1

Applications to Agriculture and Other Sectors: The Role of Weather Index Insurance

Dr. Jerry SkeesH.B. Price Professor, University of Kentucky, andPresident, GlobalAgRisk, Inc.

US CLIVAR/NCAR ASP Researcher ColloquiumStatistical Assessment of Extreme Weather Phenomena under

Climate ChangeNCAR Foothills Lab, Boulder, Colorado, USA

June 13-17, 2011

Special Thanks to GlobalAgRisk Team!

2

GlobalAgRisk, Inc.

Activities Research and development tied

to University of Kentucky research program

Technical capacity building Educational outreach

Supported by Multinational donors Governments Nongovernment organizations

MissionImprove access to financial services and the value chain for the rural poor through innovative approaches for transferring weather risk Select Country Work

Peru – El Niño/Flood Mongolia – Livestock Vietnam –

Flood/Drought Indonesia –

Earthquakes Mali – Drought Morocco – Drought Mexico – Drought Romania – Drought Ethiopia – Drought

Current Support for GlobalAgRisk, Inc.

Bill and Melinda Gates Foundation (Peru)

Ford Foundation (Vietnam and Indonesia)

Gov’t of Mongolia via Swiss Trust Fund

UNDP (Peru)GiZ (Peru)Risk Management Agency of USDA

Actuary and Underwriting Reviews3

State of Knowledge Reports from GlobalAgRisk

Supported by Bill and Melinda Gates Foundation Innovation in Catastrophic Weather Insurance to Improve the Livelihoods of Rural Households

March 2010 “Data Requirements for the Design of Weather

Index Insurance.” March 2011

“Market Development for Weather Index Insurance Key Considerations for Sustainability and Scale Up.”

Under Revision: “Legal Considerations for Index Insurance”

Forthcoming a book

4

Somebody Always Pays for Catastrophic Risk Who? How?

The poor pay through direct losses and long-term economic impacts

Financial institutions restrict services as they learn that the correlated losses of many of their borrowers and savers create significant banking problems

Governments — Disaster relief and recovery expenses, infrastructure investments, subsidized agricultural insurance

Donors forgive debt and divert funds for recovery

Society needs to understand the cost of natural disaster risk

Someone always pays

Need incentives for proper risk management and mitigation5

Designing Sustainable and Scalable Weather Index Insurance Programs Is Challenging

Products must be developed in contextCostly (technical support, capacity building, R&D)Not easily replicable

Basis riskTradeoff between transaction costs and basis risk

Limited or no data to develop productsHigh delivery costsSmall transactions/Small market volumeNascent legal and regulatory systems

6

Lessons LearnedRisk Assessment

Understanding vulnerability necessitates examining a specific context:The risk, geography, demographics, risk

management strategies, government policies, etc.

Risk assessment requires a systems approach

Proper risk assessment benefits decision makers and can guide policy decisions

Cognitive failure – decision makers often underestimate extreme risk of natural disasters

7

AdaptationMultilateral Organizations Call for Insurance

Decision makers suggest insurance can play an important role in adaptation to climate changeArticle 4.8 of the United Nations Framework

Convention on Climate Change (UNFCCC) Article 3.14 of the Kyoto ProtocolThe Hyogo Framework

Agricultural insurance, in particular, gains increased attention because agriculture is the primary livelihood of households, especially the poor, in developing countries

8

Insurance and AdaptationInsurers assess specific risksInsurers price the risk – a critical aspect of risk assessmentInsurance can encourage adaptation through

pricing – if insurance costs that much, maybe I should do something different!

Access and price of insurance can be directly linked to adaptation strategiesFire insurance: extinguishers in the house? Proximity to a fire hydrant Improved building codes

Insurance reduces variability in wealth/revenues

9

Pricing the Risk“Fairness” and Perverse Incentives

If insurance is not priced to reflect differences in

relative risk (or if more subsidies are paid to higher

risk areas), perverse incentives are likely to follow

– if you pay people to take risk, they will take more

risk

The result will be more risk exposure rather than

lessHigher risk activitiesHigher risk areasLess investment in risk reduction

10

In the U.S. Subsidies Have Been Used to Lower Loss Ratios

Coverage Level

1980 Act 1994 Act2000

ARPA

55 30.0 46.1 64.0

65 30.0 41.7 59.0

75 16.9 23.5 55.0

85 --- 13.0 38.0

11

Increased Risk-Taking

As a result of purchasing insurance, farmers tend to engage in riskier behavior (plant higher risk crops, monoculture, less irrigation, etc.)

