Www.rmsi.com Delivering a world of solutions Index Based Insurance – an innovative approach towards agricultural risk financing mechanism Inderjit Claire.

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


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


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


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


Weather Insurance Product Development – a ladder

www.rmsi.com


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


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


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


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


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


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


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


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


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


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


Data Availability

Crop Calendar

www.rmsi.com


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


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


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


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


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


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


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


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


Early RiceLate Rice

Satellite based insurance contract design

www.rmsi.com


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


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


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


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

Www.rmsi.com Delivering a world of solutions Index Based Insurance – an innovative approach towards agricultural risk financing mechanism Inderjit Claire.

Documents

weather data cleaning

proxy of weather

key weather parameters

specified weather index

world of solutions index

cover slide

weather parameters definition

evaluation of crop yield