Beruflich Dokumente
Kultur Dokumente
Distribution Method 1 Method 2
TVaRæ (𝑋𝑋) Poisson 𝑥𝑥̅ = 𝑠𝑠 M 0
𝜙𝜙¬𝑧𝑧æ √ Binomial M
𝑥𝑥̅ > 𝑠𝑠 Negative
Normal 𝜇𝜇 + 𝜎𝜎 ¿ ƒ
1 − 𝑝𝑝 Neg. Binomial 𝑥𝑥̅ < 𝑠𝑠 M Positive
Φ¬𝜎𝜎 − 𝑧𝑧æ √
∑g¥∂I ln 𝑥𝑥¥ 𝐷𝐷r = max¬Ï𝐹𝐹g ¬𝑥𝑥r √ − 𝐹𝐹 ∗¬𝑥𝑥r √Ï, Ï𝐹𝐹g ¬𝑥𝑥r.I√ − 𝐹𝐹∗ ¬𝑥𝑥r √Ï√ 𝑟𝑟
SBC/BIC 𝑙𝑙 − ln 𝑛𝑛
𝜇𝜇̂ = If data is truncated at 𝑑𝑑, then 2
𝑛𝑛
Lognormal
𝐹𝐹(𝑥𝑥) − 𝐹𝐹(𝑑𝑑)
g
∑¥∂I(ln 𝑥𝑥¥ ) M AIC 𝑙𝑙 − 𝑟𝑟
𝜎𝜎÷ M = − 𝜇𝜇̂ M 𝐹𝐹 ∗(𝑥𝑥) = , for 𝑥𝑥 ≥ 𝑑𝑑
𝑛𝑛 1 − 𝐹𝐹(𝑑𝑑)
where
Poisson 𝜆𝜆◊ = 𝑥𝑥̅ Kolmogorov-Smirnov Test Properties 𝑙𝑙: log-likelihood
𝑥𝑥̅ • Individual data only 𝑟𝑟: # of estimated parameters
Binomial,
𝑞𝑞÷ = • Continuous fit only 𝑛𝑛: sample size
fixed 𝑚𝑚 𝑚𝑚
• Lower critical value for censored data Select model with the highest SBC or AIC value.
Neg. Binomial, 𝑥𝑥̅ • If parameters are estimated, critical value
𝛽𝛽◊ =
fixed 𝑟𝑟 𝑟𝑟 should be adjusted
𝑧𝑧(I•æ)⁄M M • Bühlmann estimates are on a straight line Empirical Bayes Non-Parametric Methods
𝑛𝑛ı = ´ ≠ (𝐶𝐶𝑉𝑉 M) • Bayesian estimates are within the range of Uniform Exposures
𝑘𝑘
M
Credibility premium: 𝑃𝑃˙ = 𝑍𝑍𝑥𝑥̅ + (1 − 𝑍𝑍)𝑀𝑀
g
∑!¥∂I ∑r∂I"
𝑚𝑚¥r ¬𝑥𝑥¥r − 𝑥𝑥̅ ¥ √
= 𝑀𝑀 + 𝑍𝑍(𝑥𝑥̅ − 𝑀𝑀) Posterior • 𝛼𝛼 ∗ = 𝛼𝛼 + ∑g¥∂I 𝑥𝑥¥ 𝑣𝑣÷ =
∑!¥∂I(𝑛𝑛¥ − 1)
where I .I
• 𝜃𝜃 ∗ = Ó + 𝑛𝑛Ô
2. K
𝐿𝐿 LMN.
= 𝐿𝐿 ⋅ 𝑓𝑓ULMN. Calculates average factors to be applied to the aggregate historical premiums
2. 𝐿𝐿KLMN.
U =U
𝐿𝐿 U,Z ⋅ U,ZLMN.
