9/17/2012
Production Policies Rating Methods 2012 Minnesota Crop Insurance Conference Mankato, MN September 12, 2012
Overview • • • •
The Production Plan Concept Performance of the Program Ratemaking Methodology Implications for Crop‐Hail Data Reporting and FALC Estimation
1
9/17/2012
The Production Plan Concept
Crop‐Hail vs. MPCI Protection • Crop‐Hail (CH) – Covers all or most of a crop
• MPCI (MP) – Pays the portion of hail loss in excess of the MP deductible
• Coverage may overlap – Large hail losses can be paid under both policies
100 80 60 40 20 0 Crop‐Hail Liability (Basic)
MPCI Liabilty (75%)
Hail Damage (40%)
2
9/17/2012
Companion Plan (CP) • Caps loss at MP deductible – Any remaining loss is paid by MP policy – Pays small losses fully – Reduces overlap – Claims adjusted independently of MP – Reduces premium • Popular in high rate states
– Per acre basis
100 80 60 40 20 0 MPCI Liabilty (75%) + CP4
Hail Damage (40%)
Some Overlap Still Exists • Arising from use of hail loss adjustment charts – Represents average yield loss due to hail – Actual yield loss will differ for each claim
• Crop may recover from hail damage – Under favorable growing conditions – May compensate for loss due to hail
3
9/17/2012
Example of Overcompensation • CP policy – Hail damage of 23% = 23 bushels – Indemnity = 23 bushels
• MP policy – Guarantee = 75 bushels – Production to count = 90 bushels – Actual loss = 10 bushels – No MP indemnity
• Indemnity exceeds actual loss – Payment exceeds loss by 13 bushels
Production Plan (PP) Policy • Similar to Companion Plan • Covers losses within the MP deductible – Limit of insurance = MP deductible – Direct coordination of benefits with MP policy • Pays smaller of hail loss and MP loss – Accounts for ability of crop to recover
• Caps larger losses at the MP deductible – Any remaining hail loss is paid under MP policy
• Provides coverage on a unit basis
– In prior example, PP would pay only 10 bushels – Lower premium than CP with similar protection
4
9/17/2012
APH Modifier Feature • Increases the amount of PP hail protection – Yields tend to increase over time – APH may understate producer’s expected yield – Allow mods up to 1.20
• Hail Guarantee = APH x Selected Modifier • Total Liability = Hail Guarantee x Acres x Price • PP Amount of Insurance – Difference between Total Liability and MP Liability – Increases as Mod rises
APH Modifier Feature PP Amount of Insurance Increases as Mod Rises MP
PP
120 100 80
25
30
35
40
45
75
75
75
75
75
60 40 20 0
5
9/17/2012
Adjusting PP Losses • Complicated • NCIS Loss Adjustment procedure: – Determine the hail damage in bushels • Hail Guarantee x Damage %
– Compute the MP loss of production in bushels • Hail Guarantee – Production to Count
– Use smaller of Hail Loss and Production Loss – Convert to dollars at the base price – Apply the minimum loss % – Cap indemnity at the PP Limit of Insurance
Hail and Production Loss vs. APH Mod • Hail loss increases slowly as Mod rises (at 30% of Guar) • Production loss increases rapidly (at 100% of Guar) • PP pays the smaller amount usually the Hail loss – At high mod, PP is like CP coverage with high $ Ins per Acre
6
9/17/2012
Program Performance
Production Plan Market Share Liability: 2005 to 2011
50%
40%
30%
20%
10%
0% CO
IA
ID
IL
IN
KS MN MO ND NE OH
SD
US
7
9/17/2012
Production Plan vs. Crop‐Hail Loss Ratios (US) RMA: Is PP being used as an inducement to purchase MP? 200 175
PP Average LR: 187% CH Average LR: 77%
150 125 100 75 50 25 0 2005
2006
2007 2008 2009 Prod Plan LR Crop‐Hail LR
2010
2011
NCIS Involvement • Industry asked NCIS to file Production Plan – Standardized policy language – Rating factors
• NCIS Production Plan Filings – 2012 • • • •
IA, MN, NE, KS, ND, SD Corn, soybeans, wheat Coverage levels: 50% ‐ 85% Modifiers: 1.00, 1.05, 1.10, 1.15, 1.20
– 2013 • CO (Corn, Wheat, Potatoes) • ID (Barley, Wheat, Potatoes)
8
9/17/2012
Ratemaking Methodology
Production Plan Ratemaking Issues • Historical data isn’t usable – Too few years of data – Data is subdivided into various options • Coverage Level • APH Mod
– Wide variety of company coverages • Deductibles • Minimum Loss Percentages • Indemnification provisions (e.g., Basic, XS10IP)
– Concerns about accuracy in adjusting claims – Inability to restate all plans on common basis
9
9/17/2012
