index insurance for small-holder agriculture: what we have learned about impacts and contract design
TRANSCRIPT
Index Insurance for Small-holder Agriculture:What We Have Learned about Impacts & Contract Design
Michael R Carter
Department of Agricultural & Resource EconomicsBASIS Assets & Market Access Research Program &
I4 Index Insurance Innovation InitiativeUniversity of California, Davis
http://basis.ucdavis.edu.
AFD-FERDI Workshop, Paris
June 24, 2014
M.R. Carter Index Insurance Impacts & Design
Promise of Index Insurance
Know that risk is directly costly
Makes Households Poor when it leads them to adopt less riskyactivities, but lower returning activitiesKeeps Households Poor when it leads them accumulateunproductive ’buffer’ assets
Deepens capital constraints also as
Cuts off self-finance by holding of ’unproductive’ savingsDiscourages external finance (credit), especially when risk iscorrelated
Reducing risk via insurance should address all these problemsNote also that when twined with technology adoption,insurance becomes relatively inexpensiveBut can insurance work for smallholder agriculture?
M.R. Carter Index Insurance Impacts & Design
Logic of Index Insurance
Conventional insurance (based on individual loss adjustment)has a dismal record
Costly to verify losses for smallholdersMoral hazard if do not/cannot reliably verify lossesAdverse selection
Index insurance does not pay based on (verified) individuallosses, but instead based on a cheap to measure ’index’ that iscorrelated with individual losses (e.g., average yields in a zone,or rainfall)
Cuts costsEliminates moral hazard & adverse selection
Example from recent study of individual versus (shadow) indexinsurance in Ecuador
M.R. Carter Index Insurance Impacts & Design
Logic of Index InsuranceComparison of Index versus Conventional Insurance in Ecuador
M.R. Carter Index Insurance Impacts & Design
Challenge of Index Insurance
Ecuador study relies on government collected yield dataavailable at high density at a national levelAlso based on relatively homogenous, coastal agriculture, nothighland areasBut, more generally, can an index be found that is stronglycorrelated with individual losses? If not, index insurance is alottery ticket, not an insurance and would not be expected toreduce the costs of risk & have real development impactsReview what have we learned about whether & when indexinsurance can fulfill its promise
Development impacts of index insuranceThe challenge of basis risk & sustainable impactDesigning better index insurance contracts
Technological solutionsInstitutional solutionsBehavioral insights
M.R. Carter Index Insurance Impacts & Design
Impact of Index InsuranceMaize producers in Ghana invest more when insured
Randomly offered some farmers insurance at variable pricesOther farmers offered a capital grant for purchasing inputsFound that farmers offered insurance:
Expand area cultivated by 15%Increase input use by 40%
Capital grants by themselves have little impactSource: Karlan et al. (2014). “Agricultural Decisions after Relaxing Risk & CreditConstraints,” Quarterly J of Econ.
M.R. Carter Index Insurance Impacts & Design
Impact of Index InsuranceCotton producers in Mali invest more when insured
Designed a dual-scale index contract with radically lower basisriskInsurance offered to a random subset of villages–find that ininsured villages:
Area planted to cotton increases by 15%Input use increases by 10-20%
Currently scaling up in Burkina Faso
Source: Elabed & Carter (2014) “Ex-ante Impacts of Agricultural Insurance: Evidencefrom a Field Experiment in Mali,” Working Paper
M.R. Carter Index Insurance Impacts & Design
Impact of Index InsuranceKenyan pastoralists reduce dependence on costly coping strategies
Drought insurance for pastoralists in Kenya launched in 2010Index based on satellite measure of forage availability (NDVI)
Insurance had payout in October 2011 after a prolongeddrought that sparked 30-40% livestock mortalityOctober 2011 survey asked insured and uninsured householdshow they had been coping with the drought prior to thepayout/survey & how they anticipated coping after thepayout/survey
Source: Janzen & Carter (2014) “After the Drought: The Impact of Microinsuranceon Consumption Smoothing and Asset Protection,” NBER Working Paper No. 19702.
M.R. Carter Index Insurance Impacts & Design
Impact of Index InsuranceKenyan pastoralists reduce dependence on costly coping strategies
Striking results consistent with poverty trap theory:Initially better off households (expected to be consumptionsmoothers) show:
Before Payout: No impact on consumption reduction nor onasset sales prior to payoutAfter Payout: 65 %-point reduction in asset sales after payout
Initially worse off households (expected to be asset smoothers)show:
Before Payout: 30 %-point reduction in “meals reduced” priorto payout; No impact on asset salesAfter Payout: 43 %-point reduction in “meals reduced” afterpayout; No impact on asset sales
M.R. Carter Index Insurance Impacts & Design
Challenge of Contract Design & Sustainable Impact
These kinds of development impacts can only be sustained ifcontracts work well, paying farmers when losses occurFailure of index insurance contracts to pay when farmers havegenuine losses is damaging at multiple levels
Hurts farmersHurts lendersDestroys trust & confidence in insurance (& insurancecompany reputation)
Sometimes spectacular failure of rainfall-based contractsreflects:
Poor correlation between rainfall and average farmer outcomesLow correlation between individual farmer outcome & averagebecause of index scale
An example from an analysis of rainfall contracts in India:
M.R. Carter Index Insurance Impacts & Design
Insurance or Lottery Ticket?
Source: Clarke et al. “Weather based Crop Insurance in India,” World Bank ResearchWorking Paper 5985.
