Distribution of shifted discrete random walk generated by distinct random variables and applications in ruin theory

Simonas Gervė and Andrius Grigutis

arXiv: 2211.14629v1 · PDF · 26 November 2022
Categories: math.PR
DOI: 10.15559/24-VMSTA249
Licence: CC BY 4.0 The site shows the text.

Abstract

In this paper, we set up the distribution function

\varphi(u)=\mathbb{P}\left(\sup_{n\geqslant 1}\sum_{i=1}^{n}\left(X_i-\kappa\right)<u\right),

and the generating function of \varphi(u+1), where u\in\mathbb{N}_0, \kappa\in\mathbb{N}, the random walk \left\{\sum_{i=1}^{n}X_i, n\in\mathbb{N}\right\}, consists of N\in\mathbb{N} periodically occurring distributions, and the integer-valued and non-negative random variables X_1,\,X_2,\,\ldots are independent. This research generalizes two recent works where \{\kappa=1,\,N\in\mathbb{N}\} and \{\kappa\in\mathbb{N},\,N=1\} were considered respectively. The provided sequence of sums \left\{\sum_{i=1}^{n}\left(X_i-\kappa\right),\,n\in\mathbb{N}\right\} generates so-called multi-seasonal discrete-time risk model with arbitrary natural premium and its known distribution enables to calculate the ultimate time ruin probability 1-\varphi(u) or survival probability \varphi(u). Verifying obtained theoretical statements we demonstrate several computational examples for survival probability \varphi(u) and its generating function when \{\kappa=2,\,N=2\}, \{\kappa=3,\,N=2\}, \{\kappa=5,\,N=10\} and X_i admits Poisson and some other distributions. We also conjecture the non-singularity of certain matrices.

Items 21

Theorem

Theorem 1

M1JV

Suppose that the N-seasonal discrete-time risk model (?) is generated by random variables X_1,\, X_2,\, \ldots,\, X_N and the net profit condition \mathbb{E} S_N<\kappa N is satisfied. Then the following statements are correct:

The probability mass functions of random variables \mathcal{M}_1,\,\mathcal{M}_2,\, \ldots,\, \mathcal{M}_N and X_1,\,X_2,\, \ldots,\, X_N are related by
\begin{aligned} &\hspace{-1cm}\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(\kappa-i-j)+\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(\kappa-i-j) +\sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(\kappa-i-j)+\ldots\\ &\hspace{-1cm}+\sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j(\kappa-i-j) =\kappa N -\mathbb{E}S_N. \end{aligned}
If 0<|s|\leqslant1, then
\begin{aligned} &\hspace{-1cm}\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(s^\kappa-s^{i+j}) +\frac{G_{X_N}(s)}{s^\kappa}\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(s^\kappa-s^{i+j})\\ &\hspace{-1cm}+\frac{G_{X_N+X_1}(s)}{s^{2\kappa}} \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(s^\kappa-s^{i+j})+\ldots\\ &\hspace{-1cm}+\frac{G_{X_N+X_1+\ldots+X_{N-2}}(s)}{s^{\kappa(N-1)}} \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j(s^\kappa-s^{i+j}) =s^\kappa G_{M_N}(s)\left(1-\frac{G_{S_N}(s)}{s^{\kappa N}}\right) \end{aligned}
and, if \alpha\neq1, 0<|\alpha|\leqslant1, is a root of G_{S_N}(s)=s^{\kappa N}, then
\begin{aligned} &\hspace{-1cm}\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_N}(j-i) +\frac{G_{X_N}(\alpha)}{\alpha^\kappa}\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_1}(j-i)\\ &\hspace{-1cm}+\frac{G_{X_N+X_1}(\alpha)}{\alpha^{2\kappa}} \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_2}(j-i)+\ldots\\ &\hspace{-1cm}+\frac{G_{X_N+X_1+\ldots+X_{N-2}}(\alpha)}{\alpha^{\kappa(N-1)}} \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_{N-1}}(j-i) =0. \end{aligned}
If \alpha\neq1, 0<|\alpha|\leqslant1, is a root of G_{S_N}(s)=s^{\kappa N} of multiplicity r, r=2,\,3,\,...,\,\kappa N-1, then
\begin{aligned} &\hspace{-1cm}\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=i}^{\kappa-1}\frac{d^n}{ds^n}\left(s^j\right)\Big{|}_{s=\alpha}F_{X_N}(j-i)+ \sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=i}^{\kappa-1}\frac{d^n}{ds^n}\left(G_{X_N}(s)s^{j-\kappa}\right)\Big{|}_{s=\alpha} F_{X_1}(j-i)\\ &\hspace{-1cm}+ \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=i}^{\kappa-1}\frac{d^n}{ds^n}\left(G_{X_N+X_1}(s)s^{j-2\kappa}\right)\Big{|}_{s=\alpha} F_{X_2}(j-i)+\ldots\\ &\hspace{-1cm}+\sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=i}^{\kappa-1}\frac{d^n}{ds^n}\left(G_{X_N+X_1+\ldots+X_{N-2}}(s)s^{j-\kappa(N-1)}\right)\Big{|}_{s=\alpha} F_{X_{N-1}}(j-i) =0, \end{aligned}
for all n\in\{1,\,2,\,\ldots,\,r-1\}, where \frac{d^n}{dt^n}(\ldots)|_{t=\alpha} denotes the n'th derivative at s=\alpha.
For n=\kappa,\,\kappa+1,\,\ldots\,, the probability mass functions of random variables \mathcal{M}_1,\,\mathcal{M}_2,\, \ldots,\, \mathcal{M}_N and X_1,\,X_2,\, \ldots,\, X_N satisfy the following system of equations
\begin{aligned} \begin{cases} m^{(1)}_nx^{(N)}_0&=m^{(N)}_{n-\kappa}-\sum_{i=0}^{n-1}m^{(1)}_ix^{(N)}_{n-i}-\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j\mathbbm{1}_{\{n=\kappa\}}\\ m^{(2)}_nx^{(1)}_0&=m^{(1)}_{n-\kappa}-\sum_{i=0}^{n-1}m^{(2)}_ix^{(1)}_{n-i}-\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j\mathbbm{1}_{\{n=\kappa\}}\\ m^{(3)}_nx^{(2)}_0&=m^{(2)}_{n-\kappa}-\sum_{i=0}^{n-1}m^{(3)}_ix^{(2)}_{n-i}-\sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j\mathbbm{1}_{\{n=\kappa\}}\\ &\,\,\vdots\\ m^{(N)}_nx^{(N-1)}_0&=m^{(N-1)}_{n-\kappa}-\sum_{i=0}^{n-1}m^{(N)}_ix^{(N-1)}_{n-i}-\sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j\mathbbm{1}_{\{n=\kappa\}}\\ \end{cases}. \end{aligned}

