Hattendorff's theorem

Last updated September 13, 2020

Hattendorff's Theorem, attributed to K. Hattendorff (1868), is a theorem in actuarial science that describes the allocation of the variance or risk of the loss random variable over the lifetime of an actuarial reserve. In other words, Hattendorff's theorem demonstrates that the variation in the present value of the loss of an issued insurance policy can be allocated to the future years during which the insured is still alive. This, in turn, facilitates the management of risk prevalent in such insurance contracts over short periods of time.^[1]

Hattendorff's Theorem

The main result of the theorem has three equivalent formulations:

Hattendorff's Theorem

${\begin{aligned}&\mathrm {Var} [L_{h}|K(x)\geq h]\\&=\sum _{k=h}^{\infty }v^{2(k-h)}\mathrm {Var} [\Lambda _{k}|K(x)\geq h]\\&=\mathrm {Var} [\Lambda _{h}|K(x)\geq h]+v^{2}\mathrm {Var} [L_{h+1}|K(x)\geq h]\\&=\sum _{k=h}^{h+j-1}v^{2(k-h)}\mathrm {Var} [\Lambda _{k}|K(x)\geq h]+v^{2j}\mathrm {Var} [L_{h+j}|K(x)\geq h]\end{aligned}}$

where:

Variable	Explanation
$K(x)$	The number of whole years that a life status x survives. If $T_{x}$ is the distribution of the lifetime of an insured, then $K(x)=\lfloor T_{x}\rfloor$ .
$_{k}p_{j}$	Actuarial notation for $\mathrm {Pr} (j\leq T_{x}<j+k)$ .
$\pi _{j}$	The premium received by the insured in year j.
$b_{j}$	The benefit paid to the insured in year j.
$L_{h}$	The actuarial present value of the total loss over the remaining life of the policy at time h.
$C_{h}$	The present value of the net cash loss from the policy in the year (h, h+1).
$v$	The discount factor for one year.
$\Lambda _{h}$	The present value of the net cash loss from the policy plus the change in total liabilities in the year (h, h+1).
$V_{h}$	The benefit reserve at time h, equal to $\mathbb {E} [L_{h}\|K(x)\geq h]$ .

In its above formulation, and in particular the first result, Hattendorff's theorem states that the variance of $L_{h}$ , the insurer's total loss over the remaining life of the policy at time h, can be calculated by discounting the variances of the yearly net losses (cash losses plus changes in net liabilities) $\Lambda _{k}$ in future years.

Background^[2]

In the most general stochastic setting in which the analysis of reserves is carried out, consider an insurance policy written at time zero, over which the insured pays yearly premiums $\pi _{0},\pi _{1}\dots \pi _{K(x)}$ at the beginning of each year starting today until the year of death of the insured. Furthermore, the insured receives a benefit of $K(x)+1$ , at the end of the year of death, equal to $b_{K(x)+1}$ . No other payments are received nor paid over the lifetime of the policy.

Suppose an insurance company is interested to know the cash loss from this policy over the year (h, h+1). Of course, if the death of the insured happens prior to time h, or when $K(x)<h$ , then there is no remaining loss and $C_{h}=0$ . If the death of the insured occurs exactly at time h, or when $K(x)=h$ , then the loss on the policy is equal to the present value of the benefit paid in the following year, $vb_{h+1}$ , less the premium paid at time h. Hence in this case $C_{h}=vb_{h+1}-\pi _{h}.$ Lastly, if the death of the insured occurs after time h, or when $K(x)>h$ , then the cash loss in the year (h, h+1) is just the negative of the premium received at time h (cash inflows are treated as negative losses). Hence we summarize this result as

C_{h}={\begin{cases}0&{\mbox{if }}K(x)=0,1\dots h-1\\vb_{h+1}-\pi _{h}&{\mbox{if }}K(x)=h\\-\pi _{h}&{\mbox{if }}K(x)=h+1,h+2\dots \\\end{cases}}

Furthermore, the actuarial present value of the future cash losses in each year has the explicit formula

L_{h}={\begin{cases}0&{\mbox{if }}K(x)=0,1\dots h-1\\vb_{K(x)+1}-\pi _{K(x)}&{\mbox{if }}K(x)=h\\v^{K(x)-h+1}b_{K(x)+1}-\sum _{k=h}^{K(x)}v^{k-h}\pi _{k}&{\mbox{if }}K(x)=h+1,h+2\dots \end{cases}}

Derivation of the formula for

L_{h}

.

