Modeling Multiple Peril Crop Insurance Worldwide Jack Seaquist CARe Seminar C-7 Philadelphia, PA June 7, 2011 www.air-worldwide.com 1 AIR Agricultural Model Applications Underwriting Risk Transfer Enterprise Risk Management AIR Agricultural Model for Crop Insurance and Reinsurance Pricing & Actuarial Claims The AIR Agricultural Model for the Multiple Peril Crop Insurance is a state-of-the-art fully probabilistic model, accounting for: Changes in technology over time The evolution of crop insurance programs Commodity price volatility Sales and Producers Weather impact 2 1
Probabilistic Agricultural Model Components HAZARD DAMAGE Commodity Price Risk AWI Weather Risk Stochastic Event Set Catalog Crop Damage / Yield Estimation FINANCIAL Insured Loss Calculations Exposure Information Government Reinsurance Policy Conditions Crop Insurance Program Private Reinsurance 3 Probabilistic Agricultural Model Components HAZARD DAMAGE Commodity Price Risk AWI Weather Risk Stochastic Event Set Catalog Crop Damage / Yield Estimation FINANCIAL Insured Loss Calculations Temperature Terrain Elevation Historical Yields Crop Requirements Precipitation Historical Prices Soil Land Use Land Cover (Satellite) Typhoon Catalog Exposure Information Policy Conditions Crop Insurance Program Government Reinsurance Private Reinsurance 4 2
AWI (Agricultural Weather Index) is a Measure of Yield Variability Due To Weather Daily Temperature Daily Precipitation Crop Specific Data County- Specific AWI Index Available Water Capacity (Soil) 5 In a Normal Year Water Supply and Water Requirements Are Balanced and AWI Indicates Positive Yield Outcome Soil Moisture Runoff Evapotranspiration Detrimental to Yields Favorable for Yields Apr May Jun Jul Aug Sep Apr May Jun Jul Aug Sep 6 3
In a Normal Year Water Supply and Water Requirements Are Balanced and AWI Indicates Positive Yield Outcome Soil Moisture Runoff Evapotranspiration Detrimental to Yields Favorable for Yields Apr May Jun Jul Aug Sep Apr May Jun Jul Aug Sep 7 In A Drought Year Water Requirements Exceed the Water Supply and AWI Indicates Plant Damage Soil Moisture Runoff Evapotranspiration Detrimental to Yields Favorable for Yields Apr May Jun Jul Aug Sep Apr May Jun Jul Aug Sep 8 4
AWI Measures County-level Crop Performance During the Season End of Season AWI estimates yield variation from normal AWI AWI proportional to flood damage AWI indicates significant drought damage Calendar Days 9 AWI Removes Weather Variability from the Historic Yields to Reveal True Technology Trend Yield Time Series Yield - Bushels/Acre y 80 100 120 140 160 180 200 Trend Line Influenced by Weather Variability True Technology Trend with Weather Variability Removed 17 Bushels 1990 1995 2000 2005 2010 2011 AIR WORLDWIDE x CORPORATION 2008 10 5
The Choice of Methodology Can Have a Large Impact on Insurance Contracts Probability Density Function of Detrended Yield Time Series Probability 0.000 0.005 0.010 0.015 Yield Guarantee With AWI 100 150 200 250 100 150 200 250 Yield Bushels/Acre Without AWI 11 In The US Model AWI Technology Adjusts Historic Time Series of County Level Yields to Current Production Levels The catalogue is based on spatially correlated county level yield distributions All weather events are summarized by the final modeled harvest yield 10,000 outcomes based on variation from historical data Current technology level Not biased by recent event years 12 6
