Health Insurance Lecture s notes (Syllabus) by Aldona Skučaitė


 David Cook
 1 years ago
 Views:
Transcription
1 Health Insurance Lecture s notes (Syllabus) by Aldona Skučaitė
2 2 Course content 1. Peculiarities of Health insurance markets (see [1], [4], [8], [9], p.p , [1]) (a) Preferences and choice of individuals under risk: Utility theory; Allais parado; Prospect theory (b) Comparison of Health care market and "ideal" market (see ([1])) (c) Asymmetric information and its consequences  moral hazard and adverse selection  in insurance markets (see ([8]), ([1])) 2. Health Care Financing (see [2], [7] ) (a) "Ideal" health care system: main goals, tasks and measures (b) Two main systems for financing of health care: public  mandatory and private  voluntary (c) Health care systems in European Union and USA 3. Actuarial Models for Health Insurance (see [5], p.p ; 8142; ) (a) Multiple state model. Markov chain. Semi Markov model (b) Medical epenses insurance (c) Disability insurance (d) Critical illness insurance (e) Long term care insurance
3 Chapter 1 Health Insurance Markets 1.1 Preferences and choice of individuals under risk Utility Theory Probably, the first time suggested by D. Bernoulli, developed by J. von Neumann and O. Morgenstein, see Theory of Games and Economic Behaviour, 1944). Formal hypothesis and applications may be found, for eample, in [4]. Choices of individuals in riskless situation as well as under risk are eplained here. Let U()  utility function of individual,  wealth (income, assets, etc.), U is supposed to be non decreasing. Suppose that individual must choose between two alternatives  risky choice A, where: { I1, with probability < α < 1; (A) = I 2, with probability 1 α. Corresponding utilities are: for riskless alternative B  U(I ) for risky alternative (A)  Û(A) = αu(i 1) + (1 α)u(i 2 ) According to Formal hypothesis of Utility theory individual s choice will be: { A, if Û(A) > U(I ); B, if U(I ) > Û(A). 3
4 4 CHAPTER 1. HEALTH INSURANCE MARKETS Let Ī(A) = αi 1 +(1 α)i 2. If I = Ī(A), premium for insurance (gamble) is called "fair" ("actuarially fair"). If in this case individual chooses (A), his behavior is said to be "risk seeking", otherwise if individual chooses (B) his behavior is "risk avoiding". Let I :  state of wealth such that: U(I ) = Û(A). From inequality: we get: Û(A) > U(Ī) I > Ī and vice versa. The difference I Ī  is the maimum amount that individual will pay for insurance in ecess of actuarially fair premium. Utility function of risk averse individual is concave upwards, while of risk seeking individual  concave downwards. Utility theory eplains both risk aversion and risk seeking behavior, mainly: Risk averse individuals will buy insurance but will not gamble Risk seeking individuals will gamble but will not buy insurance Risk averse individuals agree to pay insurance premium which is greater than actuarially fair premium Since many individuals buy insurance and gamble, most probably utility function is at first concave upwards and concave downwards later Criticism of Utility theory. Allais parado (1953) Utility theory is useful for eplanation of both risk avoiding and risk seeking behavior. In many situations it eplains market behavior quite well. However, in some situations choices of individuals contradict Utility theory. Consider socalled Allais parado (see, for eample, [9], [6]) Suppose individuals are asked to select either investment strategy F 1 or F 2. Under each strategy possible wins (profits) are , 1 millions or 3 millions. Corresponding probabilities are shown in table below: $ 1 mil $ 3 mil $ F 1 1 F
5 1.2. COMPARISON OF HEALTH CARE MARKET AND "IDEAL" MARKET5 Eperiments show that majority of individuals chooses F 1, that is F 1 F 2 (F 1 dominates (is better than F 2 ). Now, again suppose that individuals to asked to select either investment strategy F 3 or F 4. Again, possible wins (profits) are , 1 millions or 3 millions and corresponding probabilities are as follows: $ 1 mil $ 3 mil $ F F In this situation majority of individuals chooses F 3, that is F 3 F 4. It is easy to see that such choices contradict Utility theory, because since F 1 F 2 : U(1), 1U() +, 89U(1) +, 1U(3), (1.1.1) while from the F 3 F 4 we get:, 9U() +, 1U(3), 89U() +, 11U(1) (1.1.2) and it is easy to see that inequalities (1.1.1) and (1.1.2) are opposite. There are more similar eperiments which results contradicts Utility theory (see, for eample, [6]). Some methods for improvement of Utility theory are proposed (see [9]), however we will not study these theories in this course. 1.2 Comparison of Health care market and "ideal" market Some concepts from Microeconomy: Ideal market Market imperfections Information asymmetry: Adverse selection Moral hazard
