FAIR VALUATION OF INSURANCE LIABILITIESjanroman.dhis.org/finance/Deflators/Deflators PPT.pdf ·...
Transcript of FAIR VALUATION OF INSURANCE LIABILITIESjanroman.dhis.org/finance/Deflators/Deflators PPT.pdf ·...
FAIR VALUATION OF INSURANCE LIABILITIES
Pierre DEVOLDERUniversité Catholique de Louvain
03/ 09/2004
Pierre Devolder 09/2004 2
Fair value of insurance liabilities
1. INTRODUCTION TO FAIR VALUE2. RISK NEUTRAL PRICING AND DEFLATORS3. EXAMPLES : THE BINOMIAL and THE
BLACK/ SCHOLES CASES4. FAIR VALUE OF LIFE INSURANCE
PARTICIPATING CONTRACTS5. FAIR VALUE OF VARIABLE ANNUITIES6. CONCLUSION
Pierre Devolder 09/2004 3
1. Introduction to FAIR VALUE
International norms IAS / IFRS for all financial institutionsin Europe soon ( …)
…A lot of discussions linked to accounting principles and artificial volatility
General principle of Fair valuation of elements for assets as well as for liabilities: market values instead of historical values
Pierre Devolder 09/2004 4
1. Introduction to FAIR VALUE
Basic principle : from an historical or statutory accountingpoint of view to fair value bases
Fair value : price at which an instrument would be tradedif a liquid market existed for this instrument
ASSETS : market values
LIABILITIES : ???If no market value : principle of estimation of futurecash flows properly discounted and taking into accountthe different kinds of risk
Pierre Devolder 09/2004 5
1. Introduction to FAIR VALUE
Need to develop good models of valuation especially foractuarial liabilities where there is no market price
Consistency between modern financial pricing theory and classical actuarial models
Important example : guarantees in life insurance andpension:one of the most challenging risk nowaday
Even if for competition reasons methods of pricing could remain very classical , fair valuation will require new insightstaking into account modern finance
Pierre Devolder 09/2004 6
2. Risk neutral pricing and Deflators
Purpose :
- introduction to the modern financial paradigmof risk neutral pricing
- link with deflator methodology
- link with classical actuarial principle of discounting
- development in a simple discrete market model
Pierre Devolder 09/2004 7
2. Risk neutral pricing and Deflators
Paradigm of risk neutral pricing:
Purpose: to compute the present price of a futurestochastic cash flow correlated with financial marketin an uncertain environment
What we could expect : price = discounted expected value of the future cash flow:
))T(L(E)i1(
1)0(L T+= NO!!!
Pierre Devolder 09/2004 8
2. Risk neutral pricing and Deflators
You have to change either the probability measure, either the discounting factor
Change of probability measure : risk neutral method:You can stay with a classical discounting factor butthe expectation of the cash flows must be doneusing another probability measure than the real one
Change of discounting factor : deflators method:You can use the real probability measure but the discounting factor has to be changed and becomesstochastic
Pierre Devolder 09/2004 9
2. Risk neutral pricing and Deflators
Risk neutral pricing :
))T(L(E)r1(
1)0(L QT+=
Q= risk neutral measure
Deflator method :
))T(L)T(D(E)0(L =
D(T) =deflator = stochastic discounting
Pierre Devolder 09/2004 10
2. Risk neutral pricing and DeflatorsSingle period market model :
one riskless asset :
rateriskfreerwithrSS =+= )1()0()1( 00
d risky assets defined on a probability space :
NjPpwith jjN ,...,1)(,...,, 21 ===Ω ωωωω
diSSSS Niiii ,...,1)),1(),...,,1(),,1(()1( 21 == ωωω
classical assumption of absence of arbitrage opportunities
Pierre Devolder 09/2004 11
2. Risk neutral pricing and Deflators
STATE PRICE : Basic property :if the market is without arbitrage there existsa random variable ψ such that for any asset :
∑=
⟩Ψ=ΨΨ=N
jjjjiji withSS
10)(),1()0( ωω
If the market is complete, the state price is unique.
