Paper Schedule Reports 指導老師:戴天時 老師 楊曉文 老師 學生: 謝昌宏 Outline • What’s Guaranteed Minimum Withdrawal Benefit (GMWB)? • Pricing Method • See Example 2 What’s Guaranteed Minimum Withdrawal Benefit (GMWB)? 1.Roll-up(複利增值) 2.Ratchet(鎖高機制) 3. Break even(保本) Reference Source:中泰人壽 金富貴外幣變額年金保險 Pricing Method • The Bino-trinomial Tree 1. 延續Milevsky and Salisbury(2006)的 設計,假設GMWB所投資的標的資 產符合幾何布朗運動 2. 帳戶價值會隨著時間有預期報酬 的增加以外,還有公平費用率的收 取,若假設公平費用率是連續收取, 且保戶不能提前解約,則我們可以 將帳戶的隨機過程改為: dW (t ) (r )W (t )dt W (t )dB1 (t ) dW (t ) (r )dt dB1 (t ) W (t ) d ln W (t ) [(r ) 2 2 d ln W (t ) [(r (t ) ) ]dt dB1 (t ) 2 2 ]dt [ dB2 (t ) 1 2 dZ (t )] Reference Source : The2Bino-trinomial Tree: A Simple Model for Efficient and Accurate Option Pricing 4 Pricing Method • Optimized withdrawal rate GMWB Value [(( A * PA ) ( B * PB ) (C * PC ))* e rt ] (include Future Annuity Value) Vs. Full withdrawal Value (Wt G)*(1 k ) G Optimized withdrawal rule reference Kwork GUARANTEED MINIMUM WITHDRAWAL BENEFIT IN VARIABLE ANNUITIES Reference Source : The Bino-trinomial Tree: A Simple Model for Efficient and Accurate Option Pricing 5 Pricing Method • Death Rate - Reduction factors GAM-94(1994): y 1994 x x q q (1 AAx) ( y 1994 ) The value of AAx refer to “1994 GROUP ANNUITY MORTALITY TABLE AND 1994 GROUP ANNUITY RESERVING TABLE”. 6 Pricing Method • Wang Risk Transform Given a distribution function F, its Wang transform is defined as Risk-Neutral Real-world Death Rate Death Rate where F(x) is the distribution function corresponding to the standard Normal distribution and λ is a parameter called the market price of risk. 7 Pricing Method • SECURITIZATION OF LONGEVITY RISK: PRICING SURVIVOR BONDS WITH WANG TRANSFORM IN THE LEE-CARTER FRAMEWORK In Belgium, is the appropriate proxy for the market price of an annuity sold to an 65-year-old individual. i = 3.25%, and is the probability that a 65-year-old annuitant does not reach age 65 + t. They get λ65(2005) = −0.4722883 for men and −0.2966378 for women. 8 Pricing Method • Death Rate - Transform Death Rate l65 ( y n) (1 nq65 ( y n)) lz ( y ) l65 ( y z )e z ( t ) dt 65 , n z - 65, z 65 • 在此我們令65歲時仍生存的人數為基準,來算出各個年齡下的瞬時死 亡率,例如:在2005年為65歲,其未來一年內瞬時死亡率為: l66 (2006) l65 (2005) (1 1 q65 (2005)) l66 (2006) l65 (2005)e 66 ( t ) dt 65 l65 (2005)e 65 l65 (2005) (1 1 q65 (2005)) l65 (2005)e 65 65 ln( l65 (2005) (1 1 q65 (2005)) ) ln(1 1 q65 (2005)) l65 (2005) 9 Pricing Method • Death Rate - Transform Death Rate 其66歲時,未來一年內的瞬時死亡率為: l67 (2007) l65 (2005) (1 2 q65 (2005)) l67 (2007) l65 (2005)e 66 ( t ) dt 67 ( t ) dt 65 66 l65 (2005)e 65 66 l65 (2005) (1 2 q65 (2005)) l65 (2005)e 65 66 66 ln( l65 (2005) (1 2 q65 (2005)) ) 65 ln(1 2 q65 (2005)) 65 l65 (2005) x 1 l65 ( 65) (1 x 64 q65 ( 65)) x ln( ) k l65 ( 65) k 65 x : Age x 1 : Birth Year = ln(1 x 64 q65 ( 65)) k , 65 x : Maximum Age k 65 10 Pricing Method • Death Rate 將一年分成m期,每期時間長度為 65 t t t t 從65歲購買GMWB的那一刻往後經過2期 的時間,投保人 的生存機率為: l65 2 t ( 65 2 t ) l65 ( 65)e 65 2 t (t ) dt 65 l65 ( 65)e 65 2 11 t Pricing Method • When