Difficult for insurer to monitor these changes in behavior

Increased risk-taking increases loss ratiosHigh loss ratios lead to higher premium ratesHigher premium rates reduce insurance

purchasingHigher premium subsidies are then needed to

maintain or increase insurance purchasing

12

U.S. Agricultural Policy Has Also Favored More Risk Areas

If you pay people to take on more risk; they will take on more risk!

Program yields for commodity payments are based on planted acre yields (not harvested acre yields); favored areas with large abandoned acres

Ad hoc disaster payments (long history)

Highly subsidized crop insurance

Premium subsidizes are a percent of premium: Once again favoring high risk areas

13

Premium Subsidy per Dollar of Liability2002

14

Loss Ratios Vary Across the Country1981–2002

Change in Crop Share

Legend

Highest Loss of Crop Share

Moderate Loss of Srop Share

Small Loss or Gain of Crop Share

Moderate Gain of Crop Share

Highest Gain of Crop Share

Missing Data 16

Farmer Knowledge and Decision Processes

Literature shows Farmers optimizeFarmers adaptFarmers are good Bayesians Farmers know central tendency on yieldsCognitive failure sets in for catastrophic

eventsChallenge for adaptation

Communicating information about climate change in a fashion that is useful for decision makers

17

Lessons LearnedPricing the Risk

Pricing risk is an important aspect of insuranceInsurance can encourage households to

reduce risksGovernment policies can motivate households

to take more risk at government’s expenseAgricultural insurance is extremely complex

Risk classification is difficultRisk varies by commodityRisk varies by regionRisk varies by producer

Difficult to monitor farmer behaviorAdministrative cost is very high

18

19

Weather Index Insurance: Gaining Popularity in Developing Countries

Payouts are based on a measurement of an event correlated with losses Weather event (e.g., rainfall, temperature, or river levels

associated with yields)

Insurance pays when index crosses a certain threshold

Payouts do not require inspection of client lossesEliminates most information asymmetry problemsTransfers extreme risk to reinsurers

Example of an Excess Rainfall Insurance Product

Extreme rainfall in India – payments would occur anytime rainfall exceeds 2000 mm

Household might buy US$100 in liability

20

500 1000 2000 3000 40000

Payout Structure for Example Excess Rainfall Contract

$1 for every 10 mm excess of 2000 mm$100 limit at 3000 mm

21

Pricing Index Insurance

Price of Index Insurance =

22

Index Insurance Creates Flexibility for Different Users

HouseholdsIndia: Drought coverage to groundnut farmers Malawi: Cooperative links insurance to farmer

loans for high-yield seed varietiesMongolia: Index-based livestock insurance

Firms (e.g., Banks/input suppliers)Increase access to credit and other servicesVietnam and Peru: Possible use of index

insurance for protecting credit riskGovernments/Donors

Mexico: State governments provide quick drought relief to farmers in need using a drought index

23

Sahel: Shifts in the Central Tendency

24

SahelSemi-arid region below the Sahara

Sahel data 1900 – 2007

• Dynamic climate largely due to oceanic oscillations

• Unlike the Sahel, climate change may lead to more permanent changes

Sahel: Hypothetical Insurance ProductInsurer entry point at 1962, 1990, and 2007 to develop a drought insurance

Insurer would re-center data arounda current forecast of the central tendency

25

Insurance contract for wheat farmersIndemnities when rain is below 425 mm

Central Tendency: 510 mm Payout Threshold: 425 mm Pure Risk: 2%

Sahel: Rainfall Distribution Using Data from 1900 to 1961 (Reference Point – 1962)

Pure risk will likely result in insurance affordable to

households

26

Central tendency is below payout thresholdInsurance is inappropriate in this setting

Sahel: Rainfall Distribution Using Data from 1962 to 1989 (Reference Point – 1990)

Pure risk would make insurance unaffordable for

households

Central Tendency: 328 mm Payout Threshold: 425 mm Pure Risk: 44%

27

Increases in rainfall reduced the weather riskInsurance would be affordable depending on other costs

(Delivery, ambiguity loading, etc.)

