MPEMathematical Problems in Engineering1563-51471024-123XHindawi Publishing Corporation85285210.1155/2011/852852852852Research ArticleApproximation for the Finite-Time Ruin Probability of a General Risk Model with Constant Interest Rate and Extended Negatively Dependent Heavy-Tailed ClaimsYangYang1,2MaXin3LinJin-guan2LiatsisP.1School of Mathematics and StatisticsNanjing Audit UniversityNanjing 210029Chinanau.edu.cn2Department of MathematicsSoutheast UniversityNanjing 210096Chinaseu.edu.cn3Golden Audit CollegeNanjing Audit UniversityNanjing 210029Chinanau.edu.cn20111306201120110603201103052011090520112011Copyright © 2011 Yang Yang et al.This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

We propose a general continuous-time risk model with a constant interest rate. In this model, claims arrive according to an arbitrary counting process, while their sizes have dominantly varying tails and fulfill an extended negative dependence structure. We obtain an asymptotic formula for the finite-time ruin probability, which extends a corresponding result of Wang (2008).

1. The Dependent General Risk Model

In this paper, we consider the finite-time ruin probability with constant interest rate in a dependent general risk model. In this model, the claim sizes {Xn,n1} form a sequence of identically distributed, not necessarily independent, and nonnegative random variables (r.v.s) with common distribution F such that F¯(x)=1-F(x)=P(X1>x)>0 for all x>0; the claim arrival process {N(t),t0} is a general counting process, namely, a nonnegative, nondecreasing, right continuous, and integer-valued stochastic process with 0<EN(t)=λ(t)< for all large t>0. The times of the successive claims are denoted by {τn,n1}. The total amount of premiums accumulated up to time t0, denoted by C(t) with C(0)=0 and C(t)< almost surely for every t>0, is another nonnegative and nondecreasing stochastic process. Assume that {Xn,n1}, {N(t),t0} and {C(t),t0} are mutually independent. Let δ>0 be the constant interest rate (i.e., after time t one dollar becomes eδt dollars), and let x0 be the initial capital reserve of an insurance company. Then, the total discounted reserve up to time t0, denoted by D(t,x), can be written as D(t,x)=x+0te-δsC(ds)-n=1N(t)Xne-δτn. For a finite time T>0, the finite-time ruin probability is defined by Ψ(x,T)=P(D(t,x)<0,  for  some  0tT)=P(supt[0,T](n=1N(t)Xne-δτn-0te-δsC(ds))>x), while the ultimate ruin probability is defined by Ψ(x)=Ψ(x,)=P(D(t,x)<0,  for  some  t0).

If the claim sizes {Xn,n1} are independent r.v.s, the model is called the independent general risk model, which was introduced by Wang . In particular, if C(t)=ct, t0, with c>0 a deterministic constant and {N(t),t0} is a Poisson process, then the model reduces to the classical one.

2. Introduction and Main Result

Hereafter, all limit relationships hold for x tending to unless otherwise stated. For two positive functions f(x) and g(x), we write f(x)~g(x) if limf(x)/g(x)=1; write f(x)g(x) if limsupf(x)/g(x)1 and f(x)=o(g(x)) if limf(x)/g(x)=0. The indicator function of an event A is denoted by 1A.

In risk theory, heavy-tailed distributions are often used to model large claim amounts. They play a key role in insurance and finance. We will restrict the claim-size distribution F to be heavy tailed. A distribution V is said to be dominatedly varying tailed, denoted by V𝒟, if limsupV¯(xy)/V¯(x)< for any y>0. A distribution V is said to be long tailed, denoted by V, if limV¯(x+y)/V¯(x)=1 for any y>0. A distribution V is said to be subexponential, denoted by V𝒮, if Vn*¯(x)~nV¯(x) for any n2, where Vn* denotes the n-fold convolution of itself. A distribution V is said to be regularly varying tailed, denoted by -α,  α>0, if limV¯(xy)/V¯(x)=y-α for any y1. A proper inclusion relationship holds that R-αLDSL, see, for example, Cline  or Embrechts and Omey . For a distribution V, denote the upper Matuszewska index of the distribution V by JV+=-limylogV¯*(y)logywith  V¯*(y)=liminfxV¯(xy)V¯(x),y>1. In the terminology of Bingham et al. , the quantity JV+ is actually the upper Matuszewska index of the function 1/V¯(x), x0, as also pointed out in Tang and Tsitsiashvili . Additionally, denote LV=limy1V¯*(y) (clearly, 0LV1 ). The presented definitions yield that the following assertions are equivalent:   (i)  VD,(ii)  V¯*(y)>0for  some  y>1,(iii)  LV>0,(iv)  JV+<.