6 6 CHAPTER 1. HEALTH INSURANCE MARKETS Comparison of Health care industry market and ideal market are presented in [1]. In this paper many features of health care industry most of which contradict theory of ideal market are described, for eample: Nature of demand and supply Epected behavior of physician as provider of services Product uncertainty Pricing process Moral hazard If we use Theory of Ideal insurance, then, under assumption that individuals are risk averse and average cost of medical care is m, individuals will prefer insurance for a price m. Actually due to risk avoidance individuals will agree to pay more than m for insurance provided than insurance premium is not "too unfair". Actually due to administrative costs insurance premium must be higher than actuarially fair premium. In such case optimal policy for insured is no longer full insurance. Two main theorems about optimal policy are proved in the paper Theorem. If an insurance company is willing to offer any insurance policy against loss desired by the buyer at a premium which depends only on the policy s actuarial value, then the policy chosen by a risk averting buyer will take the form of 1 percent coverage above the deductible minimum. Proof. Let W  initial wealth, X  possible random loss, I(X)  insurance benefit paid if loss X occurs, P  insurance premium, Y (X)  wealth of individual after loss occurred and insurance benefit was paid. So Y (X) = W P X + I(X). (1.2.1) Let utility function be U(y), where y  wealth. Then individual seek to maimize: E[U(Y (X))]. (1.2.2)
7 1.2. COMPARISON OF HEALTH CARE MARKET AND "IDEAL" MARKET7 Insurance benefit must be nonnegative, so: I(X), X. (1.2.3) If policy is optimal then it must be the best from the set of all policies with the same actuarial value E[I(X)] in the sense of Consider two insurance policies. Initial policy is I(X) with I 1 (X) > and Y 1 (X 1 ) > Y 1 (X 2 ). Suppose δ > is sufficiently small that: and I 1 (X) >, if X 1 X X 1 + δ (1.2.4) Y 1 (X ) < Y 1 (X), if X 2 X X 2 + δ, X 1 X X 1 + δ (1.2.5) Define π 1  probability that X lies in (X 1, X 1 + δ); and π 2  probability that loss X lies in (X 2, X 2 + δ). Then we may choose sufficiently small ε >, such that from (1.2.4) and (1.2.5) we get: and I 1 (X) π 2 ε, if X 1 X X 1 + δ (1.2.6) Y 1 (X ) + π 1 ε < Y 1 (X) π 2 ε, if (1.2.7) X 1 X X 1 + δ, and X 2 X X 2 + δ Define new insurance policy I 2 (X), which is: I 1 (X) π 2 ε, [X 1, X 1 + δ]; I 2 () = I 1 (X) + π 1 ε, [X 2, X 2 + δ]; I 1 (X), elsewhere. From (1.2.6) we get that I 2 (X), so (1.2.3) is satisfied. Suppose that f()  is probability density function of X. Then:
8 8 CHAPTER 1. HEALTH INSURANCE MARKETS E[I 2 (X) I 1 (X)] = X1 +δ X 1 [I 2 () I 1 ()]f()d + = ( π 2 ε) X1 +δ X 1 = π 2 επ 1 + π 1 επ 2 =, f()d + π 1 ε X2 +δ X 2 X2 +δ X 2 [I 2 () I 1 ()]f()d f()d so actuarial values of both policies are the same. It is obvious that Y 2 (X) Y 1 (X) = I 2 (X) I 1 (X). From (1.2.7) we get: Y 1 (X ) < Y 2 (X ) < Y 2 (X) < Y 1 (X) if (1.2.8) X 2 X X 2 + δ X 1 X X 1 + δ, because: Y 2 (X ) = Y 1 (X ) + I 2 (X ) I 1 (X ) = Y 1 (X ) + π 1 ε > Y 1 (X ); Y 2 (X) > Y 2 (X ) ; Y 2 (X) = Y 1 (X) + I 2 (X) I 1 (X) = Y 1 (X) π 2 ε < Y 1 (X) Since Y 1 (X) Y 2 (X) = everywhere ecept [X 1, X 1 + δ] and [X 2, X 2 + δ] we have: and E[U(Y 2 (X)) U(Y 1 (X))] = (1.2.9) X1 +δ X 1 From the Mean value theorem we get: [U(Y 2 ()) U(Y 1 ())]f()d + X 2 +δ X 2 [U(Y 2 ()) U(Y 1 ())]f()d. U(Y 2 (X)) U(Y 1 (X)) = U (Y (X))(Y 2 (X) Y 1 (X)) (1.2.1) where Y (X) is between Y 1 (X) and Y 2 (X). So from (1.2.8) we get: = U (Y (X))(I 2 (X) I 1 (X)), Y (X ) < Y (X), if X 2 X X 2 + δ, X 1 X X 1 + δ.
9 1.2. COMPARISON OF HEALTH CARE MARKET AND "IDEAL" MARKET9 Since individuals are risk averse U (y) is decreasing function of y, so U (Y (X )) > U (Y (X)). We may find u such that U (Y (X )) > u, if X 2 X X 2 + δ (1.2.11) U (Y (X)) < u, if X 1 X X 1 + δ. Putting (1.2.1) into (1.2.9), we get: E[U(Y 2 (X)) U(Y 1 (X))] = X1 +δ π 2 ε +π 1 ε X 1 X2 +δ X 2 U (Y ())f()d U (Y ())f()d. And finally from (1.2.11) we obtain: E[U(Y 2 (X)) U(Y 1 (X))] > π 2 εuπ 1 + π 1 εuπ 2 =, so second policy is better than initial policy and risk averse individual will choose the second one. So, the policy I 1 (X) cannot be optimal if one can find two values of X  X 1 and X 2 for which: I(X 1 ) > and Y (X 1 ) > Y (X 2 ). So, suppose that Y min  minimal value of wealth (Y (X)) under optimal policy. Then according to the theorem I(X) = if Y (X) > Y min. So, insurance benefit is not paid until wealth of individual does not reach Y min. This is insurance policy with 1 percent coverage above stated deductible Remark. Alternative proof may be found in [9] Theorem. If insurance company and insured person are both risk averters and there no other costs ecept of coverage of losses then any non trivial Pareto optimal policy I(X) has the property < di < 1. Such type dx of policies is called coinsurance policies.
