PIM Webinar held on 19 September 2018.
Presenters: Berber Kramer (IFPRI), Patrick Ward (Duke Kunshan University). More information at http://bit.ly/AgInsuranceWebinar
Artificial Intelligence In Microbiology by Dr. Prince C P
Helping smallholder farmers manage risks: Innovations to improve agricultural insurance
1.
2. Helping smallholder farmers
manage risks: Innovations to
improve agricultural insurance
Berber Kramer1 and Patrick S. Ward2
PIM Webinar
September 19, 2018
1 Research Fellow in the Markets, Trade and Institutions
Division of the International Food Policy Research Institute
2 Assistant Professor of Environmental Economics and Policy
in the iMEP program at Duke Kunshan University
6. Global warming: Worsening vulnerability to food insecurity
Change in Hunger and Climate Vulnerability Index relative to baseline calculated for simulated
climate states at 2°C global warming
Source: Betts et al. (Phil. Trans. R. Soc. A, 2018).
7. Challenges in providing agricultural insurance
Traditional indemnity-based insurance:
▪ High administrative and transaction costs.
▪ Asymmetric information between insured and insurer, which in turn
gives rise to adverse selection and moral hazard.
Index-based insurance was designed to overcome these challenges.
Problem:
▪ Basis risk: The index used to determine payouts does not capture
farmers’ conditions on the ground
▪ Lack of trust and understanding
8. This webinar: Outline
1. Using a stream of smartphone camera data over time to monitor crop
conditions, management and claims of crop losses and damage
▪ Formative evaluations on the feasibility and economic viability of
picture-based insurance and advisory services in India
2. Bundling insurance with climate-smart, risk-reducing, agricultural
practices and technologies
▪ Studies on insurance design and evaluating impacts in Bangladesh
and India
12. Features to ensure standardization
Follow-up pictures: Taken of same site as initial picture
Goal: More accurate loss assessment and limit moral hazard
13. Punjab
Haryana
Formative evaluation of PBI
▪ Study in 6 districts, 50 villages, 750 wheat producers
▪ In return for sending in pictures of their crops, they received insurance:
▪ Weather index-based for extreme heat, excess rainfall
▪ Wth picture-based loss monitoring, covering any visible damage from
natural disasters e.g. lodging and pest/disease
14. Four questions
1. Does picture-based loss monitoring help reduce basis risk?
2. Can pictures be used to monitor crop growth stages?
3. Does this approach improve demand for insurance?
4. Should we worry about tampering and moral hazard?
15. Lower yields among farmers with insurance payouts?
19.84 19.82
No payout Payout
CCEsYield
(Quintalsperacre)
Weather index-based insurance
17. Normalized greenness is predictive of growth stage
Application for weather index-based insurance: cover a specific
growth stage instead of a fixed calendar period, which can help
shorten the coverage period and reduce costs
Tillering Stem extension Heading + ripening
Source: Hufkens
et al., 2018
18. Four questions
1. Does picture-based loss monitoring help reduce basis risk? √
2. Can pictures be used to monitor crop growth stages? √
3. Does this approach improve demand for insurance?
4. Should we worry about tampering and moral hazard?
21. Results
1. Does picture-based loss monitoring help reduce basis risk? √
2. Can pictures be used to monitor crop growth stages? √
3. Does this approach improve demand for insurance? √
4. Without inducing tampering and moral hazard? √
22. ▪ Working with HDFC ERGO General Insurance to use picture-based loss
assessment in existing insurance products
o Insurance actors in Kenya have also expressed interest
▪ Automate image processing and damage detection to the extent possible
o Large training sets required
▪ Impact evaluation, with focus on higher value/risk crops (e.g. tomatoes)
o Emphasis on technology acceptance and inclusiveness
▪ Enabling environment: Complementarities with agro-advisory services
and other financial services (credit, price insurance)
▪ Vision: Existing index-based products (covariate risk) with picture-based
loss monitoring (idiosyncratic risk / second fail-safe trigger).
Picture-Based Insurance: What’s Next?
24. Payoff/Profits
Intensity of disaster
Conventional varieties
Drought-tolerant variety
Traditional index insurance
Optimized index insurance
None Moderate ExtremeSevere
Why bundle climate-smart agriculture with insurance?