Proof (Proof of Theorem 1) of Theorem 1

We first prove equality (?). To derive (?), we use the system of equations (3) from Lemma 3. According to the conditions of Lemma 3, s\ne0 and we rearrange the system (3) by multiplying its first equality by 1, the second one by G_{X_N}(s)/s^\kappa, the third one by G_{X_N+X_1}(s)/s^{2\kappa} and so on till the last equality which we multiply by G_{X_N+X_1+\ldots+X_{N-2}}(s)/s^{\kappa(N-1)}. We then add up all these equations and obtain

\begin{aligned} &\hspace{-1cm}\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(s^\kappa-s^{i+j})+ \frac{G_{X_N}(s)}{s^\kappa}\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(s^\kappa-s^{i+j})\\ &\hspace{-1cm}+\frac{G_{X_N+X_1}(s)}{s^{2\kappa}} \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(s^\kappa-s^{i+j})+\ldots+\frac{G_{X_N+X_1+\ldots+X_{N-2}}(s)}{s^{\kappa(N-1)}} \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}m^{(N-1)}_j(s^\kappa-s^{i+j}) \\ &\hspace{-1cm}=s^\kappa G_{\mathcal{M}_N}(s)-G_{X_N}(s)G_{\mathcal{M}_1}(s)+ \frac{G_{X_N}(s)}{s^\kappa}\left(s^\kappa G_{\mathcal{M}_1}(s)-G_{X_1}(s)G_{\mathcal{M}_2}(s)\right)\\ &\hspace{-1cm}+\frac{G_{X_N+X_1}(s)}{s^{2\kappa}}\left(s^\kappa G_{\mathcal{M}_2}(s)-G_{X_2}(s)G_{\mathcal{M}_3}(s)\right) +\ldots+\frac{G_{X_N+X_1+\ldots+X_{N-2}}(s)}{s^{\kappa(N-1)}}\left(s^\kappa G_{\mathcal{M}_{N-1}}(s) - G_{X_{N-1}}(s)G_{\mathcal{M}_N}(s)\right)\\ &\hspace{-1cm}=s^\kappa G_{\mathcal{M}_N}(s)-\frac{G_{S_N}(s)}{s^{\kappa(N-1)}}G_{\mathcal{M}_N}(s) =s^{\kappa}G_{\mathcal{M}_N}(s)\left(1-\frac{G_{S_N}(s)}{s^{\kappa N}}\right). \end{aligned}

Here we have used the fact that if any random variables X are Y independent, then

G_{X+Y}(s) = G_{X}(s) G_{Y}(s).

Thus, the equality (?) is proved. We now derive (?).

It is obvious that the right-hand-side of (Proof) equals zero if we set s=\alpha, where \alpha\neq1 is the root of G_{S_N}(s)=s^{\kappa N},\,|s|\leqslant1. We then divide both sides of (Proof) by \alpha-1, i. e.

\frac{\alpha^\kappa-\alpha^{i+j}}{\alpha-1}=\alpha^{j+i}+\alpha^{j+i+1}+\ldots+\alpha^{\kappa-1}, \,\alpha\neq1,

and get

\begin{aligned} &\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j\sum_{l=j+i}^{\kappa-1}\alpha^l+ \frac{G_{X_N}(s)}{s^\kappa}\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j\sum_{l=j+i}^{\kappa-1}\alpha^l\\ &+\frac{G_{X_N+X_1}(s)}{s^{2\kappa}} \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j\sum_{l=j+i}^{\kappa-1}\alpha^l+\ldots\\ &+\frac{G_{X_N+X_1+\ldots+X_{N-2}}(s)}{s^{\kappa(N-1)}} \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j\sum_{l=j+i}^{\kappa-1}\alpha^l\\ &=\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_N}(j-i) +\frac{G_{X_N}(\alpha)}{\alpha^\kappa}\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_1}(j-i)\\ &+\frac{G_{X_N+X_1}(\alpha)}{\alpha^{2\kappa}} \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_2}(j-i)+\ldots\\ &+\frac{G_{X_N+X_1+\ldots+X_{N-2}}(\alpha)}{\alpha^{\kappa(N-1)}} \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=i}^{\kappa-1}\alpha^j F_{X_{N-1}}(j-i) =0, \end{aligned}

which is the claimed equality (?).

We now consider the equation (?). Since s\neq1, we divide the both sides of (Proof) by s-1 and rewrite the right-hand-side of it as follows

s^{\kappa(1-N)}G_{\mathcal{M}_N}(s)\left(s^{\kappa N}-G_{S_N}(s)\right)/(s-1).

Clearly, derivatives

\frac{d^n}{ds^n}\left(s^{\kappa(1-N)}G_{\mathcal{M}_N}(s)\left(s^{\kappa N}-G_{S_N}(s)\right)/(s-1)\right)\Big|_{s=\alpha}=0

for all n\in\{1,\,2,\,\ldots,\,r-1\} if \alpha\neq1 is the root of s^{\kappa N}=G_{S_N}(s),\,|s|\leqslant1 and root’s multiplicity is r\in\{2,\,3,\,\ldots,\,\kappa N-1\}. Thus, the equality (?) is nothing but the n’th derivative of (?) at s=\alpha.