10 1 CHAPTER 1. HEALTH INSURANCE MARKETS The proof may be found in [1]. Problem of moral hazard is addressed in more detail in [8]. In this paper moral hazard is treated more like rational economical behavior rather than "moral" problem. Some suggestions how to deal with consequences of moral hazard are provided (deductibles and coinsurance). 1.3 Asymmetric information in insurance markets Probably best known work describing the impact of asymmetric information to insurance markets is [1] by M. Rothschild and J. Stiglitz (1976). It is said that insurance policy belong to equilibrium set if under assumption that individuals buy maimum one insurance policy (which must maimize their epected utility): no insurance contract in the equilibrium make negative epected profits there is no contract outside the equilibrium which if offered will make a non negative profit. Authors show that if two groups of individuals eist in the market  the low risk group and the high risk group  then there may be no equilibrium in the market. Moreover they show that eistence of high risk individuals eerts a negative eternality on low risk individuals  there are losses for low risk individuals if high risk individuals are present. However, high risk individuals are not be in a better position than they would be in isolation. For complete proof see [1]. 1.4 Sample eercises 1 Eercise. Suppose that risk averse individual whose utility function is u() is considering whether to insure against random risk or not. Individual may choose not to insure at all; to buy full insurance or to buy partial insurance. Individual s choice must maimize his / her epected utility. { Suppose that initial wealth of individual is w. Random loss is d < w, with probability p; ξ =, 1p. Show that:
11 1.4. SAMPLE EXERCISES If insurance premium is actuarially fair individual will choose full insurance. 2. If insurance premium is no longer actuarially fair and equal (1+ρ)Eξ), ρ >, then individual will not choose full insurance. Interpret (comment) these situations. 2 Eercise. Individual whose initial wealth is w = 2 and utility function is u() = 4 2 ; 2 faces random loss: ξ U[; 2]. Suppose that insurance premium is actuarially fair but individual decides to insure only against 8% of loss. Calculate his / her epected utility if: 1. Benefit paid is always equal to 8% of loss (proportional insurance) {, ξ < d; 2. Deductible d is set so that insurance benefit is : ξ d, ξ d.. It is insurance with deductible. Deductible d must be such that epected benefit I is 8% of possible loss, that is E[I] =.8E[ξ] = Maimum coverage s is set such that: to be E[I] =.8E[ξ] = 8. { ξ, ξ < s; s, ξ s. Again calculate s Which option is best for individual? Interpret results. 3 Eercise. Suppose that risk { averse individual with initial wealth w = 2 2, p =, 25; may suffer random loss ξ =, 1 p =, 75. Policies of proportional insurance are traded in the market. Under insurance conditions k ( k 1) percent of loss is compensated while insurance premium is calculated: G = (1 + ρ) k E[ξ]. Find optimal (best) insurance policy and ρ =, 2 and utility function of individual is: 1. u() = u() = ln( + 1) In both cases calculate lowest level of ρ for which optimal decision of individual will be not to insure (to accept loss himself). Compare results. Interpret them using concept of risk aversion.
12 12 CHAPTER 1. HEALTH INSURANCE MARKETS 4 Eercise. Suppose that: Individual with utility{ function u() = ln and 2,.25; initial wealth w = 21 may suffer random loss ξ =,.75. Insurance policies (α 1 ; α 2 ) are traded in the market, where α 1  insurance premium; α 2  insurance benefit minus insurance premium (see model in [1]. Suppose that due to perfect competition profits of insurers is zero. In the system of coordinates (W 1 ; W 2 ) (W 1  wealth in the case of no loss; W 2  wealth after loss) show the point of initial wealth status (without any insurance) and break even line. a) Suppose that all individuals are identical (their utility functions and accident probabilities are the same). Write down the equation of individual utility indifference curve (the one that maimizes individual s utility). Using MsEcel in the system of coordinates (W 1 ; W 2 ) graph break even line, "fair odds" line and utility curve of individual. Eplain choice of individual in this situation. b) Suppose that probability,25 is mean value of accident probabilities in the market. Actually, half (5%) of individuals belong to high risk group (p H =.35), while the other  to low risk group (p L =.15). Insurance companies know the proportion of high and low risk individuals in the market, but are unable to distinguish to which group individual belong. In this situation show the set of contracts which belong to possible separating equilibrium (You are not supposed to eplore whether or not equilibrium eist) and eplain Your decision.
13 Chapter 2 Health Care Financing 2.1 "Ideal" health care system Ehaustive eplanation of features of Ideal health care system is given in [2]. According to World Health Organization health is  "a state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity". Main goals of any health care system are  Cost, Quality and Access. Main problems of health care financing, tasks of any health care system and comparison of two main ways of financing  public mandatory and private voluntary  are presented in the paper. 2.2 Health care systems in European Union and USA Main issues: Financing system: public, private or mi of both; share of public and private ependitures in total health ependitures Role of Voluntary health insurance Achievement of Cost, Quality and Access goals See [2], [7] and / or any other reliable source that you will be able to achieve. 13
14 14 CHAPTER 2. HEALTH CARE FINANCING 2.3 Sample eercises 5 Eercise. Select any European country and make short presentation (51 min.) about its Health care system.