The asymptotic behavior of the ruin probability in the classical risk model has been extensively investigated in the literature. Klüppelberg and Stadtmüller  considered the ultimate ruin probability for the case of regularly-varying-tailed claim sizes. Using the reflected random walk theory, Asmussen  extended the study to a larger class of heavy-tailed distributions; see Corollary  4.1(ii) of his paper. Complementary discussions on the ultimate ruin probability can be found in Kalashnikov and Konstantinides , Konstantinides et al. , Tang , among others.

In this paper, we are interested in the finite-time ruin probability. In this aspect, Tang  established an asymptotic result in the classical risk model: under the condition F𝒮, he obtained that for every T>0 for which λ(T)>0, Ψ(x,T)~0-TF¯(xeδt)λ(dt). Recently, Wang  derived some important and interesting results in two independent risk models. One is the delayed renewal risk model, in which (2.4) holds if F𝒮; another is the general risk model, in which (2.4) also holds if F𝒟. We are interested in the latter, for example, the general risk model, and restate Theorem   2.2 of Wang  here.

Theorem 2.1.

In the independent general risk model introduced in Section 1, assume that the claim sizes {Xn,n1} are independent and identically distributed nonnegative r.v.s with common distribution F𝒟. Assume that for any T>0 with λ(T)-λ(0)>0, there exists some constant η=η(T)>0 such that E(1+η)N(T)<. Then, (2.4) holds.

In the present paper, we aim to deal with the extended negatively dependent general risk model to get a similar result under F𝒟. Simultaneously, the condition (2.5) can be weakened to (2.8) below.

We call r.v.s {ξn,n1} are extended negatively dependent (END) if there exists some positive constant M such that both P(k=1n{ξk>yk})Mk=1nP(ξk>yk),P(k=1n{ξkyk})Mk=1nP(ξkyk) hold for each n1 and all y1,,yn. This dependence structure was introduced by Liu . Recall that r.v.s {ξn,n1} are called upper negatively dependent (UND) if (2.6) holds with M=1, they are called lower negatively dependent (LND) if (2.7) holds with M=1, and they are called negatively dependent (ND) if both (2.6) and (2.7) hold with M=1. These negative dependence structures were introduced by Ebrahimi and Ghosh  and Block et al. . Clearly, ND r.v.s must be END r.v.s., and Example  4.1 of Liu  shows that the END structure also includes some other dependence structures.

Motivated by the work of Wang , under the END structure, we formulate our main result as follows.

Theorem 2.2.

In the dependent general risk model introduced in Section 1, assume that the claim sizes {Xn,n1} are END nonnegative r.v.s with common distribution F𝒟 and finite mean μ. Assume that for any T>0 with λ(T)-λ(0)>0, there exists some constant p>JF+ such that E(N(T))p<. Then, it holds that LF0-TF¯(xeδt)λ(dt)Ψ(x,T)LF-10-TF¯(xeδt)λ(dt). Furthermore, if F𝒟, then (2.4) holds.

The rest of the present paper consists of two sections. We give some lemmas and the proof of Theorem 2.2 in Section 3. In Section 4, we perform some numerical calculations to verify the approximate relationship in our main result.

3. Proof of Main Result and Some Lemmas

In the sequel, M and a always represent some finite and positive constants whose values may vary in different places. In this section, we start by giving some lemmas to show some properties of the class 𝒟 and the END structure. The first lemma is a combination of Proposition  2.2.1 of Bingham et al.  and Lemma  3.5 of Tang and Tsitsiashvili .

Lemma 3.1.

If a distribution V𝒟, then

for any γ>JV+, there exist positive constants a and b such that V¯(y)/V¯(x)a(y/x  )-γ holds for all xyb and

it holds for every γ>JV+ that x-γ=o(V¯(x)).