We now prove equality (?) in Theorem 1. The s derivative of both sides of equation (Proof) is

\begin{aligned} &\hspace{-0.75cm}\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(\kappa s^{\kappa-1}-(i+j)s^{i+j-1}) +\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(\kappa s^{\kappa-1}-(i+j)s^{i+j-1})\\ &\hspace{-0.75cm}+\sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(\kappa s^{\kappa-1}-(i+j)s^{i+j-1})+\ldots\\ &\hspace{-0.75cm}+\sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j(\kappa s^{\kappa-1}-(i+j)s^{i+j-1}) \\ &\hspace{-0.75cm}=G'_{\mathcal{M}_N}(s)\left(s^\kappa-G_{S_N}(s)s^{\kappa(1-N)}\right) +G_{\mathcal{M}_N}(s)\left(\kappa s^{\kappa-1}-G_{S_N}(s)\kappa(1-N)s^{\kappa(1-N)-1}-G'_{S_N}(s)s^{\kappa(1-N)}\right). \end{aligned}

We continue the proof by letting s\to1^- in (?). Because the net profit condition \mathbb{E} S_N<\kappa N holds and the random variable \mathcal{M}_N\buildrel d \over =(\mathcal{M}_1+X_N-\kappa)^+ attains finite values only, we obtain

\begin{aligned} &\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(\kappa -(i+j)) +\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(\kappa -(i+j))\\ &+\sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(\kappa -(i+j))+\ldots +\sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j(\kappa-(i+j)) \\ &=\lim_{s\to1^-}G'_{\mathcal{M}_N}(s)\left(s^\kappa-G_{S_N}(s)s^{\kappa(1-N)}\right) +\kappa N-\mathbb{E} S_N. \end{aligned}

Calculating the limit in (?) there are two separate cases to examine: \mathbb{E} \mathcal{M}_N<\infty and \mathbb{E} \mathcal{M}_N=\infty. If \mathbb{E} \mathcal{M}_N<\infty, then the limit in (?) is zero. However, this limit is zero even if \mathbb{E} \mathcal{M}_N=\infty. Indeed, if \mathbb{E} \mathcal{M}_N=\infty, then by using L’Hospital’s rule we get

\begin{aligned} &\lim_{s\to1^-}G'_{\mathcal{M}_N}(s)\left(s^\kappa-G_{S_N}(s)s^{\kappa(1-N)}\right) =\lim_{s\to1^-}\frac{s^\kappa-G_{S_N}(s)s^{\kappa(1-N)}}{1/G'_{\mathcal{M}_N}(s)} =\lim_{s\to1^-}\frac{\left(s^\kappa-G_{S_N}(s)s^{\kappa(1-N)}\right)'}{\left(1/G'_{\mathcal{M}_N}(s)\right)'}\\ &=\lim_{s\to1^-}\left(\kappa s^{\kappa-1}-G'_{S_N}(s)s^{\kappa(1-N)}-G_{S_N}(s)\kappa(1-N)s^{\kappa(1-N)-1}\right)\frac{(G'_{\mathcal{M}_N}(s))^2} {-G''_{\mathcal{M}_N}(s)}=(\kappa N-\mathbb{E} S_N)\cdot0=0, \end{aligned}

because

\begin{aligned} \lim_{s\to1^-}\frac{(G'_{\mathcal{M}_N}(s))^2}{G''_{\mathcal{M}_N}(s)}=0, \end{aligned}

see [GJS, Lem. 5.5](Proof’s direction of (?) was originally provided by Fedor Petrov.). Thus, the limit in (?) is zero and the equality (?) in Theorem 1 follows.

It remains to prove the equalities in system (?). In short, every equality in system (?) is the corresponding equality from (3) expanded at s=0. Let us demonstrate the derivation of the first equality in (?) in detail and note that the remaining ones are derived analogously. We need to show that the first equality in (3)

\begin{aligned} \sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(s^\kappa-s^{i+j})=s^\kappa G_{\mathcal{M}_N}(s)-G_{X_N}(s)G_{\mathcal{M}_1}(s) \end{aligned}

implies (the first one in (?))

m^{(1)}_nx^{(N)}_0=m^{(N)}_{n-\kappa}-\sum_{i=0}^{n-1}m^{(1)}_ix^{(N)}_{n-i}-\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^1_j\mathbbm{1}_{\{n=\kappa\}}, \,n=\kappa,\,\kappa+1,\,\ldots

or, equivalently,

\begin{aligned} \sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j\mathbbm{1}_{\{n=\kappa\}} = m_{n-\kappa}^{(N)}-\sum_{i=0}^n m_i^{(1)}x_{n-i}^{(N)},\,n=\kappa,\,\kappa+1,\,\ldots \end{aligned}

Equality (?) is implied by (?) because of the following equalities:

\begin{aligned} &\frac{1}{n!}\frac{d^n}{ds^n}\left(\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x_j^{(N)}\left(s^\kappa-s^{i+j}\right)\right)\Bigg|_{s=0}=\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j\mathbbm{1}_{\{n=\kappa\}},\\ &\frac{1}{n!}\frac{d^n}{ds^n}\left(s^\kappa G_{\mathcal{M}_N}(s)\right)\big|_{s=0}=m^{(N)}_{n-\kappa},\\ &\frac{1}{n!}\frac{d^n}{ds^n}\left(G_{\mathcal{M}_1}(s)G_{X_N}(s)\right)\big|_{s=0}=\sum_{i=0}^n m_i^{(1)}x_{n-i}^{(N)}, \end{aligned}

when n=\kappa,\,\kappa+1,\,\ldots The proof of Theorem 1 is finished.

Uses Lemma 3

Theorem

Theorem 2

9JU9

Let us assume that the probabilities m^{(j)}_0,\,m^{(j)}_1,\,\ldots,\,m^{(j)}_{\kappa-1}, j\in\{1,\,2,\,\ldots,\,N\} are known beforehand. Then, the survival probability-generating function \Xi(s), for |s|<1 and s^{\kappa N}\neq G_{S_N}(s) admits the following representation

\begin{aligned} \hspace{-0.5cm} \Xi(s)=\frac{\mathbbm{u}^T\mathbbm{v}}{G_{S_N}(s)-s^{\kappa N}},\textit{ where } \mathbbm{u}=\begin{pmatrix} s^{\kappa(N-1)}\\s^{\kappa(N-2)}G_{X_1}(s)\\s^{\kappa(N-3)}G_{X_1+X_2}(s)\\ \vdots\\s^\kappa G_{X_1+X_2+\ldots+X_{N-2}}(s)\\G_{X_1+X_2+\ldots+X_{N-1}}(s) \end{pmatrix},\, \mathbbm{v}= \begin{pmatrix} &\sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=i}^{\kappa-1}s^j F_{X_1}(j-i)\\ &\sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=i}^{\kappa-1}s^j F_{X_2}(j-i)\\ &\sum_{i=0}^{\kappa-1}m_i^{(4)}\sum_{j=i}^{\kappa-1}s^j F_{X_3}(j-i)\\ &\vdots\\ &\sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=i}^{\kappa-1}s^j F_{X_{N-1}}(j-i)\\ &\sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=i}^{\kappa-1}s^j F_{X_N}(j-i)\\ \end{pmatrix}. \end{aligned}