The present value of the loss on the policy at time h is the present value of all future cash losses

L_{h}=\sum _{k=h}^{\infty }v^{k-h}C_{h}.

Expanding this result, it is easy to see using the definition of $C_{h}$ that, when $K(x)>h$ ,

{\begin{aligned}L_{h}&=\sum _{k=h}^{K(x)-1}v^{k-h}C_{h}+v^{K(x)-h}C_{K(x)}+\sum _{k=K(x)+1}^{\infty }v^{k-h}C_{h}\\&=-\sum _{k=h}^{K(x)-1}v^{k-h}\pi _{k}+v^{K(x)-h}\left(vb_{K(x)+1}-\pi _{h}\right)\\&=v^{K(x)-h+1}b_{K(x)+1}-\sum _{k=h}^{K(x)}v^{k-h}\pi _{k}.\end{aligned}}

Similarly, when $K(x)=h$ , then $L_{h}=C_{K(x)}$ . Finally, when $K(x)<h$ , the summation, and hence the loss on the policy, is zero.

In the analysis of reserves, a central quantity of interest is the benefit reserve $V_{h}$ at time h, which is the expected loss on the policy at time h given that status x has survived to age h

V_{h}=\mathbb {E} [L_{h}|K(x)\geq h]

which admits to the closed form expression

V_{h}=\sum _{k=0}^{\infty }\left(v^{k+1}b_{k+h+1}-\sum _{j=0}^{k}v^{j}\pi _{j+h}\right){_{k}p_{x+h}}q_{x+h+k}

.

Derivation of the formula for

V_{h}

.

Here we derive the above formula for the benefit reserve.

{\begin{aligned}V_{h}&=\mathbb {E} [L_{h}|K(x)\geq h]\\&=\mathbb {E} \left[v^{K(x)-h+1}b_{K(x)+1}-\sum _{k=h}^{K(x)}v^{k-h}\pi _{k}|K(x)\geq h\right].\end{aligned}}

In order to proceed, we make the assumption that the remaining lifetime of a life status x that has lived to time h, $K(x)-h$ , follows the same (kurtate) probability distribution as another randomly chosen individual from the group of insureds but of age $x+h$ , with distribution $K(x+h)$ . This means that, in terms of expected values, $\mathbb {E} [f(K(x)-h)|K(x)\geq h]=\mathbb {E} [f(K(x+h))]$ for any function over which the expectation is defined. Then, using a clever algebraic trick, we can rewrite the benefit reserve as

{\begin{aligned}V_{h}&=\mathbb {E} \left[v^{(K(x)-h)+1}b_{(K(x)-h)+h+1}-\sum _{j=0}^{K(x)-h}v^{j}\pi _{j+h}|K(x)\geq h\right]\\&=\mathbb {E} \left[v^{K(x+h)+1}b_{K(x+h)+h+1}-\sum _{j=0}^{K(x+h)}v^{j}\pi _{j+h}\right]\\&=\sum _{k=0}^{\infty }\left(v^{k+1}b_{k+h+1}-\sum _{j=0}^{k}v^{j}\pi _{j+h}\right)\mathrm {Pr} (K(x+h)=k)\\&={\sum _{k=0}^{\infty }\left(v^{k+1}b_{k+h+1}-\sum _{j=0}^{k}v^{j}\pi _{j+h}\right)}{_{k}p_{x+h}}q_{x+h+k}\end{aligned}}

Lastly, the present value of the net cash loss at time h over the year (h, h+1), denoted $\Lambda _{h}$ , is equal to the present value of the cash loss in year h, $C_{h}$ (see above), plus the present value of the change in liabilities $PV(\Delta V_{h})$ at time h. In other words, $\Lambda _{h}=C_{h}+v\Delta Liabilities$ . If $K(x)>h$ , then $\Lambda _{h}=-\pi _{h}+(vV_{h+1}-V_{h})$ . Similarly, if $K(x)=h$ , then $\Lambda _{h}=(vb_{h+1}-\pi _{h})-V_{h}$ since there is no reserve after the year of death. Finally, if $K(x)<h$ , then there is no loss in the future and $\Lambda _{h}=0$ . Summarizing, this yields the following result, which is important in the formulation of Hattendorff's theorem

\Lambda _{h}={\begin{cases}0&{\mbox{if }}K(x)=0,1\dots h-1\\(vb_{h+1}-\pi _{h})-V_{h}&{\mbox{if }}K(x)=h\\(vV_{h+1}-V_{h})-\pi _{h}&{\mbox{if }}K(x)=h+1,h+2\dots \end{cases}}.

