teams led by:
AslihanArslan,Andre Butler, Stein Holden, Robin Lovell, Kizito Mazvimavi,
Munyaradze Mutenje, EphraimNkonya, Kai Sonder,Tor-Gunnar Vagen
Research assistance from:
Nuri Niyazi, Melanie Levine
Comments from:
Benjamin Koetz, LakshmiKrishnan,Jeff Michler, Doug Pachico,Alex Pfaff,
Matin Qaim, David Spielman, and attendees at workshops in Rome (Dec
2015) and Nairobi (July2017)
Funding from:
DFID and Bill & Melinda Gates Foundation grant for “Strengthening Impact
Assessment in the CGIAR” program (2013-17)
3. Outline
• CGIARStanding Panel on ImpactAssessment
• Motivation
• Prioritization
• Summary of nine adoption studies
• Overall results
• Discussion
• Properties of the NRMpractices
• Understanding targetdomains
• Experimental approaches to addressing constraints to adoption
4. CGIAR Standing Panel on Impact Assessment
Doug Gollin,Member
Oxford University
Karen Macours, Chair
Paris School ofEconomics
JV Meenakshi, Member
Delhi School of Economics
SPIASecretariathosted by FAO in Rome:
Nancy Johnson, James Stevenson, LakshmiKrishnan
CGIAR center-hostedpost-docs:
Frederic Kosmowski (ILRI,Ethiopia), Haruna Sekabira (IITA,Nigeria),
Stella Wambugu (IITA,Tanzania),John Ilukor (World Bank, Uganda)
5. SPIA recent NRM-relatedresearch
Impact assessment studies
1. Agroforestry (ICRAF,Univ. of Illinois, ViAgroforestry)
2. AWD (North Carolina State Univ., IRRI)
3. Conservation agriculture (Univ.of Illinois, ICRISAT)
4. Drip irrigated horticulture (Johns Hopkins, GWU,MDG Center)
5. Forest co-management (Virginia Tech, CIFOR)
6. Intercropping (Yale,ICIMOD)
7. ISFM(ParisSchool of Economics,IITA)
8. Rainwater harvesting(Tufts Univ.)
9. Soil fertility analysis based extension (UCBerkeley, World Bank,
QFD)
Adoption studies on sustainable agricultural practices
Focus of thistalk
6. Motivation
• Natural resource management (NRM)
research represents a large share of the
CGIAR’s annualbudget
• Relative to crop germplasm improvement,
NRMis under-evaluated
• “Small N problem” for much of the NRM
research portfolio
• Serious measurementchallenges
associated with estimatingenvironmental
impacts from NRMresearch
• Sustainable agriculturalpractices:
adoption is a necessary condition for
impact, is tractable, and value of
information ishigh
All CGIARresearch
Priority
practices
Research on
sustainable ag
practices
NRMresearch
7. Prioritization
• Large universe of possible practice x country combinations
• We searched for claims made about adoption of sustainable
agricultural practices through all Annual Reports (Centres and CRPs,
2003 - 2012) and Performance Monitoring System (PMS2006 - 2010)
• Database of 124 adoption claims of specific innovative new practices
(and 46policy-related claims)
• Prioritized according to scores for:
- scale ofadoption/diffusion
- multi-year, multi-country outcome claims
- role of CGIARresearch / clarity of the impact pathway
• Shortlist of 6 sustainable agricultural practices with 38 practice x
country combinations
8. n: ObjectivesOur contributio
Blank
Priority practices Relevant countries
Agroforestry
particularly “fertilizer trees”,
leguminous fodder shrubs
Malawi, Kenya, Zambia, Zimbabwe,Rwanda
Alternate wetting and drying
(AWD)
in rice productionsystems
China, Vietnam, Philippines, Indonesia,Myanmar,
Bangladesh
Conservation agriculture
primarily in maize or wheat-
based systems
Malawi, Zambia, Zimbabwe, Mozambique, India, Pakistan,
Nepal, Bangladesh, Kyrgyzstan, Uzbekistan, Tajikistan,
Turkmenistan, Kazakhstan, Iraq,Mexico
Cocoa integrated crop and
pest management (ICPM)