Proof (Proof of Theorem 2) of Theorem 2

Let us rewrite the system (3) the following way

\begin{aligned} & \begin{pmatrix} s^\kappa&-G_{X_1}(s)&0&\ldots&0&0\\ 0&s^\kappa&-G_{X_2}(s)&\ldots&0&0\\ \vdots&\vdots&\vdots&\ddots&\vdots&\vdots\\ 0&0&0&\ldots&s^\kappa&-G_{X_{N-1}}(s)\\ -G_{X_N}(s)&0&0&\ldots&0&s^\kappa \end{pmatrix} \begin{pmatrix} G_{\mathcal{M}_1}(s)\\ G_{\mathcal{M}_2}(s)\\ \vdots\\ G_{\mathcal{M}_{N-1}}(s)\\ G_{\mathcal{M}_N}(s) \end{pmatrix}\\ &\hspace{8cm}= \begin{pmatrix} \sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(s^\kappa-s^{i+j})\\ \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(s^\kappa-s^{i+j})\\ \vdots\\ \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j(s^\kappa-s^{i+j})\\ \sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(s^\kappa-s^{i+j}) \end{pmatrix} \end{aligned}

and denote this system by AB=C. Determinant of the main matrix in (Proof) is

\det A=s^{\kappa N}-G_{S_N}(s).

Thus, the main matrix in (Proof) is invertible for all such s that s^{\kappa N}\neq G_{S_N}(s) and B=A^{-1}C. Therefore, the previous thoughts and equality (?) imply

\Xi(s)=\frac{G_{\mathcal{M}_1}(s)}{1-s}=\frac{1}{s^{\kappa N}-G_{S_N}(s)} \begin{pmatrix} M_{11}&M_{21}&\ldots&M_{N1} \end{pmatrix} \frac{C}{1-s},

where M_{11},\,M_{21},\,\ldots,\,M_{N1} are the minors of A and C is the column vector of the right-hand-side of (Proof).

Uses Lemma 3

Theorem

Theorem 3

AX97

Suppose that N-seasonal discrete-time risk model (?) is generated by random variables X_1, X_2, \ldots, X_N and the net profit condition is not satisfied. In this case:

If \mathbb{E} S_N > \kappa N, then \varphi(u)=0 for all u\in\mathbb{N}_0.
If \mathbb{E} S_N = \kappa N and \mathbb{P}(S_N=\kappa N)<1, then \varphi(u)=0 for all u\in\mathbb{N}_0.
If \mathbb{P}(S_N=\kappa N)=1, then random variables X_1, X_2, \ldots, X_N are degenerate and
\begin{aligned} &u+\kappa n^*-\sum_{k=1}^{n^*}X_k\leqslant0 \Rightarrow \varphi(u)=0,\\ &u+\kappa n^*-\sum_{k=1}^{n^*}X_k>0 \Rightarrow \varphi(u)=1, \end{aligned}
here n^* is equal to such n\in\{1, 2, \ldots, N\} with which \kappa n-\sum_{k=1}^{n}X_k attains its minimum.

Proof (Proof of Theorem 3) of Theorem 3

We first show that \mathbb{E} S_N > \kappa N implies \varphi(u)=0 for all u\in\mathbb{N}_0. The recurrence (?) yields

\begin{aligned} \varphi(u)&=\sum_{\substack{i_1\leqslant u+\kappa-1\\i_1+i_2\leqslant u+2\kappa-1\\ \vdots \vspace{1mm} \\ i_1+i_2+\ldots +i_{N-1}\leqslant u+\kappa(N-1)-1\\ i_1+i_2+\ldots+i_N\leqslant u+\kappa N-1}}\hspace{-10mm}\mathbb{P}(X_1=i_1)\mathbb{P}(X_2=i_2)\cdots\mathbb{P}(X_N=i_N)\, \varphi\Bigg(u+\kappa N-\sum_{j=1}^Ni_j\Bigg)\\ &=\sum_{i=1}^{u+\kappa N}s^{(N)}_{u+\kappa N -i}\varphi(i)\\ &\quad- \sum_{\substack{i_1\leqslant u+\kappa-1\\i_1+i_2\leqslant u+2\kappa-1\\ \vdots \vspace{1mm} \\ u+\kappa (N-1)\leqslant i_1+i_2+\ldots +i_{N-1}\leqslant u+\kappa N-1\\ i_1+i_2+\ldots+i_N\leqslant u+\kappa N-1}}\hspace{-10mm}x^{(1)}_{i_1}x^{(2)}_{i_2}\cdots x^{(N)}_{i_N}\, \varphi\Bigg(u+\kappa N-\sum_{j=1}^Ni_j\Bigg)\\ &\quad\quad\vdots\\ &\quad- \sum_{\substack{u+\kappa\leqslant i_1\leqslant u+\kappa N-1\\i_1+i_2\leqslant u+\kappa N-1\\ \vdots \vspace{1mm} \\ i_1+i_2+\ldots +i_{N-1}\leqslant u+\kappa N-1\\ i_1+i_2+\ldots+i_N\leqslant u+\kappa N-1}}\hspace{-0mm}x^{(1)}_{i_1}x^{(2)}_{i_2}\cdots x^{(N)}_{i_N}\, \varphi\Bigg(u+\kappa N-\sum_{j=1}^Ni_j\Bigg)\\ &=\sum_{i=1}^{u+\kappa N}s^{(N)}_{u+\kappa N -i}\varphi(i) - \sum_{i=1}^{\kappa(N-1)} \mu_i(u)\varphi(i), \end{aligned}

where \mu_i(u) for each i\in\{1,\,2,\, \ldots,\, \kappa (N-1)\} are coefficients consisting of the products of probability mass functions of random variables X_1, X_2, \ldots, X_N. For instance, if N=2 and \kappa=1, then

\varphi(u)=\sum_{\substack{i_1\leqslant u\\i_1+i_2\leqslant u+1}}x^{(1)}_{i_1}x^{(2)}_{i_2}\varphi(u+2-i_1-i_2)=\sum_{i=1}^{u+2}s^{(2)}_{u+2-i}\varphi(i)-x^{(1)}_{u+1}x^{(2)}_0\varphi(1).

If \mu_0(u):=s^{(N)}_{u+\kappa N} and \mu_j(u):=0 when j>\kappa (N-1), then the equality in (Proof) is

\varphi(u)=\sum_{i=0}^{u+\kappa N}s^{(N)}_{u+\kappa N -i}\varphi(i) - \sum_{i=0}^{\kappa N-1} \mu_i(u)\varphi(i).