Payout Comparison for Different Insurance Plans APH CV = 0.5, APH Modifier = 1.00, MPCI Coverage = 65%, Acres% Damaged = 75% 0.40
0.35
0.30 CP(IP 2, 5%DED, 50% MPCI CL) CP(IP 3, 5%DED, 65% MPCI CL)
Payout %
0.25
CP(IP 4, 5%DED, 75% MPCI CL) Comp 1 (0 DED)
0.20
Comp 1 (5% DED) Comp 2
0.15
Comp 3 Comp 4
0.10
Comp 5 (5% DED)
0.05
0.00
0%
10%
20%
30%
40%
50%
60%
70%
80%
Hail Loss % on Unit
Selected Ratemaking Methodology • Simulation Method – Recompute indemnities for individual claims • Under each Production Plan option
– “Monte Carlo” approach • If we know now how the components behave • Combine the components to estimate the whole • Assumes we know how the components interact
– Ensures consistency between rates at various Coverage Levels and Mods – Less certainty that the overall rates will be accurate • But, the Companion Plan policy form factors can be used as a test of reasonableness
10
9/17/2012
Simulating Indemnities • Hail damage % – Use NCIS hail severity distribution
• Producer’s yield loss % – Use producer yield distribution implied by MP rate
• Givens: – Coverage level – APH modifier – Minimum loss (5%) – PP Limit of Insurance
• Size of unit needs to be considered
Hail Severity Distribution • Choices: – Actual hail losses under Basic form (selected PP method) – Fitted distribution based on all reported claims (used to develop policy form factors for all but Production Plan) Sample Hail Severity Distribution
Probability
0.05 0.04 0.03 0.02 0.01 0.00 1
11
21
31
41
51
61
71
81
91
Percent Damage
11
9/17/2012
Producer Yield Distribution • Find the average base rate for the state • Adopt the censored normal yield distribution – Used in Combo policy rating
• Determine the specific yield distribution that matches the rate
Harvested Yield Distribution Censored Normal with APH = 100 The distribution corresponds to a specific rate
0.009 0.008
Density Function
0.007 0.006 0.005 0.004 0.003 0.002 0.001 0
0
50
100
150
200
250
Yield
12
9/17/2012
Simulate Production Plan Indemnities • For each hail claim: – Determine the hail damage % – Determine the % of acres damaged by hail • Accounts for differences in insured acres • Used to include the effect of unit size
– Select the percentage yield departure • From the producer yield distribution • Apply separately to damaged and undamaged acres
– Calculate the indemnity • Repeat process for randomly selected yield departures
• Repeat for each coverage level and APH Mod
Production Plan Policy Form Factors • Find average indemnity over all claims – PP vs. Basic form
• Policy Form Factor = – PP Indemnity / Basic Indemnity – PP is by Coverage Level and APH Mod
• Compare to Companion Plan policy form factors – Coverage Levels 50, 65, 75 – PP APH Mod = 1.00
13
9/17/2012
Smoothing the Final Factors • Want consistent rates across Coverage Levels and Mods • Decompose losses into layers by Mod – 100‐105%, 105‐110%, 110‐115%, 115‐120% – Cost of each layer should be less than prior layer
• Smooth and recombine • Applied to ND, SD, KS, CO, and ID – Will apply to MN, IA, NE in next review
Rating Formulas Excludes Company Loading for Expense & Profit
• Crop‐Hail: Liability x FALC x Policy Form Factor
• Companion Plan: CP Liability x FALC x Policy Form Factor x Increasing Payment Factor (2, 3, or 4)
• Production Plan PP Liability x FALC x Policy Form Factor APH Modifier – Coverage Level
14
9/17/2012
Implications for Crop‐Hail Data Reporting and Ratemaking
Big Problem • PP data can’t be used for CH ratemaking – NCIS restates each claim to a common level • Usually, Basic form • Example for claim under XS10 form – Add the 10% deductible to the paid loss
– But PP loss is smaller of hail and production loss • We don’t know what the hail loss is • Unable to restate it as if it were written on Basic form
– We’ve ignored the problem till recently • PP volume of business was small • But volume has grown rapidly – up to 50% of exposure
– Endangers ability to set CH FALCs in the future
15
9/17/2012
Solution – Starting with 2012 Cropyear • Require companies to report PP policy data in much greater detail – Report PP and CH statistics for each PP policy – Basic form liability – Actual Hail damages in $ • Prior to deductibles & other indemnification provisions
– Data must then pass NCIS edits
• The underlying CH data can be used in our usual FALC estimation procedure • May be able to use PP data in developing rating factors for PP
End of Presentation
16