15 Chapter 3 Actuarial Models for Health Insurance The content of this chapter is based on material from [5]. 3.1 Multiple state models and Markov process The basis of majority actuarial models used for health insurance is Multiple state models. Let I  set of possible states (we will suppose that it is finite set), I = {1, 2,..., N} while J  set of direct transitions, J {(i, j) i j; i, j I}. Suppose that all states are reachable from initial state via direct or indirect transitions. The pair (I, J)  is called Multiple state model. Define S(t)  state of a system at time t and let S() = 1. We will call the process {S(t), t }  time continuous stochastic process any realization of this process  s(t) is called trajectory of the process. Stochastic process {S(t), t } is called time continuous Markov chain if n and any finite sets t < t 1 < < t n < u and i, i 1,..., i n, j, satisfying condition: P r(s(t ) = i,..., S(t n 1 ) = i n 1, S(t n ) = i n, S(u) = j) > (3.1.1) the following relation hold: P r(s(u) = j S(t ) = i, S(t 1 ) = i 1,..., S(t n ) = i n ) = P r(s(u) = j S(t n ) = i n ). (3.1.2) 15
16 16 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE (3.1.2) is called Markov property. Obvious that for any w > u > t n : P (S(u) = j, S(w) = k S(t ) = i,..., S(t n 1 ) = i n 1, S(t n ) = i n ) = P (S(u) = j S(t ) = i,..., S(t n 1 ) = i n 1, S(t n ) = i n ) P (S(w) = k S(t ) = i,..., S(t n 1 ) = i n 1, S(t n ) = i n, S(u) = j) = P (S(u) = j S(t n ) = i n ) P (S(w) = k S(u) = j). In more general case when S(z) = s(z), z t Markov property is defined: τ, u : τ < u and integer numbers i, j, s(z) ( z < τ) such that: P ([S(z) = s(z), z < τ], [S(τ) = i], [S(u) = j]) > the following must hold: P (S(u) = j [S(z) = s(z), z < τ], [S(τ) = i]) (3.1.3) = P (S(u) = j S(τ) = i). (3.1.4) Probabilites (3.1.2), e.g. P r(s(u) = j S(t) = i), t < u and i, j I are called transition probabilities and defined P ij (t, u) = P r(s(u) = j S(t) = i). Obvious: P ij (t, t) = δ ij, t, where δ ij = { if i j; 1 if i = j. If pairs t, u, t < u and i, i, j I probabilities P ij (t, u) depend only on the difference u t, the process is called homogeneous. In most actuarial applications processes are not homogeneous. Obviuos that: P ij (t, u) 1 i, j t u; (3.1.5) P ij (t, u) = 1 i t u. j I Chapman  Kolmogorov equations hold: P ij (t, u) = P ik (t, w)p kj (w, u), (3.1.6) k I
17 3.1. MULTIPLE STATE MODELS AND MARKOV PROCESS 17 where t w u. (3.1.6) is derived using Markov property: P ij (t, u) = P r(s(u) = j S(t) = i) = P r(s(u) = j, S(w) = k S(t) = i) k I = P r(s(w) = k S(t) = i) P r(s(u) = j S(t) = i, S(w) = k) k I = P r(s(w) = k S(t) = i) P r(s(u) = j S(w) = k) k I = P ik (t, w)p kj (w, u). k I Define occupancy probabilities: P ii (t, u) = P r(s(z) = i, z [t; u] S(t) = i). (3.1.7) Obvious that: t w u. P ii (t, u) = P ii (t, w) P ii (w, u), (3.1.8) Classification of states: Absorbing state i  P ii (t, u) = 1, t u; Transient state i  P ii (t, ) =, t ; Strictly transient state i  P ii (t, u) = P ii (t, u) < 1, t u. Define intensity of transition µ ij (t): µ ij (t) = lim u t P ij (t, u) u t. (3.1.9) For homogeneous Markov process: µ ij (t) = lim u t P ij (t, u) u t = µ ij.
18 18 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE We will define: µ i (t) = j:j i µ ij (t). (3.1.1) Then: µ i (t) = j:j i lim u t P ij (t, u) u t 1 P ii (t, u) = lim. u t u t = lim u t j:j i P ij(t, u) u t Kolmogorov forward differential equations: d dt P ij(z, t) = k:k j P ik (z, t)µ kj (t) P ij (z, t)µ j (t), (3.1.11) where t, z, z t  any time moments and i, j  possible states of model; boundary condition P ij (z, z) = δ ij. Obviuos that (3.1.11) may be presented: dp ij (z, t) = P ik (z, t)µ kj (t)dt P ij (z, t)µ j (t)dt. (3.1.12) k:k j Derivation of (3.1.11). Using (3.1.6) we have: P ij (z, t + t) = P ik (z, t)p kj (t, t + t) + P ij (z, t)p jj (t, t + t) then: k:k j P ij (z, t + t) P ij (z, t) t = k:k j P ik (z, t) P kj(t, t + t) t +P ij (z, t) P jj(t, t + t) 1. t Moreover: so 1 P jj (t, t + t) = P jk (t, t + t), k:k j
19 3.2. SEMI MARKOV PROCESSES AND MARKOV PROCESSES OF IIND ORDER19 P ij (z, t + t) P ij (z, t) t = k:k j P ik (z, t) P kj(t, t + t) P ij (z, t) P jk (t, t + t) t t k:k j Taking lim t + we get (3.1.11), or dp ij (z, t) = P ik (z, t)µ kj (t)dt P ij (z, t)µ j (t)dt. (3.1.13) k:k j Kolmogorov backward differential equations: d dz P ij(z, t) = P ij (z, t)µ i (z) P kj (z, t)µ ik (z) k:k i For derivation see [5]. Occupancy probabilities satisfy: with boundary condition P ii (z, z) = 1. From (3.1.14) we get: d dt P ii(z, t) = P ii (z, t)µ i (t), (3.1.14) d dt ln P ii(z, t) = µ i (t). (3.1.15) And from (3.1.15), using boundary condition P ii (z, z) = 1, we get: P ii (z, t) = ep[ t z µ i (u)du]. 3.2 Semi Markov processes and Markov processes of IInd order See [5]. In most actuarial models for health insurance it is possible to split states and use Markov process instead of semi Markov process, see [5].