By summing up both sides of the last equality by u, which varies from 0 to some natural and sufficiently large v, we obtain

\sum_{u=0}^v\varphi(u)=\sum_{u=0}^v\sum_{i=0}^{u+\kappa N}s^{(N)}_{u+\kappa N -i}\varphi(i) - \sum_{u=0}^v\sum_{i=0}^{\kappa N-1} \mu_i(u)\varphi(i).

We now change the order of summation in (Proof)

\sum_{u=0}^v\sum_{i=0}^{u+\kappa N}(\cdot)=\sum_{i=0}^{\kappa N-1}\sum_{u=0}^v(\cdot)+\sum_{i=\kappa N}^{v+\kappa N}\sum_{u=i-\kappa N}^v(\cdot)

and obtain

\begin{aligned} \sum_{u=0}^{v+\kappa N}\varphi(u)-\sum_{u=v+1}^{v+\kappa N}\varphi(u) =\sum_{i=0}^{\kappa N-1}\varphi(i)\sum_{u=0}^v s^{(N)}_{u+\kappa N -i}+\sum_{i=\kappa N}^{v+\kappa N}\varphi(i)\sum_{u=i-\kappa N}^v s^{(N)}_{u+\kappa N -i} - \sum_{i=0}^{\kappa N-1}\varphi(i)\sum_{u=0}^v \mu_i(u). \end{aligned}

Subtracting \sum_{i=0}^{v+\kappa N}\varphi(i) \sum_{u=i-\kappa N}^v s_{u+\kappa N-i}^{(N)} from both sides of the last equation and rearranging, we get

\begin{aligned} &\sum_{i=0}^{v+\kappa N}\varphi(i)\left(1 -\sum_{u=i-\kappa N}^v s_{u+\kappa N-i}^{(N)}\right)-\sum_{i=v+1}^{v+\kappa N}\varphi(i) \\ &=\sum_{i=0}^{\kappa N-1}\varphi(i)\left(\sum_{u=0}^v s^{(N)}_{u+\kappa N -i}-\sum_{u=0}^v \mu_i(u)\right) -\sum_{i=0}^{\kappa N-1}\varphi(i) \sum_{u=i-\kappa N}^v s_{u+\kappa N-i}^{(N)} \end{aligned}

\begin{aligned} \sum_{i=v+1}^{v+\kappa N}\varphi(i)-\sum_{i=0}^{v+\kappa N}\varphi(i)\left(1 -\sum_{u=0}^{v+\kappa N-i} s_u^{(N)}\right) =\sum_{i=0}^{\kappa N-1}\varphi(i)\left(\sum_{u=0}^{\kappa N -i-1} s^{(N)}_{u}+\sum_{u=0}^{v} \mu_i(u)\right), \end{aligned}

which implies

\begin{aligned}\hspace{-0.5cm} \sum_{i=v+1}^{v+\kappa N}\varphi(i)-\sum_{i=0}^{v+\kappa N}\varphi(i)\mathbb{P}(S_N>v+\kappa N - i)=\sum_{i=0}^{\kappa N-1}\varphi(i)\left(\mathbb{P}(S_N\leqslant \kappa N-i-1)+\sum_{u=0}^{v} \mu_i(u)\right). \end{aligned}

Clearly, the definition of the survival probability (?) implies that \varphi(u) is a non-decreasing function, i. e. \varphi(u)\leqslant\varphi(u+1) for all u\in\mathbb{N}_0. Thus, there exists a non-negative limit \varphi(\infty):=\lim_{u\to\infty}\varphi(u) and \varphi(\infty)=1 if the net profit condition \mathbb{E} S_N<\kappa N holds, see Lemma 1. We now let v\to\infty in both sides of (?). For the first sum in (?) we obtain

\lim_{v\to\infty}\sum_{i=v+1}^{v+\kappa N}\varphi(i)=\lim_{v\to\infty}\left(\varphi(v+1)+\ldots+\varphi(v+\kappa N)\right)=\varphi(\infty) \kappa N,

and for the second

\begin{aligned} \lim_{v\to\infty}\sum_{i=0}^{v+\kappa N}\varphi(i)\mathbb{P}(S_N>v+\kappa N - i)=\varphi(\infty)\mathbb{E} S_N. \end{aligned}

Indeed, let us recall that \mathbb{E} X=\sum_{i=0}^\infty\mathbb{P}(X>i), when X is some non-negative and integer-valued random variable. Then, the upper bound of (?) is

\begin{aligned} &\lim_{v\to\infty}\sum_{i=0}^{v+\kappa N}\varphi(i)\mathbb{P}(S_N>v+\kappa N - i)\leqslant \lim_{v\to\infty} \varphi(v+\kappa N)\sum_{i=0}^{v+\kappa N}\mathbb{P}(S_N>v+\kappa N-i)\\ &=\lim_{v\to\infty}\varphi(v+\kappa N)\sum_{i=0}^{v+\kappa N}\mathbb{P}(S_N>i)=\varphi(\infty)\mathbb{E} S_N, \end{aligned}

while the lower bound is the same due to inequality

\begin{aligned} &\sum_{i=0}^{v+\kappa N}\varphi(i)\mathbb{P}(S_N>v+\kappa N - i)\\ &=\sum_{j=0}^M\varphi(i)\mathbb{P}(S_N>v+\kappa N - i) +\sum_{i=M+1}^{v+\kappa N}\varphi(i)\mathbb{P}(S_N>v+\kappa N - i)\\ &\geqslant \inf_{i\geqslant M+1}\varphi(i)\sum_{i=0}^{v+\kappa N-M-1}\mathbb{P}(S_N>i), \end{aligned}

where M is some fixed and sufficiently large natural number. Thus, when v\to\infty, the equality in (?) is

\begin{aligned} \varphi(\infty)(\kappa N-\mathbb{E} S_N)=\sum_{i=0}^{\kappa N-1}\varphi(i)\left(\mathbb{P}(S_N\leqslant \kappa N-i-1)+\sum_{u=0}^{\infty} \mu_i(u)\right). \end{aligned}

If \mathbb{E} S_N>\kappa N, as under the case of consideration, the non-negative right-hand-side of (?) implies \varphi(\infty)=0 and consequently \varphi(u)=0 for all u\in\mathbb{N}_0. The first case that \mathbb{E} S_N>\kappa N makes survival impossible is proved.