Consequence : for asset i =0 ( risk free asset)
∑= +
=Ψ=ΨN
1jj r1
1
Pierre Devolder 09/2004 12
2. Risk neutral pricing and Deflators
RISK NEUTRAL MEASURE
Definition : artificial probability measure given by:
jj
jj )r1()(Qq Ψ+=Ψ
Ψ=ω=ω=
Properties :
∑=
=≤≤N
1jjj 1qand1q0
2°) for each asset: mean return= risk free rate
1°)
∑=
ω+
=N
1jjiji ),1(Sq
)r1(1)0(S
Pierre Devolder 09/2004 13
2. Risk neutral pricing and Deflators
DEFLATOR : random variable D defined by :
j
jjj p
DDΨ
==)(ω
Properties :
∑= +
==N
1jjj r1
1)D(EDp.i
( expected value of the deflator = classical discount )
∑=
ω=N
1jjijji ),1(SDp)0(Sii.
Pierre Devolder 09/2004 14
2. Risk neutral pricing and Deflators
∑=
Ψ=N
1jj )j,1(X)0(X:visionpriceState.i
Generalization : if X is a financial instrument on this market (replicable by the underlying assets) and giving for scenario ja cash flow X(1,j) then the initial value of this instrument can be written as :
))1(X(Er1
1)j,1(Xqr1
1)0(X:visionneutralRisk.ii Q
N
1jj +
=+
= ∑=
))1(XD(E)j,1(XDp)0(X:visionDeflator.iiiN
1jjj ==∑
=
Pierre Devolder 09/2004 15
2. Risk neutral pricing and Deflators
Multiple periods model : discrete time model ( t=0,1,…, T)Riskfree asset:
rateriskfreerwithrStS t =+= )1()0()( 00
Risky assets :
ditStStStS Niiii ,...,1)),(),...,,(),,(()( 21 == ωωω
STATE PRICE :
∑=
⟩Ψ=ΨΨ=N
jjjjiji ttwithtStS
10),()(),()()0( ωω
Pierre Devolder 09/2004 16
2. Risk neutral pricing and DeflatorsDEFLATOR :
jscenarioiftotfromfactordiscountp
ttD
j
jj 0
)()( =
Ψ=
Pricing : if X is a financial replicable instrument on thismarket generating successive stochastic cash flows :
;,...,1);,( Ω∈= ωω TttC
Then the initial price of X can be written alternatively :
∑∑= =
Ψ=T
t
N
jjj ttCX
1 1)(),()0( ω
Pierre Devolder 09/2004 17
2. Risk neutral pricing and Deflators
Or with risk neutral measure:
∑ ∑= =
ω+
=T
1t
N
1jjjt ),t(Cq
)r1(1)0(X
Or with deflators :
∑∑∑== =
=ω=T
1tjj
T
1t
N
1jj ))t(C)t(D(E)t(D),t(Cp)0(X
Pierre Devolder 09/2004 18
3.1. THE BINOMIAL CASE
Single period model:
Risky asset :with probability puSS ⋅= )0()1( 11
with probability 1-pdS ⋅= )0(1
Absence of arbitrage opportunities if:urd <+<< 10
Other form of the risky asset :µλ +++= ru 1 µλ −++= rd 1
Pierre Devolder 09/2004 19
3.1. THE BINOMIAL CASE
µ0 <λ <With condition :
volatilitypremiumrisk == µλ
Equations of the STATE PRICE :
1)1()1( 21 =Ψ++Ψ+ rrFor i=0:
121 =Ψ+Ψ duFor i=1:
Pierre Devolder 09/2004 20
3.1. THE BINOMIAL CASE
Solution for the STATE PRICE:
)1(2))(1(1
1 rdurdr
+−=
−+−+=Ψ
µλµ
up
)1(2))(1()1(
2 rdurru
++=
−++−=Ψ
µλµdown
Safety principle :
012 =Ψ=Ψ λif
)(012 casenormalif >Ψ⟩Ψ λ
Pierre Devolder 09/2004 21
3.1. THE BINOMIAL CASE
Fair value in a binomial environment – single period :
If X is a financial instrument on this market with future stochastic cash flows given respectively by :
2211 ),1(),1( XXandXX == ωω
Then the initial fair value of X is given by :
2211)0( Ψ+Ψ= XXXOr :
)r1()XX(
21
)r1(1)XX(
21)0(X 1221 +µ
λ−+
++=
Pierre Devolder 09/2004 22
3.1. THE BINOMIAL CASE
Multiple periods model:
1
2 d
2 u
d
d u ⋅
u
Risky asset
Pierre Devolder 09/2004 23
3.1. THE BINOMIAL CASE
Structure of STATE PRICES in multiple periods:
Assumption: financial product having successive cash flows depending only on the current situation of the market (no pathdependant).