20 2 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE 3.3 Actuarial values of premiums and reserves We will suppose that process satisfy Markov property. Define: p i (t)  rate of continuous premium paid at the moment t if S(t) = i, then p i (t)dt  premiums paid during the interval [t, t + dt). b j (t)  rate of annuity benefit paid at moment t if S(t) = j. c ij (t)  single (lump sum) benefit paid at moment t if transition from i to j occurs. d j (t)  single (lump sum) benefit paid at moment t if S(t) = j. { 1, if E  is true; Let: v = e δ and I E =, otherwise. Present value at the moment t of continuous annuity  b j (u) paid during interval [u, u + du) is: So: Y t (u, u + du) = v u t I S(u)=j b j (u)du. Y t (u 1 ; u 2 ) = u2 u 1 And actuarial present values are: v u t I S(u)=j b j (u)du. and E[Y t (u, u + du) S(t) = i] = v u t P ij (t, u)b j (u)du E[Y t (u 1 ; u 2 ) S(t) = i] = If b j (u) 1 then define: a ij (t, n) = t u2 u 1 v u t P ij (t, u)b j (u)du. v u t P ij (t, u). For lump sum benefit c jk (u) we have present value at moment t: Y t (u) = v u t I {S(u )=j S(u)=k}c jk (u)
21 3.3. ACTUARIAL VALUES OF PREMIUMS AND RESERVES 21 and actuarial present value: E[Y t (u) S(t) = i] = v u t P ij (t, u)µ jk (u)c jk (u)du. If c ij (u) = 1 then define: Ā ijk (t, n) = v u t P ij (t, u)µ jk (u)du t Ā i.k (t, n) = Ā ijk (t, n) j:j k Ā ij. (t, n) = Ā ijk (t, n) k:k j For lump sum benefit d j (u) present value at the moment t: and actuarial present value: If d j (u) = 1, then define: Y t (u) = v u t I S(u)=j d j (u), E[Y t (u) S(t) = i] = v u t P ij (t, u)d j (u). Ē ij (t, u) = v u t P ij (t, u). If c jk (u) = 1, u (t, n] and d j (m) = 1, then: Ā ij. (t, n, m) = Āij.(t, n) + Ēij(t, m). For n = m we get generalized formula of endowment insurance. Actuarial present value of benefits at the moment t if S(t) = i is: B i (t, n) = + t t v u t ( j v u t ( j = v u t ( u:u t j P ij (t, u)b j (u))du + P ij (t, u)µ jk (u)b jk (u))du + k:k j P ij (t, u)d j (u)). While actuarial present value of premiums is :
22 22 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE P i (t, n) = t v u t ( j P ij (t, u)p j (u))du, Then according to Equivalence principle: P 1 (, n) = B 1 (, n). Prospective reserve at time t (if S(t) = i) is calculated: V i (t) = B i (t, n) P i (t, n) Since S() = 1, then V 1 () =. If only two types of benefits b j (u) and c jk (u) are used then: V i (t) = t t t v u t j v u t j v u t j P ij (t, u)b j (u)du + (3.3.1) P ij (t, u)µ jk (u)c jk (u)du k:k j P ij (t, u)p j (u)du and: d dt V i (t) = δ V i (t) b i (t) + p i (t) j:j i µ ij (t)[c ij (t) + V j (t) V i (t)] (Use backward Kolmogorov differential equations for derivation of above formula). So: or d V i (t) = δ V i (t)dt + p i (t)dt b i (t)dt (3.3.2) j:j i µ ij (t)[c ij (t) + V j (t) V i (t)]dt (3.3.3)
23 3.3. ACTUARIAL VALUES OF PREMIUMS AND RESERVES 23 p i (t)dt = Investment premium { }} { [d V i (t) δ V i (t)dt] + Risk premium { }} { + b i (t) + µ ij (t)[c ij (t) + V j (t) V i (t)]dt j:j i = I + b i (t) + II Discrete time Markov processes Let I  set of possible states, S(t); t =, 1, stochastic process. Process S(t) is called discrete time Markov chain, if n and for any finite sets t < t 1 < < t n < u and i, i 1,..., i n the following holds: P r(s(u) = j S(t ) = i, S(t 1 ) = i 1,..., S(t n 1 = i n 1, S(t n ) = i n ) (3.3.4) = P r(s(u) = j S(t n ) = i n ), if P r(s(t ) = i, S(t 1 ) = i 1,..., S(t n 1 = i n 1, S(t n ) = i n, S(u) = j) >. In such case: P ij (t, u) = k I P ik (t, w)p kj (w, u), where t, w, u  discrete time moments (t w u). Obvious that: P ij (t, u) = k I P ik (t, t + 1)P kj (t + 1, u).