Let us now consider the case when \mathbb{E} S_N = \kappa N and \mathbb{P}(S_N=\kappa N)<1. If \mathbb{E} S_N = \kappa N and \mathbb{P}(S_N=\kappa N)<1, then at least one probability s_0^{(N)},\, s_1^{(N)},\, \ldots,\, s_{\kappa N-1}^{(N)} is larger than zero, because otherwise \mathbb{E} S_N>\kappa N. Then, from (?),

\sum_{i=0}^{\kappa N-1}\varphi(i)\left(\sum_{u=0}^{\kappa N-i-1}s_u^{(N)}+\sum_{u=0}^{\infty} \mu_i(u)\right)=0.

If s_0^{(N)}>0, then (?) implies \varphi(0)=\varphi(1)=\ldots=\varphi(\kappa N-1)=0. Using recurrence (?) we can show that \varphi(u)=0 for all u \in \mathbb{N}_0. If s_0^{(N)}=0 and s_1^{(N)}>0, then (?) implies that \varphi(0)=\varphi(1)=\ldots=\varphi(\kappa N-2)=0 and once again, using recurrence (?) we can show that \varphi(u)=0 for all u \in \mathbb{N}_0. Arguing the same we proceed up to s_0^{(N)}=s_1^{(N)}=\ldots= s_{\kappa N-2}^{(N)}=0,\, s_{\kappa N-1}^{(N)}>0 and observe that in all of these cases (?) and recurrence (?) yields \varphi(u)=0 for all u \in \mathbb{N}_0.

Finally, let us consider the case when \mathbb{P}(S_N=\kappa N)=1. If S_N=\kappa N with probability one, the random variables X_1,\, X_2,\, \ldots,\, X_N are degenerate, meaning that X_i \equiv c_i for all i \in \{1,\, 2,\, \ldots,\, N\}, where c_i \in \{0,\, 1,\,\ldots,\, \kappa N\} and c_1+c_2+\ldots+c_N=\kappa N. Thus, the model (?) becomes fully deterministic. Moreover, W(n)=W(n+N) for all n \in \mathbb{N}_0, N\in\mathbb{N}, and it is sufficient to check if the lowest of values among W(1),\,\ldots,\,W(N) is larger than zero.

Theorem

Theorem 4

4H7Q

For the finite time survival probability (?) of the N-seasonal discrete time risk model defined in (?), holds:

\begin{aligned} &\varphi(u,1)=\sum_{i\leqslant u+\kappa-1}x^{(1)}_{i},\, \varphi(u,2)=\hspace{-2mm}\sum_{\substack{i_1\leqslant u+\kappa-1\\i_1+i_2\leqslant u+2\kappa-1}}\hspace{-2mm}x^{(1)}_{i_1}x^{(2)}_{i_2},\, \ldots,\,\varphi(u,\, N)=\hspace{-6mm}\sum_{\substack{i_1\leqslant u+\kappa-1\\i_1+i_2\leqslant u+2\kappa-1\\ \vdots\\i_1+i_2+\ldots+i_N\leqslant u+\kappa N-1}} \hspace{-8mm}x^{(1)}_{i_1}x^{(2)}_{i_2}\cdots x^{(N)}_{i_N}, \end{aligned}

and

\begin{aligned} \varphi(u,T)=\hspace{-6mm}\sum_ {\substack{i_1\leqslant u+\kappa-1\\i_1+i_2\leqslant u+2\kappa-1\\ \vdots\\i_1+i_2+\ldots+i_N\leqslant u+\kappa N-1}} \hspace{-8mm}x^{(1)}_{i_1}x^{(2)}_{i_2}\cdots x^{(N)}_{i_N}\varphi\left(u+\kappa N-i_1-\ldots-i_N,\,T- N\right), \end{aligned}

if T\in\{ N+1,\, N+2,\,\ldots\}.

Moreover, for the defined probabilities (?), it holds that

\begin{aligned} &\varphi^{(j)}(u,\,1)=F_{X_j}(u+\kappa-1), \\ &\varphi^{(j)}(u,\,T)=\sum_{i=0}^{u+\kappa-1}\varphi^{(j+1)}(u+\kappa-i,\,T-1)x^{(j)}_i,\,T=2,\,3,\,\ldots \end{aligned}

Proof (Proof of Theorem 4) of Theorem 4

The proof of equalities (4) and (4) is basically the same as deriving the recurrence (?). Equalities (?) and (?) are implied by [BBS, Thm. 1].

Conjecture

Conjecture 1

EYNT

Assume that s^{(N)}_0=\mathbb{P}(X_1+X_2+\ldots+X_N=0)>0 and \alpha_1,\,\alpha_2,\,\ldots,\,\alpha_{\kappa N-1}\neq1 are the simple roots of s^{\kappa N}=G_{S_N}(s). Then, the system’s matrix in (?) is non-singular for all \kappa,\,N\in\mathbb{N}. In particular, if N=3 and \kappa=1, then

\frac{G_{S_3}(\alpha_1)}{\alpha_1^3}=\frac{G_{S_3}(\alpha_2)}{\alpha_2^3}=1

implies

\left(\frac{G_{X_3}(\alpha_1)}{\alpha_1}-1\right) \left(\frac{G_{X_3+X_1}(\alpha_2)}{\alpha_2^2}-1\right) \neq \left(\frac{G_{X_3}(\alpha_2)}{\alpha_2}-1 \right) \left(\frac{G_{X_3+X_1}(\alpha_1)}{\alpha_1^2}-1\right)

and consequently \det A\neq0, where \alpha_1,\,\alpha_2\neq1 locate in |s|\leqslant1.

Lemma

Lemma 1

UV7V

If the net profit condition is satisfied, i. e. \mathbb{E} S_N<\kappa N, then

\lim_{u\to\infty}\mathbb{P}(\mathcal{M}_j<u)=1,

for all j\in\{1,\,2,\,\ldots,\,N\}.

Used by Proof

Proof of Lemma 1

We prove the case j=1 only and note that the other cases can be proven similarly.