Cameroon, Cote d’Ivoire, Ghana, Liberia, Nigeria
Micro-dosing offertilizer
in maize-based systems
Kenya, Zimbabwe, Mozambique
Integrated soilfertility
management
Kenya, Rwanda, Burundi,DRC
9. Prioritization
•Open call for EoIs (to over 600 researchers / institutions) to essentially
“pre-qualify” parties into a hybrid collaborative-competitive model
•64 EoIs submitted; reviewed by SPIA; people from 18 EoIs invited to
workshop in Rome, Dec 2015
•Invitation to teams formed to submit proposals for 9 work packages,
with both “core” and “upgraded” budget envelopes
•12 proposals externally reviewed by panel of four (economists and
biophysical scientists)
•Funding across 9 work packages at either “core” or “upgraded” request
10. n: ObjectivesOur contributio
Blank
Priority practices Relevant countries
Agroforestry
particularly “fertilizer trees”,
leguminous fodder shrubs
Malawi, Kenya, Zambia, Zimbabwe, Rwanda
Alternate wetting and drying
(AWD)
in rice productionsystems
China, Vietnam, Philippines, Indonesia, Myanmar,
Bangladesh
Conservation agriculture
primarily in maize or wheat-
based systems
Malawi, Zambia, Zimbabwe, Mozambique, India, Pakistan,
Nepal, Bangladesh, Kyrgyzstan, Uzbekistan, Tajikistan,
Turkmenistan, Kazakhstan, Iraq,Mexico
Cocoa integrated crop and
pest management (ICPM)
Cameroon, Cote d’Ivoire, Ghana, Liberia, Nigeria
Micro-dosing offertilizer
in maize-basedsystems
Kenya, Zimbabwe, Mozambique, Niger
Integrated soilfertility
management Kenya, Rwanda, Burundi, DRC,Zambia
11. Agroforestry (AF)
Trees or shrubs grown in or around crops / pastures
Faidherbia albida / White acacia / “Fertilizer trees”
Fixes N and sheds its leaves in the rainy season
Alternate Wetting and Drying (AWD)
Lowland irrigated rice with controlled and intermittent irrigation
Periodic drying and re-flooding
Conservation agriculture(CA)
Zero tillage, permanent soil cover, rotation with legume
Fertilizer micro-dosing(MD)
Small quantities of fertilizer (bottle cap) during planting or 3-4 weeks after
emergence
Integrated soil fertility management (ISFM)
Graduated process towards combined use of improved seeds, mineral
fertilizer and organic matter
12. Authors (all2017) NRM Countries Institutions
1 Bhargava, Boudot, Butler, Chomé, Gupta,
Singh andSchulthess
CA India IFMR, UMichigan,
U C Louvain,CIMMYT
2 Sonder, Schulthess and Chomé CA Mexico CIMMYT
3 Mutenje, Marenya, Fantayeand
Mazvimavi
CA Mozambique
Zambia
CIMMYT, ICRISAT
4 Holden, Katengetza, Fisher and
Thierfelder
CA Malawi NMBU, UIdaho,
CIMMYT
5 Arslan, Alfani, Scognamillo,Ignaciuk,
Asfaw, Conti, Grewer, Kokwe, Kozlowska,
McCarthy, Phiri andSpairani
CA
AF
Malawi
Zambia
FAO (EPIC),IFAD,
ICRAF
6 Michler, Mazvimavi, Kairezi,Liverpool-
Tasie andSanou
CA
MD
Zimbabwe
Niger
ICRISAT, CIMMYT,
U Illinois
7 Vågen, Masikati,Chiputwa,
Parmutai, Franzel, Hughes,Jacobson,
Kuntashula, Nhlane, Alfani and Arslan
AF Zambia ICRAF, PennState,
FAO (EPIC)
8 Lovell,Thuy and Phong AWD Vietnam U NongLam,
UC SantaCruz
9 Nkonya, Azzarri, Kato, Koo, Nziguheba
and Van Lauwe
ISFM Kenya
Rwanda
Zambia
IFPRI, IITA,GeoPoll
13. Country Full
adoption
(%)
Partial
adoption
(%)
Data source(s) and methods
1 India N/A <3 Remote sensing (Sentinel 1A radar) and new
HH survey(Punjab) N/A <16
2 Mexico N/A N/A Remote sensing (Sentinel 1A radar)
3 Mozambique <6 <8 Secondary data and new qualitative data
Zambia <4 <10 Secondary data and new qualitative data
4 Malawi <1 <6 Three surveys – a) lead farmers; b) followers;
c) random panel
5 Malawi <1 <5 IHPS / SAPP (2013/2014)
Zambia <0.5 <4 RALS (2012 and 2015)