Sahel: Rainfall Distribution Using Data from 1990 to 2006 (Reference Point – 2007)

Pure risk may result in affordable insurance for

households

Central Tendency: 456 mm Payout Threshold: 425 mm Pure Risk: 6%

28

Climate Change Increases Ambiguity and Catastrophe Loads

Misestimating the central tendency is very costlyGreen distribution Insurer forecast in 1962Blue distribution Forecasted loss experience based on 1990 Reference

29

Adaptation, Policy Interventions, and Climate Change

Households must adapt or experience increasing farm losses1.Change farming practices2. Invest in infrastructure3.Transition out of farming

Insurance, by itself, is not a means of adaptationPolicy interventions should assess the opportunity

costs when supporting insurance programs – other adaptation investments may be more important

Insurance can be used to facilitate adaptation (e.g., linking insurance, credit, and improved seed varieties

in Malawi)

30

Premium Subsidies Will likely Slow Adaptation

Significant caution is needed for insurance premium subsidies

If climate change shifts the central tendency, premium subsidies could slow adaptation

Households account for subsidies when comparing profits from farming and other activities

These subsidies can encourage farmers to take more risk and delay adapting to climate change

31

How to Improve Access to Catastrophic Weather Insurance? Our Experience Suggests . . .Index insurance is best suited for catastrophic and

consequential lossesIndex insurance that addresses weather risk of

firms that serve the poor (risk aggregators) presents a feasible avenue for market growth; build a sustainable market first and then move to micro products

Household products must find innovative delivery mechanisms to improve product affordability and offer value to clients (insurance-linked products)

Solutions that involve public-private partnerships must clearly delineate the role for markets and the role for governmentUnderstanding cognitive failure for extreme risk

can helpRisk layering – Putting catastrophic insurance into

a broader conceptual framework33

Risk Aggregator Products Are Less Costly to Develop and Implement than Household Products

Risk aggregator products face lower basis riskRisk aggregators effectively diversify much of the

idiosyncratic risks born by their clients

Data constraints are less binding for risk aggregator products

It is more cost effective for the insurer to establish a partnership with a risk aggregator than to market and distribute products to small holders

Risk aggregators are more likely to understand hedging and basis risk

34

El Niño Insurance for FloodInnovation in Northern Peru

35

Piura and other areas in the NorthSeverely affected by 1998 El Niño

Extreme rains (Jan – Apr 1998) 40x normal rainfall

Severe floods 41x normal river volume

Widespread losses Many disrupted markets Agricultural production, ↓ 1/3 Public infrastructure losses Cash-flow, debt repayment

problems Health problems

Total losses in Piura estimated at USD 200 million

Contract is Written Using NOAA Data

Nino estimates are derived from satellite data, observations of buoys and readings of the temperature on the surface and at deeper levels.

The data are publicly available monthly from NOAA (The U.S. National Oceanic and Atmospheric Administration)

http://www.cdc.noaa.gov/Correlation/nina1.data

38

Strong El Niño in 1982-83 and 1997-98

21

22

23

24

25

26

27

19

79

19

80

19

81

19

82

19

83

19

84

19

85

19

86

19

87

19

88

19

89

19

90

19

91

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

ENSO 1.2 Nov and Dec

0

200

400

600

800

1000

1200

1400

1600

1800

2000

19

79

19

80

19

81

19

82

19

83

19

84

19

85

19

86

19

87

19

88

19

89

19

90

19

91

19

92

19

93

19

94

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

Precipitaciones Aeropuerta Piura

2 extreme events in the last 32 years

37

31

Correlation Matrix for top 10 percentile El Niños

40

Correlation MatrixSouth North ENSO1.2ENSO3

South 100% 90% 92% 89%North 100% 90% 90%ENSO1.2 100% 100%ENSO3 100%

Example of a Payout from the 1997 Event- Nino 1.2 (Nov-Dec) temperature = 26.28°C