24 24 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE 3.4 Disability insurance Generally  3 state model: 1st state  "active", 2nd state  "invalid", 3rd  "dead" Hamza notation P 11 (, t) t p aa P 11 (z, t) t z p aa +z P 11 (, t) t p aa P 12 (, t) t p ai µ 12 (t) µ ai +t a 11 (, n) = a aa :n a 11 (, ) = a aa a 22 (t, n) = a ii +t:n t Suppose that age of insured person at start of policy is, S( + t)  state of policy at the moment + t, t >, S() = a. Disability annuity is paid, its present value: Define: Then: Y = v u I {S(+u)=i} du φ(, u) = P (S( + u) = i S() = a) ā ai = E(Y S() = a) = Let Γ  set of policy conditions: where: Γ = [n 1, n 2, f, m, r], v u φ(, u)du (3.4.1) n 1 n 2  waiting period (measured from start of policy)  end of policy period
25 3.4. DISABILITY INSURANCE 25 f  waiting period (measured from inception of illness) m  maimum annuity payment period (in years from start of payment) r  last moment when annuity is paid Define: φ Γ (, u)  probability that individual who bought insurance at () is ill and is paid annuity at time + u (according to policy conditionsγ). Obviuos: φ Γ (, u) φ(, u). And: For eample: φ [,,,, ] (, u) = φ(, u) φ [,n,,,n] (, u) = { φ(, u), if u < n;, if u n, and: ā ai,γ = ā ai : n = v u φ [,n,,,n] (, u)du = v u φ(, u)du We will suppose that conditions of Markov process are satisfied. Let: tp gh y = P (S(y + t) = h S(y) = g); h = a, i, d; g = a, i, and µ gh y tp gh y t t = lim ; h = a, i, d; g = a, i; h g, tp hh y = P (S(y + u) = h, u [, t] S(y) = h); h = a, i. Define probability: tp ai y (τ) = P (S(y + u) = i, u [t τ, t] S(y) = a); τ t Obviuos that: tp ai y () = t p ai y tp ai y (τ) =, τ t
26 26 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE or: If τ t, then: It is easy to show that: From (3.4.2) get: t τ tp ai y (τ) = up aa y µ ai y+u t u p ii y+udu. (3.4.2) ā ai = tp ai v t dt (3.4.3) t tp ai y = up aa y µ ai y+u t u py+udu. ii (3.4.4) Then from (3.4.3) and (3.4.4) we have: ā ai = t v t up aa After changing integration order: Define: ā ai = Define function: v u up aa µ ai +u t u p ii +ududt. (3.4.5) [ ] µ ai +u v t u t up+udt ii du. (3.4.6) u ā ii y = zpy ii v z dz. ψ(, u, t) = u p aa ψ(, t z, t) = t z p aa Then, for eample, if Γ = [, n,,, ]: µ ai +u t u p ii +u, µ ai +t z z p ii +t z, a ai,γ = = = tp ai t t v t dt + tp ai n ψ(, u, t)v t dudt + (t n)v t dt ψ(, t z, t)v t dzdt + n t n t n ψ(, u, t)v t dudt ψ(, t z, t)v t dzdt.
27 3.4. DISABILITY INSURANCE 27 If insurance premium is paid continuously at the rate P (u), if S(u) = a, then: For constant premium rate: v u up aa P (u)du = ā ai,γ. P = āai,γ. ā aa :m Let Γ = [, n,,, n]. Prospective reserve for active members at time moment + t is: V a +t,γ = t P m t v u up aa +t µ ai +t+ua ii +t+u:n t u du up aa +tv u du, and reserve for disabled members: V i +t,γ = t m t v u up ii +tdu P Eample of approimation formula: From (3.4.6)we have: ā ai = v u up aa up ia +tv u du. [ ] µ ai +u v t u t up ii +udt du u ā ai = Define: h= 1 h+up aa µ ai +h+uv h+u [ t h up ii +h+u vt h u dt]du (3.4.7) h+u Then: a +h+u = t h up ii +h+u vt h u dt. h+u
28 28 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE So: ā ai = h= 1 h+up aa µ ai +h+uv h+u a +h+udu (3.4.8) ā ai v h+1/2 a +h+1/2 h= 1 From Chapman  Kolmogorov equation: h+up aa µ ai +h+udu (3.4.9) and then h+up aa = h p aa up aa +h + h p ai up ia +h So finally: h+up aa h p aa up aa +h ā ai = = = v h+1/2 a +u+1/2 h= v h+1/2 a +u+1/2 h= h= h= 1 1 h+up aa µ ai +h+udu hp aa up aa +hµ ai +h+udu 1 v h+1/2 a +u+1/2 hp aa up aa +hµ ai +h+udu v h+1/2 a +u+1/2 hp aa ω +h. +udu  probability for inception of illness for indi where ω = 1 vidual aged (). u p aa µ ai 3.5 Critical illness insurance Define mortality probabilities µ ai ; µ ad = µ ad(o) ; µ id(o) ; µ id(d). Sometimes duration of illness r is important, then use probabilities µ id(o),r ; µ id(d),r. Then