By direct verification, END r.v.s have the following properties similar to those of ND r.v.s; see Lemma  3.1 of Liu . For some refined properties of END r.v.s, one can refer to Chen et al. . The following lemma can also be found in Lemma  2.2 of Chen et al. .

Lemma 3.2.

(i) If r.v.s {ξn,n1} are nonnegative and END, then for any n1, there exists some positive constant M such that E(k=1nξk)Mk=1nEξk.

(ii) If r.v.s {ξn,n1} are END and {fn(·),n1} are either all monotone increasing or all monotone decreasing, then {fn(ξn),n1} are still END.

The following two lemmas play important roles in the proof of our main result.

Lemma 3.3.

Let {ξn,n1} be identically distributed and END r.v.s with common distribution V and μV+=Eξ11{ξ10}<. Then, for any θ>0, x>0 and n1, there exists some positive constant M such that P(k=1nξk>x)nV¯(θx)+M(eμV+nx)θ-1.

Proof.

Following the proof of Lemma  2.3 of Tang , we employ a standard truncation argument to prove this lemma. For simplicity, we write Snξ=k=1nξk, n1. If μV+=0, then ξn is almost surely nonpositive for each n1, implying P(Snξ>x)=0 for any positive x, and thus (3.1) holds.

Let, in the following, μV+>0. For any fixed θ>0 and positive integer n, define ξ̃n=min{ξn,θx},ξ̃n+=max{ξ̃n,0}=ξn1{0ξnθx}+θx1{ξn>θx}. According to Lemma 3.2(ii), {ξ̃n,n1} and {ξ̃n+,n1} are still END r.v.s, respectively. Denote S̃nξ=k=1nξ̃k, n1. Clearly, P(Snξ>x)=P(Snξ>x,max1knξk>θx)+P(Snξ>x,max1knξkθx)nV¯(θx)+P(S̃nξ>x). It remains to estimate the second summand in (3.3). For a positive h, by Lemma 3.2(ii), {ehξ̃n+,n1} are END nonnegative r.v.s. Hence, using identity Eehξ̃1+=0θx(ehu-1)V(du)+(ehθx-1)V¯(θx)+1, by Markov inequality and Lemma 3.2(i) we have P(S̃nξ>x)e-hxEehS̃nξe-hxEehk=1nξ̃k+e-hxM(Eehξ̃1+)n=Me-hx(0θx(ehu-1)V(du)+(ehθx-1)V¯(θx)+1)n. Since 1+ueu for all u and (ehu-1)/u is strictly increasing in u>0, from (3.5), we obtain P(S̃nξ>x)Mexp{n0θxehu-1uuV(du)+n(ehθx-1)V¯(θx)-hx}Mexp{n    ehθx-1θx(0θxuV(du)+θxV¯(θx))-hx}Mexp{n    ehθx-1θxμV+-hx}. Choose h=(θx)-1log(x(μV+n)-1+1), which is positive. For such h, by (3.6), we have P(S̃nξ>x)Mexp{1θ-1θlog(xμV+n+1)}Mexp{1θlogeμV+nx}. The last estimate and (3.3) imply the desired estimate (3.1). The lemma is proved.

Lemma 3.4.

In the dependent general risk model introduced in Section 1, assume that the claim sizes {Xn,n1} are END nonnegative r.v.s with common distribution F𝒟. Let Z be an arbitrary nonnegative r.v. and assume that {Xn,n1}, {N(t),t0} and Z are mutually independent. Then, for any T>0 and any positive integer n0, LFk=1n0j=1kP(Xje-δτj>x,N(T)=k)k=1n0P(j=1kXje-δτj>x+Z,N(T)=k)LF-1k=1n0j=1kP(Xje-δτj>x,N(T)=k). Furthermore, if F𝒟, then k=1n0P(j=1kXje-δτj>x+Z,N(T)=k)k=1n0j=1kP(Xje-δτj>x,N(T)=k).

We remark that if F is consistently varying tailed (see the definition in Chen and Yuen ), then by conditioning (3.9) easily follows from Theorem  3.2 of Chen and Yuen . Note that this case is in a broader scope, since there is no need to assume independence between (τ1,,τn0) and Z.

Proof.