According to the law of large numbers, almost surely

\begin{aligned} \frac{1}{n}\sum_{i=1}^{n}\left(X_{i}-\kappa\right)&=\frac{1}{N}\left(\frac{N}{n}\sum_{\substack{{i=1}\\ i\equiv 1\,{\rm mod}\,N}}^{n}\left(X_{i}-\kappa\right) +\ldots+\frac{N}{n}\sum_{\substack{{i=1}\\ i\equiv N\,{\rm mod}\,N}}^{n}\left(X_{i}-\kappa\right)\right)\\ &\overset{n \to \infty}{\longrightarrow} \frac{1}{N}\left((\mathbb{E}X_1-\kappa)+\ldots+(\mathbb{E}X_N-\kappa)\right) =\frac{\mathbb{E}S_N-\kappa N}{N}=:-\mu<0. \end{aligned}

Therefore,

\begin{aligned} \lim_{n\to\infty}\mathbb{P}\left(\sup_{j\geqslant n}\left|\frac{1}{j}\sum_{i=1}^{j}\left(X_i-\kappa\right)+\mu\right| < \frac{\mu}{2}\right) = 1. \end{aligned}

Consequently, for any arbitrarily small \varepsilon>0, there exists such number N_\varepsilon\in\mathbb{N} that

\begin{aligned} &\mathbb{P}\left(\bigcap_{j=n}^{\infty}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)\leqslant 0\right\}\right) \geqslant \mathbb{P}\left(\bigcap_{j=n}^{\infty}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)< -\frac{\mu}{2}\right\}\right)\\ &\geqslant\mathbb{P}\left(\bigcap_{j= n}^{\infty}\left\{\left|\frac{1}{j}\sum_{i=1}^{j}\left(X_i-\kappa\right)+\mu\right| < \frac{\mu}{2}\right\}\right) = \mathbb{P}\left(\sup_{j\geqslant n}\left|\frac{1}{j}\sum_{i=1}^{j}\left(X_i-\kappa\right)+\mu\right| < \frac{\mu}{2}\right) \geqslant 1-\varepsilon \end{aligned}

for all n \geqslant N_\varepsilon.

It follows that for any such \varepsilon and any u\in\mathbb{N} we have

\begin{aligned} \mathbb{P}(\mathcal{M}_1< u)&=\mathbb{P}\left(\bigcap_{j=1}^{\infty}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)<u\right\}\right)\\ &\geqslant\mathbb{P}\left(\left\{\bigcap_{j=1}^{N_\varepsilon-1}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)<u\right\}\right\}\cap\left\{\bigcap_{j=N_\varepsilon}^{\infty}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)\leqslant0\right\}\right\}\right)\\ &\geqslant\mathbb{P}\left(\bigcap_{j=1}^{N_\varepsilon-1}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)<u\right\}\right) +\mathbb{P}\left(\bigcap_{j=N_\varepsilon}^{\infty}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)\leqslant0\right\}\right)-1\\ &\geqslant\mathbb{P}\left(\bigcap_{j=1}^{N_\varepsilon-1}\left\{\sum_{i=1}^{j}\left(X_i-\kappa\right)<u\right\}\right)-\varepsilon. \end{aligned}

The last inequality implies

\lim_{u \to \infty}\mathbb{P}(\mathcal{M}_1<u)\geqslant 1-\varepsilon,

where \varepsilon>0 is as small as we want and the assertion of Lemma follows.

Lemma

Lemma 2

24OC

If the net profit condition is satisfied, i. e. \mathbb{E} S_N<\kappa N, it holds that (\mathcal{M}_j+X_{j-1}-\kappa)^+\buildrel d \over = \mathcal{M}_{j-1}, for all j=2,\,3,\,\ldots,\,N, and (\mathcal{M}_1+X_N-\kappa)^+\buildrel d \over = \mathcal{M}_N.

Used by Proof

Proof of Lemma 2

We prove the equality (\mathcal{M}_1+X_N-\kappa)^+\buildrel d \over = \mathcal{M}_N only and note that the other ones can be proved by the same arguments. According to Lemma 1, the random variable \mathcal{M}_1 is not extended, i. e. it does not attain infinity. Let us denote \tilde{X}_j=X_j-\kappa for all j\in\{1,\,2,\,\ldots,\,N\}. Then

\begin{aligned} &(\mathcal{M}_1+\tilde{X}_N)^+\\ &\buildrel d \over =(\max\{0,\,\max\{\tilde{X}_{1},\,\tilde{X}_{1}+\tilde{X}_{2},\,\tilde{X}_{1}+\tilde{X}_{2}+\tilde{X}_{3},\,\ldots\}\}+\tilde{X}_N)^+\\ &\buildrel d \over =(\max\{0,\,\tilde{X}_{1},\,\tilde{X}_{1}+\tilde{X}_{2},\,\tilde{X}_{1}+\tilde{X}_{2}+\tilde{X}_{3},\,\ldots\}+\tilde{X}_N)^+\\ &\buildrel d \over =(\max\{\tilde{X}_{N},\,\tilde{X}_{N}+\tilde{X}_{1},\,\tilde{X}_{N}+\tilde{X}_{1}+\tilde{X}_{2},\,\tilde{X}_{N}+\tilde{X}_{1}+\tilde{X}_{2}+\tilde{X}_3,\,\ldots\})^+\\ &\buildrel d \over =(\max\{\tilde{X}_{N},\,\tilde{X}_{N}+\tilde{X}_{N+1},\,\tilde{X}_{N}+\tilde{X}_{N+1}+\tilde{X}_{N+2},\,\tilde{X}_{N}+\tilde{X}_{N+1}+\tilde{X}_{N+2}+\tilde{X}_{N+3},\,\ldots\})^+\\ &\buildrel d \over =\max\{0,\max\{\tilde{X}_{N},\,\tilde{X}_{N}+\tilde{X}_{N+1},\,\tilde{X}_{N}+\tilde{X}_{N+1}+\tilde{X}_{N+2},\,\ldots\}\}\\ &\buildrel d \over = \mathcal{M}_N^+. \end{aligned}