29 3.5. CRITICAL ILLNESS INSURANCE 29 tp aa = t p aa = ep [ [ τp ii +u, = ep or, if we ignore illness duration r, : [ tp ii = ep t t (µ ai τ y+u + µ ad(o) y+u ] )du (µ id(o) y+u+r,r + µ id(d) y+u+r,r)dr (µ id(o) y+u ] + µ id(d) y+u )du Stand alone, or additional, benefit If transition a i occurs benefit of 1 is paid, define its present value by Ā (DD),n, then Ā (DD),n = up aa µ ai +uv u du. For continuous premium we have: P,n = Ā(DD),n, ā aa,n Reserve for active members: V a +t,n t = Ā(DD) +t,n t P,n ā aa +t,n t, where: Ā (DD) +t,n t = u tp aa +tµ ai +uv u t du. t In this case reserve in state i is not needed. Suppose that benefit is paid in installments α, β, 1 α β at time moments + u; + u + τ 1 ; + u + τ 1 + τ 2, second and third parts of benefit are paid only if insured person is alive at benefit payment moment. Then: [ ] Ā (DD;α,β),n = up aa µ ai +u αv u + τ1 p ii +u,βv u+τ 1 + τ1 +τ 2 p+u,(1 ii α β)v u+τ 1+τ 2 du, so premium is : P,n = Ā(DD;α,β),n ā aa,n ],
30 3 CHAPTER 3. ACTUARIAL MODELS FOR HEALTH INSURANCE and reserves: V a +t,n t = Ā(DD;α,β) +t,n t P,n ā aa +t,n t V i +t = u+τ1 tp ii +t,t uβv u+τ 1 t + u+τ1 +τ 2 tp ii +t,t u(1 α β)v u+τ 1+τ 2 t ; u t < τ 1 V i +t = u+τ2 tp ii +t,t u(1 α β)v u+τ 2 t ; u + τ 1 t < u + τ 1 + τ Accelerated insurance Suppose that fraction λ 1 of benefit is paid upon diagnosis of illness and rest part  (1 λ) on death of individual (if death occurs during policy period). Then: [ Ā (D+DD;λ),n = up aa µ ad(o) +u v u + µ ai +u(λv u + (1 λ) Define: Ā (DD;λ),n = Ā (D;λ),n = so: If λ = 1, then up aa µ ai up aa [ +uλv u du µ ad(o) +u v u + µ ai +u(1 λ) Ā (D+DD;λ),n Ā (DD;1),n = Ā (D;1),n = = Ā(D;λ),n u + Ā(DD;λ),n up aa µ ai +uv u du up aa µ ad(o) +u Formulas for discrete case may be found in [5]. 3.6 Long term care insurance v u du u ] v u+r rp ii +u(µ id(o) +u+r + µ id(d) +u+r)dr) d ] v u+r rp ii +u(µ id(o) +u+r + µ id(d) +u+r)dr du, Again, multiple state model is used to model LTC insurance policy conditions. For types of cover and calculation of premiums and / or reserves see [5].
A Competitive Model of Annuity and Life Insurance with Nonexclusive Contracts
A Competitive Model of Annuity and Life Insurance with Nonexclusive Contracts Roozbeh Hosseini Arizona Stat University Abstract I study a two period economy in which altruistic individuals have uncertain
More informationThe Role of Commitment in Dynamic Contracts: Evidence from Life Insurance
The Role of Commitment in Dynamic Contracts: Evidence from Life Insurance Igal Hendel and Alessandro Lizzeri Abstract We use a unique data set on life insurance contracts to study the properties of long
More informationOptimal Annuitization with Stochastic Mortality Probabilities
Working Paper Series Congressional Budget Office Washington, DC Optimal Annuitization with Stochastic Mortality Probabilities Felix Reichling Congressional Budget Office Felix.Reichling@cbo.gov Kent Smetters
More informationReducing asymmetric information with usagebased automobile insurance
Reducing asymmetric information with usagebased automobile insurance Sara Arvidsson VTI*/CTS** Abstract: Automobile insurers currently use available information about the vehicle, the owner and residential
More informationAnnuities versus Bonds Optimal Consumption and Investment Decisions in a Continuous Life Cycle Model
Annuities versus Bonds Optimal Consumption and Investment Decisions in a Continuous Life Cycle Model Bernhard Skritek Research Report 201109 September, 2011 Operations Research and Control Systems Institute
More informationA Generalization of the MeanVariance Analysis
A Generalization of the MeanVariance Analysis Valeri Zakamouline and Steen Koekebakker This revision: May 30, 2008 Abstract In this paper we consider a decision maker whose utility function has a kink
More informationTO QUEUE OR NOT TO QUEUE: EQUILIBRIUM BEHAVIOR IN QUEUEING SYSTEMS
TO QUEUE OR NOT TO QUEUE: EQUILIBRIUM BEHAVIOR IN QUEUEING SYSTEMS REFAEL HASSIN Department of Statistics and Operations Research Tel Aviv University Tel Aviv 69978, Israel hassin@post.tau.ac.il MOSHE
More informationCan The Private Annuity Market Provide Secure Retirement Income?
WP/04/230 Can The Private Annuity Market Provide Secure Retirement Income? G.A. Mackenzie and Allison Schrager 2004 International Monetary Fund WP/04/230 IMF Working Paper Research Department Can The Private
More informationFederal Reserve Bank of Chicago
Federal Reserve Bank of Chicago Accounting for nonannuitization Svetlana Pashchenko WP 201003 Accounting for nonannuitization Svetlana Pashchenko University of Virginia March 25, 2010 Abstract Why don
More informationNotes  Gruber, Public Finance Section 12.1 Social Insurance What is insurance? Individuals pay money to an insurer (private firm or gov).