𝑓𝑓 U Calculates average factors to be applied to the aggregate historical premiums
3. 𝑅𝑅 K= KLMN. − S to make them on-level
LMN.𝐿𝐿 S 𝐿𝐿
Indicated Base Rate = Current Base Rate 1+ ⋅ Indicated Avg. Rate Change
𝑅𝑅 = 𝑤𝑤 ⋅ 𝑅𝑅fg +fg(1 − 𝑤𝑤) ⋅ 𝑅𝑅gi where 𝑤𝑤 = LMN.𝑓𝑓 Indicated Base Rate = Current Base Rate ⋅ Off-Balance Factor
𝑓𝑓
Indicated Avg. RelativityOff-Balance Factor
Frequency-Severity Method
o kmpqr. so
kULMN. − 𝑁𝑁U,Ws\z Adj. 𝐿𝐿† =
Adj. 𝐿𝐿†U = U Avg. RelativityU ⋅ Exposure
m,ntu
2. 𝑛𝑛wU,W = 𝑐𝑐̂kWLMN. y𝑁𝑁
2. 𝑛𝑛wU,W = 𝑐𝑐̂W y𝑁𝑁U − 𝑁𝑁U,Ws\z
Avg. RelativityU ⋅ ExposureU U
Aggregate
Adj. 𝐿𝐿†U
Aggregate Indicated Relativity =Adj. 𝐿𝐿†U
|1. 𝑙𝑙|U,W = 𝑛𝑛wU,W ⋅ 𝑥𝑥wU,W Indicated RelativityU = U Adj. 𝐿𝐿†òôöõ
1. 𝑙𝑙U,W = 𝑛𝑛wU,W ⋅ 𝑥𝑥wU,W
2. 𝑅𝑅 = ∑U[WÄ 𝑙𝑙|U,W , where 𝑦𝑦 is the valuation CY
Adj. 𝐿𝐿†òôöõ
2. 𝑅𝑅 = ∑U[WÄ 𝑙𝑙|U,W , where 𝑦𝑦 is the valuation CY
Indicated Avg. Rate
Indicated Base Rate =Indicated Avg. Rate
Data Preparation Indicated Base Rate = Indicated Avg. Relativity
Data Preparation Indicated Avg. Relativity
Losses
Losses
Credibility-Weighted Relativities
Credibility-Weighted Relativities
Losses New Relativity = 𝑍𝑍(Indicated Relativity) + (1 − 𝑍𝑍)(Current Relativity)
New Relativity
= 𝑍𝑍(Indicated Relativity) + (1 − 𝑍𝑍)(Current Relativity)
Other Topics
Other Topics
Increased Limit Factor
Increased Limit Factor
𝐿𝐿𝐿𝐿𝐿𝐿(𝑢𝑢) + 𝑅𝑅𝐿𝐿ß
Aggregation Develop to Trending
𝐼𝐼𝐼𝐼𝐼𝐼𝐿𝐿𝐿𝐿𝐿𝐿(𝑢𝑢)
= + 𝑅𝑅𝐿𝐿ß
• Calendar Year (CY) Ultimate • Trend Period 𝐼𝐼𝐼𝐼𝐼𝐼 =
𝐿𝐿𝐿𝐿𝐿𝐿(𝑏𝑏) + 𝑅𝑅𝐿𝐿
𝐿𝐿𝐿𝐿𝐿𝐿(𝑏𝑏) + 𝑅𝑅𝐿𝐿® ®
• Accident Year (AY) • Loss Development • Trend Factor • 𝑏𝑏: original limit
• 𝑏𝑏: original limit
• Policy Year (PY) Factors • 𝑢𝑢: increased limit
• 𝑢𝑢: increased limit
Rate of policy variation with limit 𝑢𝑢 = 𝐼𝐼𝐼𝐼𝐹𝐹ß ⋅ Indicated Base Rate
Rate of policy variation with limit 𝑢𝑢
= 𝐼𝐼𝐼𝐼𝐹𝐹ß ⋅ Indicated Base Rate
Loss Elimination Ratio
Loss Elimination Ratio
𝐿𝐿𝐿𝐿𝐿𝐿(𝑑𝑑) − 𝐿𝐿𝐿𝐿𝐿𝐿(𝑏𝑏)
Projected Losses
𝐿𝐿𝐿𝐿𝑅𝑅 𝐿𝐿𝐿𝐿𝐿𝐿(𝑑𝑑)
= − 𝐿𝐿𝐿𝐿𝐿𝐿(𝑏𝑏)
𝐿𝐿𝐿𝐿𝑅𝑅© =© 𝑥𝑥̅ − 𝐿𝐿𝐿𝐿𝐿𝐿(𝑏𝑏)
𝑥𝑥̅ − 𝐿𝐿𝐿𝐿𝐿𝐿(𝑏𝑏)
Incurred losses for CY 𝑖𝑖: 𝐿𝐿É = 𝐿𝐿S + 𝑅𝑅 − 𝑅𝑅 • 𝑏𝑏: original deductible
Incurred losses for CY 𝑖𝑖: 𝐿𝐿ÉU = U𝐿𝐿SU +U 𝑅𝑅U − U𝑅𝑅Us\ Us\ • 𝑏𝑏: original deductible
where 𝑅𝑅U is the reserves at the end of CY 𝑖𝑖 • 𝑢𝑢: increased deductible
where 𝑅𝑅U is the reserves at the end of CY 𝑖𝑖 • 𝑢𝑢: increased deductible
Rate of policy variation with deductible 𝑑𝑑 = (1 − 𝐿𝐿𝐿𝐿𝑅𝑅 ) ⋅ Indicated Base Rate
Incurred losses for AY or PY 𝑖𝑖: 𝐿𝐿É = 𝐿𝐿S + 𝑅𝑅 Rate of policy variation with deductible 𝑑𝑑 = (1 − 𝐿𝐿𝐿𝐿𝑅𝑅© ) ⋅©Indicated Base Rate
Incurred losses for AY or PY 𝑖𝑖: 𝐿𝐿ÉU = U𝐿𝐿SU +U 𝑅𝑅U U
where 𝑅𝑅U is the reserves as of the valuation date
where 𝑅𝑅U is the reserves as of the valuation date