1
11 Ψ
21Ψ
22Ψ
23Ψ
12 Ψ
State price
jscenarioifttimeatpricestatejt =Ψ
Pierre Devolder 09/2004 24
3.1. THE BINOMIAL CASE
Value of the STATE PRICES:
12
11
1 −−−− ΨΨ=Ψ jtjjtjt C
Where j-1 = number of up (j=1,..,t+1)t-j-1 = number of down
1−jtCAnd: is the number of paths in the tree with
j-1 up in t periods
Pierre Devolder 09/2004 25
3.1. THE BINOMIAL CASE
Fair valuation in multiple periods – binomial :
If X is a financial instrument having successive cash flowsin the tree given by :
jscenarioifttimeatflowcashX jt =
Then the initial fair value of X is given by :
∑∑=
+
=Ψ=
T
t
t
jjtjtXX
1
1
1)0(
Pierre Devolder 09/2004 26
3.2. THE BLACK / SCHOLES CASEContinuous extension : Black and Scholes model
Riskless asset :
dt)t(Sr)t(dS 00 =
Risky asset :
)t(dw)t(Sdt)t(S)t(dS 111 σ+δ=
Where:- w is a standard brownian motion- δ is the mean return of the asset (δ>r)- σ is the volatility of this return
Pierre Devolder 09/2004 27
3.2. THE BLACK / SCHOLES CASERisk neutral measure :
)t(*dw)t(Sdt)t(Sr))t(dwdt)r()(t(Sdt)t(Sr)t(dS
11
111
σ+=σ+−δ+=
tr)t(w)t(*w:withσ−δ
+=
Q = risk neutral measure ( Girsanov Theorem) Under Q the process w*is a standard brownian motion
Pierre Devolder 09/2004 28
3.2. THE BLACK / SCHOLES CASE
Deflators: stochastic process D such that for i=1(risk free asset) and for i=2 ( risky asset):
))t(S)t(D(E)0(S ii =
Solution:
tr21)t(wr
rt
2
ee)t(D⎟⎠⎞
⎜⎝⎛σ−δ
−⎟⎠⎞
⎜⎝⎛σ−δ
−−=
Safety principle :Lower values of « w » give higher values for deflator
Pierre Devolder 09/2004 29
4. Fair value of participating contracts
4.1. Liability side: Life insurance contract with profit :guaranteed interest rate + participation
4.2. Asset side :Strategic asset allocation: Cash + Bonds + Stocks
4.3. Valuation of the contract :Fair value and equilibrium condition
Pierre Devolder 09/2004 30
4. Fair value of participating contracts
Need for a consistent ALM approach:Double link between asset and liability in this
kind of product : Liability Asset :
Investment strategy must take into accountthe specificities of the underlying liability
Liability :AssetParticipation liability linked with investmentresults
Pierre Devolder 09/2004 31
4.1. Liability sidePure Endowment contract :
- initial age at t=0 : x- maturity : N- Benefit if alive at time t=N : 1- Benefit in case of death before N : 0- Contract with single or periodical premiums
(pure premium)- Technical parameters :
-mortality table: - guaranteed technical rate : i- participation rate on surplus: η
l x
Pierre Devolder 09/2004 32
4.1. Liability sideBonus systems :
general definition: percentage of the surplus (Assets – Liabilities)
Three possible schemes :Terminal bonus : bonus computed at maturityReversionary bonus : bonus computed each
year and fully integrated in the contractas additional premium
Cash bonus : bonus computed each year and paid directly to the client
Pierre Devolder 09/2004 33
4.2. Asset sideCash – Bonds- Stocks model ( CBS model):The underlying portfolio of the insurer is supposed to be invested in three big classes of asset:
- Cash : short term position ( money account)- Bonds : zero coupon bonds with a maturity