hit the boundary Option Value : (age : x, long of a period : T ) G G * e rT * e xT G * e r 2T * e x1T Discount factor Conditional probability of living Living Living G Death Death G 0 0 12 See Exapmle CRR • Find BTT Middle Point d ln W (t ) [(r ) 2 2 ]dt dB(t ) ln W (0) E[ln W (t ) ln W (0)] => B CRR steps is odd: (ln W (t ) d ln W (t )) l 1.38 2 t Index 0.5 1.5 Index Middle Po int l Index * 2 t 4.972 4.605 2*Cell Heigh 4.548 4.499 4.124 3.912 Boundary Reference Source : The Bino-trinomial Tree: A Simple Model for Efficient and Accurate Option Pricing 13 See Example • First year 5.18(178.54) 4.97(144.41) 2*CellHeigh 4.605(100) 4.55(94.48) 4.76(116.81) 4.34(76.42) 4.12(61.82) 3.91(50) CellHeigh Boundary 14 See Example • Withdrawal G 5.18(178.54) 4.97(144.41) 4.86(128.54) 4.605(100) 4.55(94.48) 4.76(116.81) 4.2(66.81) 4.34(76.42) 4.12(61.82) 3.27(26.42) 3.91(50) Boundary 0 15 See Example • Second year 5.18(178.54) 178.54 4.97(144.41) 4.86(128.54) 4.605(100) 4.76(116.81) 116.81 4.34(76.42) 76.42 4.55(94.48) 4.12(61.82) 4.2(66.81) 3.91(50) 50 Boundary 3.27(26.42) 0 32.71 21.4 14 16 See Example • Calculate final value 5.18(178.54) 178.54 4.97(144.41) 4.86(128.54) 4.605(100) 4.76(116.81) 116.81 4.34(76.42) 76.42 4.55(94.48) 4.12(61.82) 4.2(66.81) 3.91(50) 3.27(26.42) 50 Boundary 32.71 50 0 21.4 14 17 See Example • Forecast probability of death(Age>=65) 18 See Example • Forecast probability of death(Age>=65) y 1994 x x q q (1 AAx) ( y 1994 ) Ex: q65(2005)=q65(1994)x(1-AA65)(2005-1994)=0.019016 q66(2006)=q66(2004) x(1-AA66)(2006-1994)=0.0207688 19 See Example • Calculate risk-neutral nq65*(n=1,2,3,…) 1.calculate lx ( y ) (總生存率), x>=65 lx1 ( y 1) (1 qx1 ( y 1)) lx ( y) 2. 1q65 l65 (2005) l66 (2006) 0.017485 l65 (2005) l65 (2005) l67 (2007) 2q65 0.0356861 l65 (2005) ... 20 See Example • * nq Calculate risk-neutral 65 λ65(2005) = −0.4722883 for men 3. * 65 nq nq65 1q65* 0.004926 2q65* 0.0114413 ... 21 See Example • Calculate risk-neutral conditional death force x 1 x = ln(1 x 64 q65 ( 65)) k , 65 x k 65 65* = ln(1 1 q65 (2005)) 0.00494 66* = ln(1 2 q65 (2005)) 65 0.006569 ... Conditional Survival Probability: e 65 t e 66 t 0.997534131, t 0.5 0.996720697 22 See Example • Backward induction - CRR Pu 178.54 Survival value Value ((178.54* Pu ) (116.81* Pd ))* e r t * e 66 t Pd 116.81 ((178.54* Pu ) (116.81* Pd ))* e r t *(1 e 66 t ) Death value 76.42 Value ((50* Pu) (50* Pd ))* e r t * e 66 ((32.71* Pu ) (21.4* Pd ))* e r t *(1 e t 66 t 50 ) Boundary 32.71 50 21.4 14 23 See Example • Backward induction – first term (178.54) 130.67(144.41) Pu Survival value Value ((130.67 * Pu ) (85.49* Pm) (55.93* Pd )) *e r t * e 66 t ((144.41* Pu ) (94.48* Pm) (61.81* Pd )) * e r t *(1 e 66 t ) G Pm 85.49(94.48) Pd 55.93(61.81) Death value Vs. (we choice the higher) Full withdrawal value (1- k )*(178.54 - G) G 24 See Example • Backward induction – hit the boundary (178.54) 130.67(144.41) Pu Pm 85.49(94.48) Pd Value G (G * e r * e 66 ) 55.93(61.81) 25 Thanks for your attation 26