Proofs

The proof of the first equality is written as follows. First, by writing the present value of future net losses at time h,

{\begin{aligned}\sum _{k=h}^{\infty }v^{k-h}\Lambda _{k}&=\sum _{k=h}^{\infty }v^{k-h}[C_{k}+v\Delta {\text{Liabilities in year }}(k,k+1)]\\&=\sum _{k=h}^{\infty }v^{k-h}C_{k}+\sum _{k=h}^{\infty }v^{k-h+1}\Delta {\text{Liabilities in year }}(k,k+1)\\&=L_{h}+\sum _{k=h}^{\infty }v^{k-h+1}(vV_{k+1}-V_{k})\\&=L_{h}+\sum _{k=h}^{\infty }v^{(k+1)-h+1}V_{k+1}-\sum _{k=h}^{\infty }v^{k-h+1}V_{k}\\\end{aligned}}

from which it is easy to see that

L_{h}=\sum _{k=h}^{\infty }v^{k-h}\Lambda _{k}+V_{h}.

It is known that the individual net cash flows in different years are uncorrelated, or $\mathrm {Cov} (\Lambda _{h}\Lambda _{j}|K(x)\geq k)=0$ when $k\leq h<j$ (see Bowers et al., 1997, for a proof of this result). Using these two results, we conclude that

$\mathrm {Var} [L_{h}|K(x)\geq h]=\mathrm {Var} \left[\sum _{k=h}^{\infty }v^{k-h}\Lambda _{k}+V_{h}|K(x)\geq h\right]=\sum _{k=h}^{\infty }v^{2(k-h)}\mathrm {Var} [\Lambda _{k}|K(x)\geq h]$

which proves the first part of the theorem. The reader is referred to (Bowers et al., pg 241) for the proof of the other equalities.

Related Research Articles

In mathematics, the pentagonal number theorem, originally due to Euler, relates the product and series representations of the Euler function. It states that

In the theory of stochastic processes, the Karhunen–Loève theorem, also known as the Kosambi–Karhunen–Loève theorem is a representation of a stochastic process as an infinite linear combination of orthogonal functions, analogous to a Fourier series representation of a function on a bounded interval. The transformation is also known as Hotelling transform and eigenvector transform, and is closely related to principal component analysis (PCA) technique widely used in image processing and in data analysis in many fields.

In statistics and information theory, a maximum entropy probability distribution has entropy that is at least as great as that of all other members of a specified class of probability distributions. According to the principle of maximum entropy, if nothing is known about a distribution except that it belongs to a certain class, then the distribution with the largest entropy should be chosen as the least-informative default. The motivation is twofold: first, maximizing entropy minimizes the amount of prior information built into the distribution; second, many physical systems tend to move towards maximal entropy configurations over time.

The birth–death process is a special case of continuous-time Markov process where the state transitions are of only two types: "births", which increase the state variable by one and "deaths", which decrease the state by one. The model's name comes from a common application, the use of such models to represent the current size of a population where the transitions are literal births and deaths. Birth–death processes have many applications in demography, queueing theory, performance engineering, epidemiology, biology and other areas. They may be used, for example, to study the evolution of bacteria, the number of people with a disease within a population, or the number of customers in line at the supermarket.

In mathematics, the Weyl character formula in representation theory describes the characters of irreducible representations of compact Lie groups in terms of their highest weights. It was proved by Hermann Weyl. There is a closely related formula for the character of an irreducible representation of a semisimple Lie algebra. In Weyl's approach to the representation theory of connected compact Lie groups, the proof of the character formula is a key step in proving that every dominant integral element actually arises as the highest weight of some irreducible representation. Important consequences of the character formula are the Weyl dimension formula and the Kostant multiplicity formula.

In mathematics, the Chebyshev function is either of two related functions. The first Chebyshev function $ϑ (x)$ or $θ (x)$ is given by

In mathematics, the Schur orthogonality relations, which is proven by Issai Schur through Schur's lemma, express a central fact about representations of finite groups. They admit a generalization to the case of compact groups in general, and in particular compact Lie groups, such as the rotation group SO(3).

