www.rmsi.com elivering a world of solutions Index Based Insurance – an innovative approach towards agricultural risk financing mechanism Inderjit Claire October 2007
Mar 28, 2015
www.rmsi.com
Delivering a world of solutions
Index Based Insurance – an innovative approachtowards agricultural risk financing mechanism
Inderjit Claire
October 2007
www.rmsi.com
Delivering a world of solutions
Background – a general perspective
High exposure of low income countries to weather risks (drought, floods…), pests and diseases
Lack of insurance and other risk management tools Need of innovative approaches to deal with the nature of
agricultural risks
-40%
-35%
-30%
-25%
-20%
-15%
-10%
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Rain
fall
/AG
DP
-10%
-8%
-6%
-4%
-2%
0%
2%
4%
6%
8%
10%
12%
14%
GS
DP
%Dev of rainfall from Normal
%growth rate of AGDP
%growth rate of GSDP
Rainfall and economic performance in Andhra Pradesh
www.rmsi.com
Delivering a world of solutions
Background – Traditional Vs Index Based Insurance
Financial protection against adverse weather conditions Contracts can be structured as insurance or derivatives Based on the performance of a specified weather index during the risk
period Payouts are made if the index crosses a specified trigger level at the end
of the contract period Protect against yield volatility
Multi-peril Crop Insurance
High Administrative Costs
Moral Hazard Adverse Selection
Index-Based Weather Insurance
Rainfall is a proxy for damage
Objective triggers and structured rules for payouts
Improved correlation between need and provision
More on Index Based. . .
www.rmsi.com
Delivering a world of solutions
Goals
The goal is to meet the demands of various stakeholders
– Farmers
– Insurers
– Delivery channels
– Marketing agencies/organizations etc
Steps involved in the design and validation of the weather insurance pilots
– Demand assessment» Ensuring initiatives were in response to perceived and expressed needs of farmers and their
interest groups
– Identification of key insured parties : compulsory or voluntary
– Determination of key perils» The most important factor in insurance design
– Decision on crops to be covered
– Loss assessment procedures
– Rating
– Identify possible complementary role of Government
www.rmsi.com
Delivering a world of solutions
Weather Insurance Product Development – a ladder
www.rmsi.com
Delivering a world of solutions
Desk Research
Evaluation of weather and crop yield data– Availability of weather data
» Temporal» Spatial, » Limitations
– Availability of crop yield data» Temporal» Spatial» Limitations
Collection of relevant research– IMD Crop calendar
– Refer ICAR publications on Field and Plantation crops
– Meeting with agrometeorologists, agronomists» To identify essential crop pheonphase risks» Critical crop growth stages and their dependencies to rainfall affecting yield
– Review of insurance products being offered, if any!
– Local climatology
Stage 1 - Desk Research
Literature review-draw lessons from past experiences
Data availability
Crop yield dataCCE data/insurance units levelMandal/district level
Selection of area
weather data-Daily-weekly
Selection of crops- Identify preliminary risks
www.rmsi.com
Delivering a world of solutions
Data Analysis
Stage 2 - Data analysis
Sanity checks-Data consistency
-Data quality-Data limitation
1. Weather Data cleaning and enhancement- Replacement of missing and erroneous valuesthrough spatial and temporal interpolation/correlations
2. Detect data discontinuities-Through one or multiple stations values
4. Evaluation of crop yield data- For mandal/ district level validation as proxy of weather
3. Weather simulation (WXGEN)-To generate 100 yrs of data using Markov chain model
www.rmsi.com
Delivering a world of solutions
Field Research
Stage 3 - Field Research
1. Demand assessment- Examine the risk structure of specific key crops
2. Determination of key perils- a key factor in insurance design
3. Decision on crops to be covered- another key factor
4. Identify crop growing season- Key crop growth stages and duration
5. Significance of specific weather parameters- Analysis on index able weather perils
6. Critical/ strike for weather parameters- Definition of strike and exits for the payouts- Definition of daily rainfall floor- Periodic rainfall caps- Daily level excess rainfall limits etc
www.rmsi.com
Delivering a world of solutions
Contract Design
Stage 4 – Preliminary contract design
Identification of insurable risks (Key Perils)Production risks due to adverse weather conditions1. Drought/ Deficient rainfall
Sowing cover e.g. failed sowing/ germination failurePhenophase – wise cover e.g. Vegetative Growth, Flowering, Grain Filling etc.
2. Excess rainfall3. Frost (temperature based index)4. High Winds5. Satellite NDVI based cover
www.rmsi.com
Delivering a world of solutions
Weather Contract Design
Overarching Objective
To design contracts that cover both magnitude and frequency of deviations of key weather parameters
from requisite levels.