We follow the line of the proof of Lemma  3.6 of Wang  with some modifications in relation to the properties of the class 𝒟 and the END structure. Clearly, for each k=1,,n0, P(j=1kXje-δτj>x+Z,N(T)=k)={0t1tkT,tk+1>T}0-P(j=1kXje-δtj>x+z)×P(Zdz,τ1dt1,,τk+1dtk+1). We first show the upper bound. For any fixed l>0, P(j=1kXje-δtj>x+z)P(j=1k{Xje-δtj>x+z-l})+P(j=1kXje-δtj>x+z,max1jkXje-δtjx+z-l):=I1+I2. By F𝒟, for any 0<θ<1 and each k=1,,n0, we have uniformly for all t1,,tk[0,T] and z[0,), I1j=1kF¯(θ(x+z)eδtj)LF-1j=1kF¯((x+z)eδtj), by firstly letting x then θ1. We note that {Xn,n1} are END r.v.s. Then, by F𝒟, there exists some positive constant M=M(n0) such that for sufficiently large x, each k=1,,n0, all t1,,tk[0,T] and z[0,), I2=P(j=1kXje-δtj>x+z,x+zk<max1jkXje-δtjx+z-l)P(i=1k{jiXje-δtj>l,Xie-δti>x+zk})i=1kjiP(Xje-δtj>lk-1,Xie-δti>x+zk)Mi=1kjiF¯(leδtjk-1)F¯((x+z)eδtik)MF¯(ln0-1)j=1kF¯((x+z)eδtj). Since l can be arbitrarily large, it follows that limsupllimsupxsupt1,,tk[0,T],z[0,)I2j=1kF¯((x+z)eδtj)=0. Hence, from (3.10)–(3.14), we obtain for each k=1,,n0, P(j=1kXje-δτj>x+Z,N(T)=k)LF-1j=1k{0t1tkT,tk+1>T}0-F¯((x+z)eδtj)×P(Zdz,τ1dt1,,τk+1dtk+1)=LF-1j=1kP(Xje-δτj>x+Z,N(T)=k)LF-1j=1kP(Xje-δτj>x,N(T)=k). As for the lower bound for (3.10), since {Xn,n1} are END r.v.s, we have for sufficiently large x and each k=1,,n0, P(j=1kXje-δtj>x+z)P(j=1k{Xje-δtj>x+z})j=1kF¯((x+z)eδtj)-1i<jkP(Xie-δti>x+z,Xje-δtj>x+z)j=1kF¯((x+z)eδtj)-M1i<jkF¯((x+z)eδti)F¯((x+z)eδtj)=(1-o(1))j=1kF¯((x+z)ertj) holds uniformly for all t1,,tk[0,T] and z[0,). By F𝒟 and Fatou's lemma, we have for any θ̃>1 and all j=1,2,, liminf1F¯(x)P(Xj>x+ZeδT)=liminf0-F¯(x+zeδT)F¯(x)P(Zdz)0-liminfF¯(θ̃x)F¯(x)P(Zdz)=F*¯(θ̃)LF,θ̃1, which means P(Xj>x+ZeδT)LFF¯(x). Similar to (3.15), from (3.10), (3.16), and (3.18), we obtain for each k=1,,n0, P(j=1kXje-δτj>x+Z,N(T)=k)j=1kP(Xje-δτj>x+Z,N(T)=k)j=1k{0t1tkT,tk+1>T}P(Xj>xeδtj+ZeδT)P(τ1dt1,,τk+1dtk+1)LFj=1k{0t1tkT,tk+1>T}F¯(xeδtj)P(τ1dt1,,τk+1dtk+1)=LFj=1kP(Xje-δτj>x,N(T)=k). The desired relation (3.8) follows now from (3.15) and (3.19).

If F𝒟, (3.9) follows by using the properties of the class to establish analogies of relations (3.12) and (3.17). This ends the proof of the lemma.

Proof of Theorem <xref ref-type="statement" rid="thm2.2">2.2</xref>.

We use the idea in the proof of Theorem  2.2 of Wang  (e.g., Theorem 2.1 of this paper) to prove this result. Clearly, F𝒟 and μ< imply JF+1. By (2.8), we have for any ϵ>0, there exists some positive integer n1=n1(T,ϵ) such that E(N(T))p1{N(T)>n1}ϵ.