A ratio distribution is a probability distribution constructed as the distribution of the ratio of random variables having two other known distributions. Given two random variables X and Y, the distribution of the random variable Z that is formed as the ratio Z = X/Y is a ratio distribution.

In mathematics, the Fraňková–Helly selection theorem is a generalisation of Helly's selection theorem for functions of bounded variation to the case of regulated functions. It was proved in 1991 by the Czech mathematician Dana Fraňková.

In mathematics, the Remez inequality, discovered by the Soviet mathematician Evgeny Yakovlevich Remez, gives a bound on the sup norms of certain polynomials, the bound being attained by the Chebyshev polynomials.

In probability theory and statistics, the Conway–Maxwell–Poisson distribution is a discrete probability distribution named after Richard W. Conway, William L. Maxwell, and Siméon Denis Poisson that generalizes the Poisson distribution by adding a parameter to model overdispersion and underdispersion. It is a member of the exponential family, has the Poisson distribution and geometric distribution as special cases and the Bernoulli distribution as a limiting case.

In mathematics, the multiple zeta functions are generalisations of the Riemann zeta function, defined by

In mathematics, Maass forms or Maass wave forms are studied in the theory of automorphic forms. Maass forms are complex-valued smooth functions of the upper half plane, which transform in a similar way under the operation of a discrete subgroup $of as modular forms. They are Eigenforms of the hyperbolic Laplace Operator defined on and satisfy certain growth conditions at the cusps of a fundamental domain of . In contrast to the modular forms the Maass forms need not be holomorphic. They were studied first by Hans Maass in 1949.$

In the branch of mathematics called graph theory, the strength of an undirected graph corresponds to the minimum ratio edges removed/components created in a decomposition of the graph in question. It is a method to compute partitions of the set of vertices and detect zones of high concentration of edges, and is analogous to graph toughness which is defined similarly for vertex removal.

In mathematics, particularly linear algebra, the Schur–Horn theorem, named after Issai Schur and Alfred Horn, characterizes the diagonal of a Hermitian matrix with given eigenvalues. It has inspired investigations and substantial generalizations in the setting of symplectic geometry. A few important generalizations are Kostant's convexity theorem, Atiyah–Guillemin–Sternberg convexity theorem, Kirwan convexity theorem.

In mathematics, Khabibullin's conjecture, named after B. N. Khabibullin, is related to Paley's problem for plurisubharmonic functions and to various extremal problems in the theory of entire functions of several variables.

Explicit formulas for eigenvalues and eigenvectors of the second derivative with different boundary conditions are provided both for the continuous and discrete cases. In the discrete case, the standard central difference approximation of the second derivative is used on a uniform grid.

In mathematical analysis, Krein's condition provides a necessary and sufficient condition for exponential sums

In mathematics, Ramanujan's master theorem is a technique that provides an analytic expression for the Mellin transform of an analytic function.

In mathematics, singular integral operators of convolution type are the singular integral operators that arise on Rⁿ and Tⁿ through convolution by distributions; equivalently they are the singular integral operators that commute with translations. The classical examples in harmonic analysis are the harmonic conjugation operator on the circle, the Hilbert transform on the circle and the real line, the Beurling transform in the complex plane and the Riesz transforms in Euclidean space. The continuity of these operators on L² is evident because the Fourier transform converts them into multiplication operators. Continuity on L^p spaces was first established by Marcel Riesz. The classical techniques include the use of Poisson integrals, interpolation theory and the Hardy–Littlewood maximal function. For more general operators, fundamental new techniques, introduced by Alberto Calderón and Antoni Zygmund in 1952, were developed by a number of authors to give general criteria for continuity on L^p spaces. This article explains the theory for the classical operators and sketches the subsequent general theory.

References

↑ Bowers, N.L.; et al. (1997). Actuarial Mathematics, 2nd Edition . pp. 229–245.
↑ Gerber, H.U., Leung, B.P.K., Shiu, E.S.W. "Indicator Function and Hattendorff's Theorem". North American Actuarial Journal. 7 (1). pp. 41–42.CS1 maint: multiple names: authors list (link)

External links

YouTube video explanation

This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.

[1] Bowers, N.L.; et al. (1997). Actuarial Mathematics, 2nd Edition . pp. 229–245.

[2] Gerber, H.U., Leung, B.P.K., Shiu, E.S.W. "Indicator Function and Hattendorff's Theorem". North American Actuarial Journal. 7 (1). pp. 41–42.CS1 maint: multiple names: authors list (link)