The insured selects the sum insured

Minimum payment = 5%

Sum insured = 10,000,000 Soles

1998 payment = 76% x 10,000,000 = 760,000 Soles

Primary Goal:Improve Access to and Terms of LoansCapacity building with

Financial institutionsPeruvian banking regulatorPeruvian credit rating agenciesSources of social capital flows into Peruvian Institutions

Case to be made1)Strengthen the resiliency of the financial

institution 2)Financial institution can be ready to lend when

the community needs capital the most – post disaster

42

Risk Aggregator StrategyNatural Disaster Effects on BankingLoan portfolio — Systemic repayment problems

for borrowers, problems can remain for yearsDeposits — Depositors withdraw fundsCosts increase — Costs of funds (e.g., Interbank

loans), administrative costsResulting problems

LiquidityProfitabilityCapital Adequacy

Lending institutions have many ways of managing these risks (e.g., Provisions, restructuring loans, etc.)

43

1997–1998 El Niño Spike and Recovery

10% Spike 3.5-Year

Recovery

With this event every 1 in 15 years, 300 basis points must be added

PROBLEM

LOANS

44

45 45

Default Risk Significantly Affects Interest Rates!

π – Expected profitsp – Exogenous probability of non-

defaulti – Interest rater – Lender’s opportunity costsL – Amount of funds loaned

LrLip 11 11 p

ri

Example (No default risk)

r = 10%p = 100%

Example (10% default risk)

r = 10%p = 90%

El Nino may add 300 basis points to interest rates

Historical Pattern of Agricultural Lending in Piura1994–2006

0%

2%

4%

6%

8%

10%

12%

14%

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

Year

Per

cen

t o

f L

oan

s to

Ag

ricu

ltu

re

El Niño

Lenders say they have“fixed the problem” by not making

loans when they see El Niño coming

46

El Niño Reduces Capital Adequacy and the Ability to Leverage and Make Profits

47

Comparison of Sum Insureds using Monte Carlo

48

% is the sum insured corresponding to a percent of credit portfolioe.g., Sum insured = 5% of credit portfolio

Risk Assessment Includes Evaluating Current Risk Management Strategies

Potential strategies for managing these risks and their costs

Liquidity Hold higher portion of assets in cashEffect — Reduces investment in productive assets

Profitability Avoid exposed regions and sectorsEffect — Limits growth opportunities, especially for

untapped marketsCapital adequacy Leverage a lower amount of

equity to provide a “cushion” for the riskEffect — Limits growth

49

50 50

Mongolia — Massive Deaths of Animals

Mongolia: 45 million animals at the start of 2010Sheep, goats, cattle and yak, horses, camelValue of animals = US $2 BillionSome 11 million animals were lost in 2001–2002

due to dzud (harsh winter weather).. 9.7 million were lost in early 2010!

Animal husbandry in Mongolia is 20+% of the GDP and over 85% of all agriculture

Census is done every year — Mortality data are available by soum (county) from 1970 onwards

51

Index-based Livestock Insurance — Risk Layering A New Model for Public-Private Partnerships

GCC — Social Insurance

A layer of very infrequent risk where decision makers may have a cognitive failure problem

LRI — Commercial Insurance

Offered by private companies with reinsurance from government and now a global reinsurer

Government Catastrophe Cover (GCC)

Paid by governmentusing a World Bank

contingent loan

Livestock RiskInsurance (LRI)

Retained by Herders

6% mortality

30% mortality

100% mortality

If the government can’t continue to pay for extreme losses, the

commercial layer can continue

52

Climate Drivers Matter: Arctic Oscillation

-5

-4

-3

-2

-1

0

1

2

3

1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 2009

1944-45 33 percent

2009-201022 percent mortality - Feb value = -4.2

Underwriting MattersIndex insurance can indeed address many

of the adverse selection and moral hazard problems associated with traditional forms of insurance

However, sitting sales closing dates still maters and the insured can adversely select using weather forecast information!

54

Scale MattersNone of our efforts will succeed unless we

design products that capture the attention of the regulator and the market from the outset

Start with biggest risk targeted to risk aggregators Rural lendersValue ChainFarmer Associations

Carefully move to micro products with concept of livelihoods insurance for consequential losses suffered by small households; challenges will remain for demand and delivery

55

Visit our Website for Papers, Projects and More

www.globalagrisk.com

56