Uses Lemma 1

Lemma

Lemma 3

43DI

Let s\in\mathbb{C} be the complex number and 0<|s|\leqslant1. Then the probability-generating functions of X_1,\,X_2,\,\ldots,\,X_N and \mathcal{M}_1,\,\mathcal{M}_2\ldots,\,\mathcal{M}_N are related the following way

\begin{aligned} \begin{cases} \sum_{i=0}^{\kappa-1}m_i^{(1)}\sum_{j=0}^{\kappa-1-i}x^{(N)}_j(s^\kappa-s^{i+j})&=s^\kappa G_{\mathcal{M}_N}(s)-G_{X_N}(s)G_{\mathcal{M}_1}(s)\\ \sum_{i=0}^{\kappa-1}m_i^{(2)}\sum_{j=0}^{\kappa-1-i}x^{(1)}_j(s^\kappa-s^{i+j})&=s^\kappa G_{\mathcal{M}_1}(s)-G_{X_1}(s)G_{\mathcal{M}_2}(s)\\ \sum_{i=0}^{\kappa-1}m_i^{(3)}\sum_{j=0}^{\kappa-1-i}x^{(2)}_j(s^\kappa-s^{i+j})&=s^\kappa G_{\mathcal{M}_2}(s)-G_{X_2}(s)G_{\mathcal{M}_3}(s)\\ &\,\,\vdots\\ \sum_{i=0}^{\kappa-1}m_i^{(N)}\sum_{j=0}^{\kappa-1-i}x^{(N-1)}_j(s^\kappa-s^{i+j})&=s^\kappa G_{\mathcal{M}_{N-1}}(s)-G_{X_{N-1}}(s)G_{\mathcal{M}_N}(s) \end{cases}. \end{aligned}

Used by Proof, Proof

Proof of Lemma 3

Let us demonstrate how the first equality in (3) is derived and note that the remaining ones follow the same logic.

By Lemma 2, the distributions’ equality (\mathcal{M}_1+X_{N}-\kappa)^+\buildrel d \over = \mathcal{M}_{N} implies the equality of probability-generating functions G_{\mathcal{M}_N}(s)=G_{(\mathcal{M}_1+X_{N}-\kappa)^+}(s). Then, applying the law of total expectation for the last equality, we obtain

\begin{aligned} G_{\mathcal{M}_N}(s) &=\mathbb{E} s^{(\mathcal{M}_1+X_{N}-\kappa)^+}=\mathbb{E} \left( \mathbb{E} \left( s^{(\mathcal{M}_1+X_{N}-\kappa)^+}|\mathcal{M}_1\right) \right) =\sum_{i=0}^{\infty}m^{(1)}_i\mathbb{E} s^{(X_N-\kappa+i)^+}\\ &=\sum_{i=0}^{\kappa-1}m^{(1)}_i\mathbb{E}s^{(X_N-\kappa+i)^+} +G_{X_N}(s)s^{-\kappa}\sum_{i=\kappa}^{\infty}m^{(1)}_is^i\\ &=\sum_{i=0}^{\kappa-1}m^{(1)}_i\left(\mathbb{E}s^{\left(X_N-\kappa+i\right)^+} -s^{i-\kappa}G_{X_N}(s)\right)+s^{-\kappa}G_{X_N}(s)G_{\mathcal{M}_1}(s). \end{aligned}

Multiplying both sides of the last equality by s^\kappa when s\neq0 and observing that

\begin{aligned} \sum_{i=0}^{\kappa-1}m^{(1)}_i\left(s^\kappa\mathbb{E} s^{(X_N-\kappa+i)^+}-s^iG_{X_N}(s)\right)=\sum_{i=0}^{\kappa-1}m^{(1)}_i \sum_{j=0}^{\kappa-1-i}x_j^{(N)}(s^\kappa-s^{i+j}) \end{aligned}

we get the desired result.

Uses Lemma 2

Lemma

Lemma 4

MHNQ

Assume that the net profit condition G'_{S_N}(1)=\mathbb{E} S_N<\kappa N is valid. Then there are exactly \kappa N-1 roots, counted with their multiplicities, of s^{\kappa N}=G_{S_N}(s) in |s|\leqslant1, s\neq1.

Proof of Lemma 4

We follow the proof of [GJ, Lemma 9]. Due to the estimate

|G_{S_N}(s)|\leqslant1<\lambda |s|^{\kappa N}

on |s|=1 when \lambda>1, Rouché’s theorem implies that both functions G_{S_N}(s)-\lambda s^{\kappa N} and \lambda s^{\kappa N} have the same number of roots in |s|<1 and this number is \kappa N due to the fundamental theorem of algebra. When \lambda\to1^+ some roots of G_{S_N}(s)-\lambda s^{\kappa N} remain in |s|<1 and some migrate to the boundary points |s|=1. Obviously, s=1 is always the root of s^{\kappa N}=G_{S_N}(s) and it is the simple root because the net profit condition holds, i. e.

\left(G_{S_N}(s)-s^{\kappa N}\right)'\Big|_{s=1}=\mathbb{E} S_N-\kappa N<0.

Thus, there remain \kappa N-1 roots of s^{\kappa N}=G_{S_N}(s) in |s|\leqslant1, s\neq1 and additionally one can say that |s|=1 , s \neq 1 is the root of s^{\kappa N}=G_{S_N}(s) if the greatest common divisor of \kappa N and the powers of s in G_{S_N}(s) is greater than one.

Example

Example 1

XSWM

Let \kappa=2 and consider the bi-seasonal (N=2) discrete time risk model (?) where X_1\sim\mathcal{P}(1, 0) and X_2\sim\mathcal{P}(2, 0). We set up the survival probability-generating function \Xi(s) and calculate \varphi(u), when u=0,\,1,\,\ldots,\,15.

Example

Example 2

413O

Let us consider the model (?) when \kappa=2 and X_1\sim\mathcal{P}(1, 1) and X_2\sim\mathcal{P}(9/10, 1). We find \varphi(u) when u=0,\,1,\,\ldots,\,50 and set up the survival probability-generating function \Xi(s).

Example

Example 3

GJJ0

Let us consider the bi-seasonal model (?) with \kappa=3 where claims are represented by two independent random variables X_1 and X_2, whose distributions are given in Table 3 and Table3.

The paper has a table here, which this page does not show.

We find the survival probability \varphi(u) for all u\in\mathbb{N}_0 and its generating function \Xi(s).

Example

Example 4

VSDZ

In the last example, we consider ten season model with a premium rate of 5, i. e. N=10, \kappa=5, and we assume claims to be generated by independent random variables X_k\sim\mathcal{P}(k/(k+1)+4,\,0), k\in\{1,\, 2,\,\ldots,\, 10\}, where \mathcal{P}(\lambda,\,0) denotes Poisson distribution with parameter \lambda. We calculate both the finite time survival probability \varphi(u, T) and the ultimate time survival probability \varphi(u) and provide a frame of ultimate time survival probability-generating function \Xi(s).

Source and licence

The text on this page is from arXiv:2211.14629v1 by Simonas Gervė and Andrius Grigutis, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/2211/2211.14629.jsonl.