Notes  Gruber, Public Finance Section 12.1 Social Insurance What is insurance? Individuals pay money to an insurer (private firm or gov). These payments are called premiums. Insurer promises to make a
More informationAN INTRODUCTION TO PREMIUM TREND
AN INTRODUCTION TO PREMIUM TREND Burt D. Jones * February, 2002 Acknowledgement I would like to acknowledge the valuable assistance of Catherine Taylor, who was instrumental in the development of this
More informationADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET
ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET Amy Finkelstein Harvard University and NBER James Poterba MIT and NBER Revised August 2002 ABSTRACT In this paper,
More informationEDUCATION AND EXAMINATION COMMITTEE SOCIETY OF ACTUARIES RISK AND INSURANCE. Copyright 2005 by the Society of Actuaries
EDUCATION AND EXAMINATION COMMITTEE OF THE SOCIET OF ACTUARIES RISK AND INSURANCE by Judy Feldman Anderson, FSA and Robert L. Brown, FSA Copyright 25 by the Society of Actuaries The Education and Examination
More informationNBER WORKING PAPER SERIES PRICE ELASTICITY OF DEMAND FOR TERM LIFE INSURANCE AND ADVERSE SELECTION
NBER WORKING PAPER SERIES PRICE ELASTICITY OF DEMAND FOR TERM LIFE INSURANCE AND ADVERSE SELECTION Mark V. Pauly Kate H. Withers Krupa SubramanianViswanathan Jean Lemaire John C. Hershey Katrina Armstrong
More informationPitfalls in the implementation of nondiscriminatory premiums the case of unisex tariffs in the German automobile insurance market
Pitfalls in the implementation of nondiscriminatory premiums the case of unisex tariffs in the German automobile insurance market Vijay Aseervatham Institute for Risk Management and Insurance, University
More informationExpected Value and Variance
Chapter 6 Expected Value and Variance 6.1 Expected Value of Discrete Random Variables When a large collection of numbers is assembled, as in a census, we are usually interested not in the individual numbers,
More informationNew insights on the meanvariance portfolio selection from de Finetti s suggestions. Flavio Pressacco and Paolo Serafini, Università di Udine
New insights on the meanvariance portfolio selection from de Finetti s suggestions Flavio Pressacco and Paolo Serafini, Università di Udine Abstract: In this paper we offer an alternative approach to
More informationCostly Voting when both Information and Preferences Di er: Is Turnout Too High or Too Low?
Costly Voting when both Information and Preferences Di er: Is Turnout Too High or Too Low? Sayantan Ghosal and Ben Lockwood University of Warwick October, 2008 Abstract We study a model of costly voting
More informationAccounting for nonannuitization
MPRA Munich Personal RePEc Archive Accounting for nonannuitization Svetlana Pashchenko Uppsala University 19. November 2012 Online at http://mpra.ub.unimuenchen.de/42792/ MPRA Paper No. 42792, posted
More informationThe taxation of retirement saving is an important and. Taxing Retirement Income: Nonqualified Annuities and Distributions from Qualified Accounts
Taxing Retirement Income Taxing Retirement Income: Nonqualified Annuities and Distributions from Qualified Accounts Jeffrey R. Brown John F. Kennedy School of Government, Harvard University, and NBER,
More informationHuntington s Disease, Critical Illness Insurance and Life Insurance
Scand. Actuarial J. 2004; 4: 279/313 æoriginal ARTICLE Huntington s Disease, Critical Illness Insurance and Life Insurance CRISTINA GUTIÉRREZ and ANGUS MACDONALD Gutiérrez C. and Macdonald A. Huntington
More informationReference point adaptation: Tests in the domain of security trading q
Available online at www.sciencedirect.com Organizational Behavior and Human Decision Processes 105 (2008) 67 81 www.elsevier.com/locate/obhdp Reference point adaptation: Tests in the domain of security
More informationNot Only What But also When: A Theory of Dynamic Voluntary Disclosure
Not Only What But also When: A Theory of Dynamic Voluntary Disclosure Ilan Guttman, Ilan Kremer, and Andrzej Skrzypacz Stanford Graduate School of Business September 2012 Abstract The extant theoretical
More informationFederal Reserve Bank of New York Staff Reports
Federal Reserve Bank of New York Staff Reports Tax Buyouts Marco Del Negro Fabrizio Perri Fabiano Schivardi Staff Report no. 467 August 2010 This paper presents preliminary findings and is being distributed
More informationSocial Security, Life Insurance and Annuities for Families
Social Security, Life Insurance and Annuities for Families Jay H. Hong University of Pennsylvania JoséVíctor RíosRull University of Pennsylvania, CAERP, CEPR, NBER April Abstract In this paper we ask
More informationRegulating withdrawals from individual pension accounts. consumption
World Bank Pension Reform Primer Annuities Regulating withdrawals from individual pension accounts P ension, to most people, implies a regular payment from a specific age such as retirement until death.
More informationContracts as Reference Points Experimental Evidence
American Economic Review 101 (April 2011): 493 525 http://www.aeaweb.org/articles.php?doi=10.1257/aer.101.2.493 Contracts as Reference Points Experimental Evidence By Ernst Fehr, Oliver Hart, and Christian
More informationSelling assets: When is the whole worth more than the sum of its parts?
Selling assets: When is the whole worth more than the sum of its parts? Robert Marquez University of California, Davis Rajdeep Singh University of Minnesota October, 214 Abstract When is it better to sell
More informationTesting for Adverse Selection with Unused Observables
Testing for Adverse Selection with Unused Observables Amy Finkelstein MIT and NBER James Poterba MIT and NBER February 2006 ABSTRACT This paper proposes a new test for adverse selection in insurance markets
More informationTax Buyouts. Marco Del Negro Federal Reserve Bank of New York
Tax Buyouts Marco Del Negro Federal Reserve Bank of New York Fabrizio Perri University of Minnesota, Federal Reserve Bank of Minneapolis CEPR and NBER Fabiano Schivardi Universita di Cagliari, Einaudi
More information