not necessarily matched with the durationof the contract
- Stocks : stock index
Pierre Devolder 09/2004 34
4.2. Asset side Cash model :
Money market account :
ratefreerisk)t(rwithdt)t()t(r)t(d
=β=β
Risk free rate :Ornstein-Uhlenbeck process
motionbrowniandardtanswwith)t(dwdt))t(rb(a)t(dr
1
1r
=σ+−=
Pierre Devolder 09/2004 35
4.2. Asset sideBond model
P(t,T)= price at time t of a zero coupon with maturity T General evolution equation :
)t(dw)T,t()T,t(Pdt)T,t()T,t(P)T,t(dP 1σ−µ=Particular case : VASICEK Model
)e1(a1)s(B:with
)tT(B)T,t()tT(B)t(r)T,t(
as
r
r
−−=
−σ=σ−λσ+=µ
Pierre Devolder 09/2004 36
4.2. Asset sideStocks model :
S(t)= value at time t of a stock index
12
22
1S
woftindependanmotionbrowniandardtanswrateserestint/stocksncorrelatiowith
)))t(dw1)t(dw(dt()t(S)t(dS
==ρ
ρ−+ρσ+µ=
Pierre Devolder 09/2004 37
4.2. Asset sidePortfolio :
1stocksinproportionbondsinproportioncashinproportion
SBC
S
B
C
=α+α+α=α=α=α
2 main assumptions : - proportions remain constant
( continuous rebalancing)- self financed strategy
Pierre Devolder 09/2004 38
4.2. Asset sideBond strategy :
Assumption : at each time t zero coupon bonds of onlyone maturity can be held but the maturity has not necessarily to match the duration of the liability( price of mismatching-long duration of life insurance contract)
Strategy 1 : matched strategy : T=NStrategy 2 : shorter maturity of the bondand successive reinvestments till end of the contract
Pierre Devolder 09/2004 39
4.2. Asset sideBond strategy
tn Ns=0 t1 t2 t3
ntreinvestmeoftstanins:t,...,t,t,0tntsreinvestmeofnumber:n,...,1,0s
n210 ==
Particular case : matched strategy : n=0
Pierre Devolder 09/2004 40
4.2. Asset sideEvolution of the portfolio :
portfoliounderlyingtheofvaluemarket)t(V =Evolution equation :
)tttfor(
)t()t(d
)t,t(P)t,t(dP
)t(S)t(dS
)t(V)t(dV
i1i
Ci
iBS
<<
ββ
α+α+α=
−
Pierre Devolder 09/2004 41
4.2. Asset side
Value at maturity of the assets :
ddistributenormally))t(Vln)t(V(lntermeachwith
))t(Vln)t(V(ln)0(Vln)N(Vln
s1s
s
n
0s1s
=−
−=−
+
=+∑
))0(Vln)N(Vln(var)N())0(Vln)N(V(lnE)N(
2 −=σ
−=µ
Pierre Devolder 09/2004 42
4.2. Asset side
Explicit form :
∫
∫∫
∫
+
++
+
−ρσσαα−−σα+σα−
−σα−ρσα+ρ−σα+
−λσα+α+α+µα=−
++
+
++
1s
s
1s
s
1s
s
1s
s
t
t1srSBS1s
22B
2B
2S
2S
t
t11srBSS
t
t2
2SS
1srB
t
tCBSs1s
du))ut(B2)ut(B(21
)u(dw))ut(B()u(dw1
du))ut(B)u(r)(()t(Vln)t(Vln
Pierre Devolder 09/2004 43
4.3. Valuation of the contractExample of a contract with single premium andterminal bonus :
Fair value at maturity = pay off of the contract
)0;)i1()N(Vmax()i1()N(FV NN +−η++=
Initial fair value : (? In line with the single premium)
)))i1(;N;V(Call)N,0(P)i1((p)0(FV NNxN +η++=
Pierre Devolder 09/2004 44
4.3. Valuation of the contractEquilibrium condition :
1))i1(;N;V(call)N,0(P)i1( NN =+η++
Consequences :1°)
)))N,0(R1(
1)N,0(Pwith(
)N,0(Ri
N+=
<
2°) depends on the investment strategy throughthe value of V(N)
Pierre Devolder 09/2004 45
4.3. Valuation of the contract3°) implicit relation for the technical rate i ;4°) explicit relation for the participation rate η:
))i1(;N;V(call)N,0(P)i1(1
N
N
++−
=η
An explicit formula of the call can be obtained in the CBS model presented before.
