trends, decadal processes, and rainfall insurance for ......trends, decadal processes, and rainfall...
TRANSCRIPT
Trends, Decadal Processes, and RainfallInsurance for Farmers in Ethiopia
Presented by Daniel Osgood, [email protected] International Research Institute for Climate and SocietyColumbia University
with Tufa Dinku, Arthur Greene, Eric Holthaus, KristopherKarnauskas, Brad Lyons, Michael Norton, Asher Siebert
What is Index Insurance?
Problems with traditional insurance havekept it from being available to most ofthe world
• Recent index innovation– Insure weather index not crop– Cheap, easy to implement, good incentives– Only partial protection– Only a couple of years old– Can a couple dozen pilots transition to a viable
industry that meaningfully addresses poverty atlarge scales?
• Insurance Physics: Ma$$ conservation law– Premium money covers payouts + cost of risk– Probability estimates of payouts necessary– Can include uncertainty in probabilities in price
IRI Insurance experience (and partners)• Some places:
– Ethiopia, Ghana, Honduras, Indonesia, Kenya,Malawi, Mali, Nicaragua, Nigeria, Rwanda,Senegal, Tanzania, Uganda
• Some partners:– CUCRED, DECSI, Ethiopia Met Service,
FIDES, ICRSAT, IIASA, Malawi InsuranceAssociation, Malawi Met Service, MRFC,Millennium Villages Project, NASFAM, NYALA,OIBM-MicroEnsure, Oxfam America, REST,SwissRe, World Bank-CRMG, The RockefellerFoundation
• Some summaries– Nearly 50 contracts designed– Purchased by thousands of farmers
Climate trends and insurance pricing• Prices must reflect best knowledge of
trends
• Care about altered probability of droughtnext year (not 2100)
• Can update prices each year withimproved probabilities
• Insured climate variability typicallydominates trend (~2%)
• Decadal processes often dominate trend
• Partners and policy-makers have limitedknowledge of decadal processes
• Insurance products in Morocco and theSahel have been challenged by decadalprocesses
Adapting insurance for adaptation
• Insurance as climate change adaptationmechanism– Increase resiliency– Build wealth– Price signals for transition
• Farmers in Malawi said that index insurance was toadapt to climate change
• Policy difference between long term trend anddecadal processes
• Need insurance adaptation based on best info
Nicaragua Insurance and precipitationtrends
• IPCC ¼% drying per year• Drying begins approximately 2025• No significant trends in historical rainfall,
index payments• If anything, payouts go down over time• Farmers may have
– Decades to build wealth– Decades to slowly adapt or transition as
prices slowly increase (and droughts slowlyincrease)
Historical Rainfall Historical burn payouts
Example in Ethiopia
• Insurance as part of a Development andAdaptation project
• Horn of Africa Risk Transfer for Adaptation(HARITA)
• Parthers: Oxfam America (PSD, HARO, HRD),SwissRe, IRI, REST, DECSI, NYALA, EthiopiaMet Service, Rockefeller, crop parametersfrom LEAP, Mekele University, builds ongroundwork by and software with WB CRMG,WFP, critique and input from Michael Carter
Insurance company concerned by trend in Adi Ha• 15 years of data
(NOAA ARC Remote sensing product triggers payouts)
• Non-significant trend in rainfall totals• Trend increased price by ~25%• Farmers in Ethiopia paid for this
• Is the trend due to– Climate Change?– Decadal Process?– Spurious noise?
Rainfall during contract window for Adi Ha ARC pixel
y = -1.7662x + 151.52
0
50
100
150
200
250
19951996
19971998
19992000
20012002
20032004
20052006
20072008
Year
Rai
nfa
ll (m
m)
Rainfall Total
Trigger
Linear (Rainfall Total)
Assuming trend increases Adi Ha Payouts
14.8%7.8%Mean--2008--2007--2006--200569.2%53.5%2004--20034.7%-2002--200144.4%13.0%2000--1999--199886.3%43.1%1997--19962.1%-1995DetrendedObserved
IPCC does not predict drying trend
But recent history suggests drying trend in Spring
• Climate Change?• Decadal Process?• Spurious?
• No insurance currently offered for Spring• What about fall when insurance offered?
Mar Apr May
Fall (index insurance season), little evidence of drying
• Regional rainfall analysis shows no fall drying
• “Trend” in indexed rainfall unstable– Goes from negative to positive
• With subtle changes in contract period• Removing a single payment year
• Fall insurance ‘trend’ may be spurious– Lower price?– Need to build evidence– Need improved pricing methods
• Ps.– Spring rainfall important for many other crops in
Ethiopia• Trend/decadal process is important
– Don’t forget temperature• Consensus on increasing temperature• Increases ET• Increases crop water processing
Large Region
Small Region
Thank you