6 Zimbabwe <2.5 <15 ICRISAT panel (2007/8 - 2010/11)
Overall results: Conservation agriculture
14. Overall resultscont.
Country Innovation Full
adoption
(%)
Partial
adoption(%)
Data source(s) and methods
6 Zimbabwe Micro-dosing N/A N/A Existing HH survey
Niger Micro-dosing <3 <18 2013-14 survey by ICRISAT
7 Zambia Agroforestry N/A <15 Remote sensing and new HH
survey, with links out RALS 2015
8 Vietnam Alternate
Wetting and
Drying
<10 N/A Remote sensing (SAR) with new
qualitative data
9 Rwanda ISFM N/A N/A New SMS survey (GeoPoll)
Zambia ISFM <6 N/A RALS (2012)
Kenya ISFM <29 N/A Ag Sector Baseline Survey (2013)
15. Discussion
1) Nature of the innovation itself
2) Representativeness ofdata
3) Experimental agenda on constraints
16. 1. Nature of the innovations
Rogers (1962) scheme for examining diffusion of innovations:
• The nature of the innovation itself
• Communication channels
• Time
• Social / economic system in which innovation is embedded
Nature of the innovation broken down into the following:
• Relative advantage (private benefits)
• Compatibility
• Complexity
• Trialability
• Observability
17. CA MD AF AWD ISFM
Basis for
relative
advantage
(private
benefit)
Cost-savings if
already
mechanized
Yield
increases
vs no
fertilizer
Depends on
type of tree
(e.g. fodder,
fertilizer, fruit)
Lower fuel
costs if
pumping in
Yield
increases vs
no fertilizer
Compatibility Major
changes
needed
Fits easily On margins:
Easy fit;
Within plot:
more difficult
Moderate
changes
needed
Fits easily
Complexity Complex
(3 component
practices)
Simple Simple (Deceptively)
complex
Complex
(graduated
process)
Trialability Single plot or
part can be
trialled
Single plot
or part can
be trialled
Not possible –
need to commit
Often
constrained
by collective
action
Single plot
or part can
be trialled
Observability Neighbor’s
adoption can
be observed
Hard to
observe
others
Visible once
established;
Planting out is a
rare event
Perforated
pipe not
necessary
condition
Hard to
distinguish
from good
ag practice
18. 2. Representativeness of data
Selected samples
Doss (2003, 2006) highlights problem of single-purpose samples in
adoption studies with no clear relationship to underlying population
Will bias adoption upwards as unable to unbundle effects of promotion effort
(often combined with incentives) from adoption
Representative samples
e.g. all maize farmers in country X
Stable (canbe made into a panel) and policy-relevant
“Recommendationdomains”
Are we able to reliably screen / filter the population of farmers to reliably
predict ex-ante which farmers could benefit from the practice?
Adoption shares would then be an expression of this domain
Challenging to operationalize: heterogeneity matters and comes from
multiple domains (current ag practice, soils, rainfall,social, economic)
19. 3. Research agenda on constraints
Reasonable hypothesis: these practices do not provide a private return
(in the absence of incentive mechanisms)
• Has land productivity been prioritized over labour productivity or
profitability in evaluating new technologies?
• Have the implications of deeply heterogeneous agricultural conditions
been under-appreciated?