Pierre Devolder 09/2004 46
4.3. Valuation of the contractRisk forward neutral measure method :
measureneutralriskforwardQwhere
)0);)i1()N(V(((maxE)N,0(Pcall
N
TQN
=
+−=
In the CBS model we have :
υυ±−+−
=
Φ+−Φ=
±
−+
2
N
2/1)T,0(Pln)i1(lnN)i(D
:with))i(D()N,0(P)i1())i(D(call
Pierre Devolder 09/2004 47
4.3. Valuation of the contractWith for instance for the matched strategy :
)e1(a2)e1(
a21
aN)N(B
)e1(a1
aN)N(B
:with)N(B)1(2)N(B)1(N
aN3
aN2322
aN21
1rSBS22
r2
B2
S2
S2
−−
−
−−−+=
−−=
ρσσα−α+σα−+σα=υ
Pierre Devolder 09/2004 48
5. FAIR VALUE OF VARIABLE ANNUITIES
Purposes :
How to valuate pension annuities not in terms of technical basisbut in terms of market fair values;
Influence of reversionary bonus ( variable annuities) on the level of provision;
Sensitivity of the provision with respect to financial parameters;
How to fix the technical interest rate.
Pierre Devolder 09/2004 49
5.1. Liability side
- Immediate lifetime annuity for an affiliate to a pension fund - x : initial age at time t=0- Liability to pay: 2 cases :
jscenarioforttimeatpaytoamountL j,t =
( possibility to increase yearly the pension dependingon the financial performances – asset side)
1 ) fixed annuity : L 2) variable annuity :
- Payment at the end of the year till death or during a fixed period of n years
Pierre Devolder 09/2004 50
5.1. Liability side
Actuarial first order bases :
ratediscounttechnicali=ttimeatyprobabilitsurvivalpxt =
Technical provision for a constant pension ( case 1):
LL jt =,
txt
n
tnxxn i
pLaLV)1(
11 +
== ∑=
Pierre Devolder 09/2004 51
5.2. Asset side
Binomial model :mixed financial strategy of the pension fundbetween riskless asset (r= riskfree rate)and risky asset (binomial model u / d)
γ : part invested in the risky asset : part invested in the riskless asset γ−1
)10( ≤≤ γ
Pierre Devolder 09/2004 52
5.3. Bonus scheme
Definition of the reversionary bonus for variable <annuities
Used rule of bonus : comparison each year between the effective return of the assets and the riskfree rate;a part of this surplus is given back to the affiliate:
: participation rate10 ≤≤ β
Pierre Devolder 09/2004 53
5.3. Bonus schemeYearly rate of increase of the pension :
If the risky asset is up :
+−+
+−++=+ )1)1(
)1)(1((11r
ruk γγβ
)1
(11r
k+++=+ µλβγor
If the risky asset is down :
1)1)1(
)1)(1((11 =−+
+−++=+ +
rrdl γγβ
Pierre Devolder 09/2004 54
5.3. Bonus scheme
Final form of the liabilities of the variable annuity:
( ) 1, 1 +−+⋅= jtjt kLL
Where t-j+1 is the number of timesof up permitting to give a bonus.