M.R. Carter Index Insurance Impacts & Design
Solutions to the Problem of Basis Risk
3 classes of solutions:
Better predictors of average outcomes (area yield or satelliteinstead of rainfall)Institutional solutions (redistribution within groups at scale ofindex)Dual-scale index & audit rules
In addition, behavioral economics insights suggest furthersolutions, and willingness to pay for them
M.R. Carter Index Insurance Impacts & Design
Contract Design InnovationsRemote sensing: Evapotranspiration & other biomass growth measures
Several efforts to employ satellite measures that reflectvegetative growth
IBLI rangeland contracts in Northern Kenya and SouthernEthiopiaPlaNet Guarantee working with evapotranspiration measures inSenegal & Burkina
Strength is that it is very high resolution (e.g., 250 m x 250 mpixels)Still need more exploration about relationship between grainyields & biomass (skepticism from precision ag experts inCalifornia)New projects by IFAD/WFP (Senegal) & I4 (in cooperationwith the SI systems engineering company in Tanzania)
M.R. Carter Index Insurance Impacts & Design
Contract Design InnovationsAudit rules & gap insurance
Dual scale contract implemented in Ethiopia under IFPRI/I4projectPrimary index is rainfall basedFarmers can appeal for crop cut if fell index inaccurateSee how trust evolves over time
M.R. Carter Index Insurance Impacts & Design
Contract Design InnovationsDual-scale Contracts
Reducing scale of an insurance index has two impacts:
Reduces uninsured basis risk; but,Increases moral hazard
So why not Dual-scale contract?
One scale close to the farmer to reduce basis risk (reducesfalse negative & false positives)One at higher scale as an audit rule to check moral hazard(only pay if higher scale says likely that payout could haveoccurred naturally)
First effort in Mali showed results could be large:
M.R. Carter Index Insurance Impacts & Design
Contract Design InnovationsEquivalently priced single & dual-scale contracts (Mali)
M.R. Carter Index Insurance Impacts & Design
Contract Design InnovationsInsights from behavioral economics
Behavioral economics places individuals in controlledexperiments & observes how people behave, say, in the face ofriskFind that people do NOT consistently behave in wayseconomics traditionally predictsCalls into question models used to design insurance andpredict its uptakeLook at two areas where behavioral economics has been usedto understand contract design:
“Ambiguity aversion” & insurance demand in Mali“Certainty preference” & willingness to pay for insurance inBurkina Faso
M.R. Carter Index Insurance Impacts & Design
Index Insurance as an Ambiguous, Compound lottery
M.R. Carter Index Insurance Impacts & Design
Aversion to Ambiguity & Compound Lotteries
Long-standing evidence (Ellsberg paradox) that people areaverse to ambiguity & act much more conservatively in itspresenceSimilar empirical evidence of a similar reaction to compoundlotteriesPsychologically:
ComplexityIf people cannot reduce the lottery, then final probabilitiesseem unknown –> akin to ambiguity
Halvey (2007) shows in an experiment a link betweenambiguity aversion and compound risk attitudesImplemented field experiments in Mali to measure risk &compound risk aversion60% of the sample is compound risk averse, implying demandfor index insurance becomes hyper-sensitive to basis risk:
M.R. Carter Index Insurance Impacts & Design
Predicted Impact of Compound Risk Aversion on IndexInsurance Demand
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
Index Insurance Uptake as a Function of FNP
Probability of False Negative(%)
Fra
ction o
f P
opula
tion that W
ould
Purc
hase C
ontr
act (%
)
Assuming Expected Utility Theory
Assuming Compound−Risk Aversion
M.R. Carter Index Insurance Impacts & Design
Certain vs. Uncertain Utility
Andreoni & Sprenger propose a simple way to account forcommonly observed behavioral paradoxesSuggest that people strongly overvalue a sure thing more thanan almost sure thing (99.9% probability)If this ’overvaluation’ of outcomes that are certain is correct,implies that individuals undervalue insurance because the badthing (the premium) is certain and hence overvalued relative tothe good thing (payments), which are uncertain andundervaluedNote that overvaluation is above and beyond what would beexpected based on standard risk aversionConsistent with farmer complaints in the field about payingpremium in bad years
M.R. Carter Index Insurance Impacts & Design
Field Experiment in Burkina Faso
Working with 577 farmer participants in the area where we areworking with Allianz, HannoverRe, EcoBank, Sofitex andPlaNet Guarantee to offer area yield insurance for cottonfarmers, played two incentivized behavioral games:
Measured risk aversion over uncertain outcomes (α) andextent of certainty preference (β )Measured willingness to pay for insurance under two randomlyoffered alternative, actuarially equivalent contract framings:
Standard framing (certain premium)Novel framing (premium forgiveness in bad years)
Found that:
One-third of farmers exhibit certainty preferenceAverage willingness to pay is 10% higher under novel framingCertainty preference farmers value the alternative framing by25%
M.R. Carter Index Insurance Impacts & Design
Field Experiments in Burkina
M.R. Carter Index Insurance Impacts & Design
Willingness to Pay for insurance
All Certainty Others ExpectedPreference Utility
Average WTP (both frames) 15,771 15,208 15,573 16,492Frame A WTP 15,051 13,526 15,631 15,989Frame B WTP 16,493 17,397 15,521 16,950T-test (p-value) 0.09 0.01 0.9 0.5
Regression analysis (controlling for covariates, clusteringstandard errors, etc.) confirms these findings that Frame Bhas a large & significant impact on demand for the 30% of thepopulation that exhibits a strong preference for certaintySuggests that a simple re-framing of insurance can matter a lot
M.R. Carter Index Insurance Impacts & Design
In Conclusion
Know that risk is costly to poor rural householdsExciting new evidence that insurance can make a differenceHowever, cannot place rain gauges everywhere & think that wehave a viable basis for insuranceStill work to be done to craft technological and institutionalsolutions that will yield contracts that will work & that willappeal to people given how we behave in the face of risk
M.R. Carter Index Insurance Impacts & Design