In some cases, market failures may be limiting farmers’ ability to adopt
potentially profitable practices (and thereby limiting our ability to enjoy the
public benefits)
Growing experimental literature on whether supplementary interventions
can help farmers overcome constraints regarding credit, information etc
(AgriculturalTechnologyAdoption Initiative: atai-research.org)
21. Calculated from RALS survey 2012
Is 3 components in a technology too complex?
ISFM for small farmers in SSA
22. Conservation agriculture
Mexico
• Approach developed for detecting tillage in Belgium based on surface
roughness
• Calibrate it for Northern Mexico through ground-truthing
• Requested ESAshift mode of data collection (VV to VH) in Sentinel mission
• Remote sensing predicts correctly 94% of the time
India
New, regionally representative HH survey across Indo-Gangetic Plains:
• 4 states, 240 villages, and 3600 households
• Remote sensing using radar images from Sentinel-1&2 to identify ZT
(approach asabove)
• Good correspondence between RSand survey for Punjab and Haryana
• Need to re-train for each state: numbers mismatched for UPand Bihar
Malawi, Mozambique, Zambia, Zimbabwe
• Malawi (NMBU): Survey of lead farmers and their followers
• Malawi and Zambia (FAO EPIC/ IFAD): Two rounds each of spatially-explicit
nationally representative panel (IHPS, RALS)+climate data to examine
correlates ofadoption
23. Fertilizer micro-dosing
Zimbabwe
• Analysis of ICRISATdata in two-wave panel structure 2013 and 2016 from
458 HHs
• Links to input-subsidy program promoting CA
• Widespread adoption of the spatial pattern of MD (spot application)
• Quantities applied much greater than intended in concept of micro-dosing
Niger
• ICRISATsurvey of 800 HHs in 40 villages in 4 regions
• Adoption of any fertilizer (43%) much higher than 2011/12 nationally
representative LSMSdata(13%)
• Most common approach (29% farmers) for fertilizer application is to mix
with seeds at low rates (equivalent to 2 – 8 kg / ha)
24. Alternate Wetting and Drying
Vietnam
• Government policy to promote the practice in Mekong Delta for rice
production “1 Must-Do, 5 Gains”
• Provinces in which a Farmer Organization (FO) controls water use and
distribution, AWD adoption is tied to the FO fee structure
• Each farm that falls within an FO is systematically linked to neighbors’
watering regimes: farmer must water if the FO decides it is watering time
• Synthetic Aperture Radar (SAR) can detect shifts in soil moisture over large
areas
• Calculated wetness index data to understand change over time for specific
cells, assuming that areas that show consistent change (i.e. flooding then
drydown) in the growing season areAWD adopters
• Use of a perforated pipe initially promoted to help farmers monitor water
table, but farmers don’t need it – not a necessary condition for adoption
• Note – 100-year drought in 2015/16
25. Agroforestry
Zambia
Geospatial model +new 400 HH survey to validate
• Geospatial methods using Landsat 8 work pretty well
– can predict Faidherbia albida in Zambia to 85%
accuracy
• Maps of Faidherbia albida across study area can be
used to assess adoption of species
• 15% of cultivated plots in study area had F.albida but
may be as little as a single tree in a plot
• Typically way below density required for measurable
impact on cropproductivity
• ICRAF carried out research and promotion of
improved fallows for almost 20 years in the Eastern
Province
• ICRAF leaving in 2006 created a void of agroforestry
activities across Zambia
26. Integrated Soil Fertility Management
Rwanda
• Team experimented with using an SMS-based survey via GeoPoll service
provider
• Two waves of data collection from 1000 HHs using SMS
• 93% of the respondents were under 35 years (instead of approx. 40%)
• Over 54 year-olds virtually absent from sample
• Measurement error for those that did respond appear to be very high –
reported yields are tiny fraction of nationally representative, driven by
weird plotarea numbers
Kenya
• Ag Sector Household Baseline in Kenya from 2013
Zambia
• RuralAgricultural Livelihoods Survey in Zambia from 2012