The Price Risk Component Takes Into Account Correlations Between Total US Production and Harvest Price CORN - PRICE RATIO Harvest Price Stochastic Catalog Historical 180% 160% 140% 2010 1988 1983 1993 120% 100% 80% 2008 60% 40% 60% 70% 80% 90% 100% 110% 120% 130% Yield - % of Normal 13 New SRA Reduced the Overall Profitability of the Industry and Increased Protection for Catastrophic Events 200-6% -3% Loss Ratio -1% 0% 1% 100 100 3% 4% Old SRA New SRA 0 Average 2 years (Median) 10 yr (10%) 20 yr (5%) 50 yr (2%) 100 yr (1%) 500 yr (0.2%) Return Period 14 7
Impact of 2011 SRA Change - Illinois 200 3% 1% -2% 8% 7% Loss Ratio 100 100 11% 16% Old SRA New SRA 0 Average 2 years (Median) 10 yr (10%) 20 yr (5%) 50 yr (2%) 100 yr (1%) 500 yr (0.2%) Return Period 15 Impact of 2011 SRA Change - Iowa 200 5% 0% -1% -3% 4% Loss Ratio 100 100 12% 19% Old SRA New SRA 0 Average 2 years (Median) 10 yr (10%) 20 yr (5%) 50 yr (2%) 100 yr (1%) 500 yr (0.2%) Return Period 16 8
Impact of 2011 SRA Change North Dakota 200-7% -12% -14% -18% -2% Loss Ratio 100 100-2% -1% Old SRA New SRA 0 Average 2 years (Median) 10 yr (10%) 20 yr (5%) 50 yr (2%) 100 yr (1%) 500 yr (0.2%) Return Period 17 Impact of 2011 SRA Change - Texas 200 Loss Ratio 100 100-2% -4% 0% 0% 0% -1% -3% Old SRA New SRA 0 Average 2 years (Median) 10 yr (10%) 20 yr (5%) 50 yr (2%) 100 yr (1%) 500 yr (0.2%) Return Period 18 9
New 2011 SRA Fund Allocation Rules Commercial Fund* (by State) Policies Minimum of 25% of premium in each State + 35% minimum * No more sub-funds (Cat/Buy/Rev) Maximum of 75% of premium in each State Retained Premium Ceded to Government Assigned Risk Fund (by State) 20% = Total Net Retained Premium 19 Fund Allocation Process with AIR MPCI Model Policy Records Probability of Loss State-Level Fund Limits Allocation Strategies Allocate to Funds Alternative Retention Levels Calculate Pre-SRA and Post-SRA Loss Distributions Probability of Loss Distributions for Each Policy Bundle -- State, County, Crop, Insurance Plan, Coverage Level AIR MPCI Model Catalog 20 10
Crop Insurers SRA Fund Allocation Strategies Greatly Impact Reinsurer Risk 60% Gain/Loss Excedence Probability Comparison (percentage of premium) 40% 20% Gain -% of Premiums 0% -20% Median A B C D E -40% -60% -80% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Probability 21 China Crop Insurance Program Covers Cost of Production Rather than the Value of the Crop Separate programs by province Perils covered and policy terms vary by crop and province Named perils might include Rainstorm, flood, windstorm, hail, waterlog, drought, frost, diseases, insect pests, tornado, mud-rock flow, landslide, low temperature Indemnity varies by plant growing stage Losses shared and limited by governments 22 11
AWI (Agricultural Weather Index) is a Measure of Crop Damage Due to Weather Daily Temperature Daily Precipitation Available Water Capacity (Soil) AWI Crop Specific Data Land Use Land Cover Terrain Elevation 23 A Comprehensive Approach Captures Flood Losses from Multiple Sources Derived from historical years 1980-present Snowmelt Derived from Asia-Pacific Typhoon model Typhoon Low Locally Excessive Moisture Risk High Risk AWI Soil Moisture AWI Runoff 24 12
Higher Resolution Damage Estimates Based on AWI Are Validated against Province Level Losses Wet Dry Damage to AWI Model for Local Effects Crops on Local Level Losses for 2007 Drought by Province Wet Dry Accumulated Effects High Resolution Crop Mask 25 Closing Thought What if the 1993 Flood Happened in 2011? 1993 Corn Yield (Percent of Expected Value) State Actual 1993 RMA Gross Loss Ratios AIR Modeled (Recasted) Gross Loss Ratios AIR Modeled (Recasted) Post-SRA Loss Ratios Iowa 465% 247% 168% Illinois 63% 21% 66% Indiana 55% 22% 67% Ohio 91% 54% 61% Minnesota 610% 214% 161% Nebraska 188% 80% 83% US Total 219% 106% 97% Key Differences Between 1993 Flood and 2011 Flood: Timing, Location, Policy Mix, Premium Rates, SRA Program Terms 26 13