As expected THE LIABILITY DEPENDS ON TIME AND
IS STOCHASTIC
Pierre Devolder 09/2004 55
5.4. Valuation of the contract
Computation of the fair value of the liabilities :(fixed annuity)
( ) ⎥⎦
⎤⎢⎣
⎡Ψ= ∑∑
+
==tj
t
j
n
txtnx
LpLFV1
11,
⎥⎦
⎤⎢⎣
⎡Ψ= ∑∑
+
==tj
t
j
n
txt pL
1
11
tn
txt r
pL )1
1(1 +
= ∑=
nxraL=
Pierre Devolder 09/2004 56
5.4. Valuation of the contractComputation of the fair value of the liabilities
(variable annuity)
• Actuarial valuation : not so simple: liabilities not deterministic• Fair valuation : general formula of valuation :
( ) ⎥⎦
⎤⎢⎣
⎡Ψ= ∑∑
+
==tj
t
jjt
n
txtnxk LpLFV
1
1,
1,
( )( )[ ]tn
txt kpL +Ψ+Ψ= ∑
=112
1
( )( ) ⎥⎦
⎤⎢⎣
⎡+ΨΨ= ∑∑
+
=
+−−−
=
1
1
11
12
1
11
t
j
jtjjt
n
txt kCpL
Pierre Devolder 09/2004 57
5.4. Valuation of the contract
Computation of the fair value of the liabilities (variable annuity)
tn
t
t
xtnxk rrpLLFV ⎥
⎦
⎤⎢⎣
⎡+−
+⎥⎦⎤
⎢⎣⎡+
= ∑= )1(2
11
1)(22
1, µ
λµβγ
tn
txt i
pL )*1
1(1 +
= ∑=
nxiaL *=
Pierre Devolder 09/2004 58
5.4. Valuation of the contract
Equilibrium relation :
i* = equilibrium constant discount rate given by :
))1(2
1(/))1(2
(*2222
rrri
+−
++−
−=µ
λµβγµ
λµβγ
riorif ===→ *:00 γβ
riandif <>>→ *:00 γβ
Pierre Devolder 09/2004 59
5.5. Numerical illustration
Central scenario: u=1.1 d=0.99r=0.03
i=0.025Mortality: GRM 95
Risk premium : λ =0.02Volatility : µ = 0.06
n=5
Pierre Devolder 09/2004 60
5.5. Numerical illustration
0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1Fixed annuity
40%stocks
100%stocks
4000
4100
4200
4300
4400
4500
4600
4700
4800
4900
0% <12,25%stocks 20%stock 40%stocks 60%stocks 80%stocks 100%stocks
Pierre Devolder 09/2004 61
5.5. Numerical illustration
Volatility sensitivity analysis : ( 60% in risky asset)
4400
4500
4600
4700
4800
4900
5000
0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09 0,1
"Volatility factor"
Actuarial valuat ion Deflato r valuat ion B= 0 ,15B= 0 ,3 B= 0 ,5 B= 0 ,65B= 0 ,8 B=1
5.5. Numerical illustration
Value of the equilibrium discount rate : central scenario
1° for β = 0.5 and γ =0.6 :i*= 2.21%
2° for β = 1 and γ =0.6 :i*= 1.42%
3° for β = 0.9 and γ =0.4 :i*= 2.05%
4° for β = 1 and γ =1 :i*= 0.4%
%6%2
==
µλ
r=3%i=2.5%
Pierre Devolder 09/2004 62
5.5. Numerical illustration
Value of the equilibrium discount rate : other scenario( less volatile)
1° for β = 0.5 and γ =0.6 :i*= 2.60% versus 2.21%
2° for β = 1 and γ =0.6 :i*= 2.21% versus 1.42%
3° for β = 0.9 and γ =0.4 :i*= 2.52% versus 2.05%
4° for β = 1 and γ =1 :i*= 1.68% versus 0.4%
%3%1
==
µλ
r=3%i=2.5%
Pierre Devolder 09/2004 63
Pierre Devolder 09/2004 64
5.5. Numerical illustrationEquilibrium technical rate
0,1
0,3
0,5
0,7
0,9
100%stocks
20%stocks0
0,005
0,01
0,015
0,02
0,025
0,03
Participation rate
100%stocks 80%stocks 60%stocks 40%stocks 20%stocks 0%stocks
Pierre Devolder 09/2004 65
5.5. Numerical illustration
0
0,005
0,01
0,015
0,02
0,025
0,03
0,035
0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09 0,1
"Volatility factor"
B=1 B=0.8 B=0.65 B=0.5
B=0.3 B=0.15 B=0
EQUILIBRIUM TECHNICAL RATE(II)
Pierre Devolder 09/2004 66
6.CONCLUSION
State prices , risk neutral measures and deflators are easy tools in order to valuate stochastic future cash flows correlated to future financial markets.
This is exactly the situation of life insurance products
One of the most important result is the way to valuate bonus and to define properly a technical interest rate in a complete ALM framework