The challenge was to do it in an integrated manner (under a single cover) while ensuring robustness of the contracts
www.rmsi.com
Delivering a world of solutions
Weather Contract Design (Choice of Key Product Components)
Choice of caps and floors
Choice of triggers/ limits
Choice of indemnity payment levels
Choice of period & stages of cover
Interchangeability between indices and actual weather parameter values
Choice of sum assured under various covers
Overall flexibility in setting product parameters
www.rmsi.com
Delivering a world of solutions
Coverage of Weather Risks – Cotton, Mahabubnagar District
Harvest
Stages Sowing & Germination
Vegetative Growth & FloweringBoll Formation & Boll
Development
Deficient Rainfall for Vegetative Growth & Flowering
Deficient Rainfall for Boll Formation & Boll Development Perils
Inadequate Rainfall for Sowing
Covers for Crop Stages after Germination
15-55 days 50 days 40 daysPloughing
www.rmsi.com
Delivering a world of solutions
Bundled Product Payouts - Historical Data
Bundled Product Payouts (Historical Data)
0.0
1000.0
2000.0
3000.0
4000.0
5000.0
6000.0
7000.0
8000.0
1976 1979 1982 1985 1988 1991 1994 1997 2000 2003
Year
Bu
nd
led
Pro
du
ct P
ayo
uts
(M
ax.
Su
m A
ssu
red
Rs
8000
)
Bundled Product Payouts
www.rmsi.com
Delivering a world of solutions
Risk assessment
Risk profile of the areas – importance of weather risk in this profile
Availability of yield data and agronomic information
Issues related to basis risk – topographic make-up, presence of microclimates
Weather data and infrastructure – presence of weather stations, satellite information, historical data
Time period – key seasonal milestones
www.rmsi.com
Delivering a world of solutions
Data Availability
Rainfall data for Anantapur and Mahabubnagar Blocks– Field visit - 4 mandals in Anantapur and 6 Blocks in Mahabubnagar
– IMD weather data
– Anantapur » Rainfall data available for 18-25 years (upto 2003)
– Mahabubnagar » Rainfall data available for 16-18 years (upto 2003)
Crop Yield Data– Mandal level
– District level
Primary/Secondary information collected– Field survey
– Agro meteorological/ agronomic information
– Literature review – e.g., ICAR publications, iKisan etc
www.rmsi.com
Delivering a world of solutions
Data Availability
Crop Calendar
www.rmsi.com
Delivering a world of solutions
Probabilistic Drought Assessment – a framework
Production losses
Stochastic normal & drought events
Historical weather
Hazard Module
Simulated Weather Generator
Vulnerability ModuleCrop Yield
ModelPlanting Area
Model
CropsSoilMgmt.
Customized from rapid onset disaster modeling framework
Probabilistic drought risk assessment model
– Hazard module– Vulnerability module– Economic module
www.rmsi.com
Delivering a world of solutions
Hazard Module – Historical Weather
Mandal level Rainfall data used Is 1988 – 2003 data enough (18-25 years) for non-parametric
analysis?
Simulations produce more tail (extreme) events
www.rmsi.com
Delivering a world of solutions
Historical Weather – analyzing hazard
Data cleansing
- Sanity checks
– Spatial and temporal consistency
– No de-trending
Rainfall is lowest in Anantapur District
Coefficient of Variation is highest
– Anantapur
– Mahabubnagar
Rainfall risk is very high in these two districts
www.rmsi.com
Delivering a world of solutions
Hazard module – validation of stochastic weather events
Simulated Data
0
0.2
0.4
0.6
0.8
1
1.2
0 200 400 600 800 1000 1200
Rainfall (mm)
Ex
ce
ed
an
ce
Pro
ba
bili
ty
Historical Data
0
0.2
0.4
0.6
0.8
1
1.2
0 100 200 300 400 500 600 700 800
Rainfall (mm)
Exc
eed
ance
Pro
bab
ilit
y
The exceedance probability curves for both annual rainfall in historical and simulated data show the same trend (zero exceedance probability is equal to 800 in both the cases and rainfall with exceedance probability as 1 is just below 200).