25. Drought-tolerant rice
▪Several varieties developed under the Stress-Tolerant
Rice for Africa and South Asia (STRASA) program
▪Sahbhagi dhan released in 2010 and approved for
cultivation in upland and rainfed lowland rice-growing
regions
oShort duration: allows the crop to escape early/late season
drought (late monsoon onset, early monsoon cessation)
oCan withstand longer dry spells than most widely-used local
varieties
26. Optimized weather index insurance
▪Structured to begin paying out to farmers during
“severe” droughts
▪Payouts structured to be roughly equivalent to the value
of lost production during “severe” droughts
oCoverage area per unit of insurance 0.1 ac
27. Product marketing
▪Product was marketed through information sessions at
the village level
▪Year 1: Randomized subsidies (up to 30%)
▪Year 2: Flat effective price, but some villages were
offered a subsidy (on top of an inflated price)
28. Patterns of uptake across years
Purchased in
Year 1
Did not
purchase in
Year 1
Purchased in
Year 2
25.5% 11.8%
Did not
purchase in
Year 2
33.1% 29.6%
29. Determinants of DT-WII demand
▪Sequential decisionmaking process
1. Should I purchase? (participation)
2. How much should I purchase? (intensity)
30. What might we expect to influence uptake?
▪ Price (subsidy level)
▪ Drought expectations
▪ Behavioral characteristics:
o Time preferences
o Risk preferences
o Ambiguity preferences
▪ Trust in the insurance provider
▪ Neighbors’ purchases (peer effects)
▪ Year 1 ▪ Year 2
▪ Receipt of a “subsidy”
▪ Drought expectations
▪ Behavioral characteristics
▪ Trust in the insurance provider
▪ Neighbors’ purchases
▪ Experiences from Y1
o Did they purchase in Y1?
o Did they experience a drought in Y1?
▪ Neighbors’ experiences in Y1
o Did neighbors purchase in Y1?
o Did neighbors experience a drought in
Y1?
31. What actually influences uptake in Year 1
▪Price affects participation in the market (-)
▪Peer effects crowd in participation (+)
▪Risk aversion does not affect participation, but limits the
intensity (-)
▪Trust in the insurer increases intensity of participation (+)
32. What actually influences uptake in Year 2
▪ Own experience in Y1 matters a lot
oIf farmer purchased DT-WII in Y1 and experienced a drought (but
also rec’d insurance payout), more likely to participate (+) and
purchase more units (+) in Y2
▪ Experience of neighbors not important
oHard to observe?
▪ Peer effects crowd in participation and intensity in Y2 (+)
33. No effect of price reduction framed as subsidy
▪Year 1 results indicate that farmers are sensitive to price
▪Year 2 results indicate that this price-sensitivity is not
sensitive to how it is framed
▪Policy implications: More sustainable/less distortionary
investments to reduce the cost of the product could be
successful at stimulating demand
34. Why does this matter?
▪ Can DT-WII offset the ex ante and ex post impacts of droughts?
▪ Evidence from a similar WII product in Bogra district, Bangladesh suggests it
can:
o Ex ante
o Increased investments in agricultural production during the monsoon
season (coverage period) as a result of having risk management (risk
management effect)
o Ex post
o Increased investments in agricultural production during the dry season
(not covered by insurance) as a result of increased income from the
insurance payout (income effect)
35. Concluding remarks
▪ Exciting time to work on agricultural insurance
▪ There have been many exciting innovations in recent years that can
contribute to the development of improved agricultural insurance products
that can meet the needs of farmers
▪ But we must not rest on our laurels – there is more work to be done:
o Lots of excitement around the use of remote sensing technologies, but
current generation of open-access or affordable remote sensing too
coarse to detect plot-level losses
o Picture-based insurance project demonstrates possibility of engaging
with farmers to document crop losses
o Potential for positioning insurance as part of a portfolio of risk
management products, including CSA
36. Thank you!
For more information:
(Project notes available at https://www.ifpri.org/project/PBInsurance)
Contact Berber Kramer (B.Kramer[at]cgiar.org)