Therefore the corresponding yield and rainfall curves from historical and simulated data are in line with each other.
www.rmsi.com
Delivering a world of solutions
EPIC simulated yield
– Generated at Mandal level
– Mandal level rainfall data used
– Management inputs taken from ANGRAU
– Field based inputs used
Reported yield
– Available at both Mandal and district level
Results validated
– drought years
Agro-met model
Vulnerability Model - Crop model
www.rmsi.com
Delivering a world of solutions 22
Vulnerability Mapping
NORMAL YEAR AVG YIELDMAIZE (Tonnes per hectare)
Less than 0.50.5 to 1.01.0 to 2.02.0 to 3.03.0 to 4.04.0 to 5.05.0 to 6.0Greater than 6.0Crop Not Grown
State of Andhra Pradesh
Crop Yield Simulations
- For each model
– Using historical data and then
– Simulated weather events
Vulnerability mapping @ Mandal level
Across six districts of AP
Covers both high and low vulnerable areas
0% 10% 20% 30% 40% 50% 60% 70%
Anantapur
Mahbubnagar
Kurnool
Cuddapah
Chittoor
Prakasam
Rangareddy
Nalgonda
Yield Losses in Drought Years (% normal year)
Severe
Moderate
Minor
www.rmsi.com
Delivering a world of solutions
Vulnerability Model - validations
Yield Deviation - Historical Data
-80%
-60%
-40%
-20%
0%
20%
40%
60%
80%
100%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Year
The yield deviations of the simulated data is in the same range as the historical yield data
Yield Deviations - Simulated Data
-60%
-50%
-40%
-30%
-20%
-10%
0%
10%
20%
30%
40%
1 35 69 103 137 171 205 239 273 307 341 375 409 443 477
Year
www.rmsi.com
Delivering a world of solutions
Annual Average Loss of Production Value, % normal year
0.0%1.0%2.0%3.0%4.0%5.0%6.0%7.0%
Prakasham_Maize_Production Loss_EP
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
Production Loss (% normal yield)
EP
Losses
Exceedence Probability loss
Average Annual loss
Probable Maximum loss
www.rmsi.com
Delivering a world of solutions
Key Issues
Technical Issues– Basis risk
» Introduction of Mandal-level weather stations can help to mitigate intra-district heterogeneities
» Farm level issue may prolong until density of weather stations improves substantially
– Daily level measurement of rainfall and its dissemination to end-clients» Regular and timely communication of weather data to facilitate better tracking of indemnities
– Asymmetries in geographical demarcation of insurance units needs consideration» Interpolation of weather data between existing and proposed weather stations to offset asymmetries due to
administrative demarcation
– Traditional MPCI will not get covered» Fire, hail storm etc
– Pricing at every Mandal will NOT carry the same degree of confidence» Shall depend on quality and availability of weather data applied
» Will require supporting field data and local knowledge
» Allow premium adjustment based on experiences and underlying risk
Administrative Issues– Geographical and administrative equivalent of Mandal does not exist in any other pilot state
except A.P.
– Banks covered during the field research have expressed difficulties in two insurance schemes
–
www.rmsi.com
Delivering a world of solutions
Early RiceLate Rice
Satellite based insurance contract design
www.rmsi.com
Delivering a world of solutions
Overview
12 June 2006 15 July 2006 18 August 200614 May 2006
FCC
Classified
Basis for approaching the new Classification – Spectral signatures
www.rmsi.com
Delivering a world of solutions
Crop mapping– Supervised Classification technique
– Area Estimation
– Accuracy Assessment
Ex: Wheat NDVI Analysis
Graph showing reported vs mapped acreage figures for wheat crop in the district. The difference is of 15.21 %.
NDVI analysis– NDVI value computation using available values
www.rmsi.com
Delivering a world of solutions
NDVI based insurance product for wheat in Farrukhabad tehsil of Farrukhabad district
Ex: Indemnity Payout Structure for NDVI Based Cover
Payout for NDVI Ranges in Farrukhabad Tehsil
0
10
20
30
40
50
60
174 to 173 176 to 175 180 to 179.5 181 to 180.5 182 to 181.5
NDVI ranges
Payo
t as P
erc
en
tag
e o
f S
um
Assu
red
Payouts
Payout Structure Crops having satellite derived rescaled mean NDVI values greater than 181 will receive no compensation
Between 181 and 180.5, farmers will receive 2 rupees for 0.5 value of deficit
Between 180.5 and 179, farmers will receive 4 rupees for 0.5 value of deficit
Between 179 and 175, farmers will receive 5 rupees per value of deficit
Between 175 and 173, farmers will receive 10 rupees per value of deficit
Payout structure is based on the slabs for a standard sum assured of Rs 100
www.rmsi.com
Delivering a world of solutions
At the household level:
– Gives farmers greater flexibility in investment decisions
– Banks have greater interest in lending
– Farmers see potential in investing in their farms
For governments:
– Provides government contingent financing
– Allows the cost of drought risk to be smoothed over time
– Provides some predictability to drought financing and buys time for other emergency responses to take affect
– Provides government a level of autonomy
Conclusions