To estimate the upper bound of Ψ(x,T), we split it into two parts as Ψ(x,T)P(j=1N(T)Xje-δτj>x)=(k=1n1+k=n1+1)P(j=1kXje-δτj>x,N(T)=k):=I3+I4. According to Lemma 3.4 of this paper and Lemma  3.5 of Wang , we have for sufficiently large x, I3(1+ϵ)LF-1k=1n1j=1kP(Xje-δτj>x,N(T)=k)(1+ϵ)LF-1j=1P(Xje-δτj>x,N(T)j)=(1+ϵ)LF-1j=1P(Xje-δτj>x,τjT)=(1+ϵ)LF-10-TF¯(xeδt)λ(dt). By Lemma 3.3, F𝒟, Lemma 3.1(ii), (3.20), and p>JF+1, there exists some positive constant M such that for sufficiently large x, I4k=n1+1P(j=1kXj>x)P(N(T)=k)F¯(p-1x)k=n1+1kP(N(T)=k)+M(eμ)px-pk=n1+1kpP(N(T)=k)MF¯(x)(EN(T)1{N(T)>n1}+E(N(t))p1{N(T)>n1})=MϵF¯(x). By Lemma 3.1(i), for any γ>JF+, there exists some positive constant a such that for sufficiently large x, 0-TF¯(xeδt)λ(dt)a-1F¯(x)0-Te-γδtλ(dt)a-1e-γδT(λ(T)-λ(0))F¯(x), which, combining (3.23) and λ(T)-λ(0)>0, implies I4Mϵ0-TF¯(xeδt)λ(dt). From (3.21), (3.22), and (3.25), we derive the right-hand side of (2.9).

As for the lower bound of Ψ(x,T), by Lemma 3.4, we have for the above given ϵ>0 and sufficiently large x, Ψ(x,T)P(j=1N(T)Xje-δτj>x+0Te-δsC(ds))k=1n1P(j=1kXje-δτj>x+0Te-δsC(ds),N(T)=k)(1-ϵ)LFk=1n1j=1kP(Xje-δτj>x,N(T)=k)=(1-ϵ)LF(j=1P(Xje-δτj>x,τjT)-k=n1+1j=1kP(Xje-δτj>x,N(T)=k)):=(1-ϵ)LF(0-TF¯(xeδt)λ(dt)-I5). Analogously to the estimate for I4, we have for sufficiently large x, I5F¯(x)EN(T)1{N(T)>n1}Mϵ0-TF¯(xeδt)λ(dt). From (3.26) and (3.27), we obtain the left-hand side of (2.9).

If F𝒟, then (2.4) follows by using (3.9) in the proof of (3.22) and (3.26).

4. Numerical Calculations

In this section, we perform some numerical calculations to check the accuracy of the asymptotic relations obtained in Theorem 2.2. The main work is to estimate the finite-time ruin probability defined in (1.2).

We assume that the claim sizes {Xn,n1} come from the common Pareto distribution with parameter κ=1, β=2, F(x;κ,β)=1-(κκ+x)β,x0, which belongs to the class 𝒟, and {(X2n-1,X2n),n1} are independent replications of (X1,X2) with the joint distribution FX1,X2(x,y)=-1αln(1+(e-αF(x)-1)(e-αF(y)-1)e-α-1), with parameter α=1, where the joint distribution FX1,X2(x,y) is constructed according to the Frank Copula. It has been proved in Example  4.2 of Liu  that X1 and X2 are END r.v.s. Since {(X2n-1,X2n),n1} are independent copies of (X1,X2), the r.v.s {Xn,n1} are END as well.

Assume that the claim arrival process N(t) is the homogeneous Poisson process with intensity parameter λ. Clearly, such an integer-valued process N(t) satisfies the condition (2.8). Choose λ=0.1. The total amount of premiums is simplified as C(t)=ct with the premium rate c=500, and the constant interest rate δ=0.02. Here, we set the time T as T=10 and the initial capital reserve x=500,103,2×103,5×103, respectively. We aim to verify the accuracy of relation (2.4). The procedure of the computation of the finite-time ruin probability Ψ(x,T) in Theorem 2.2 is listed here.

Step 1.

Assign a value for the variable x and set l=0.

Step 2.

Divide the close interval [0,T] into m=1000 pieces, and denote each time point as ti, i=1,,m.

Step 3.

For each ti, generate a random number ni from the Poisson distribution P(λti), and set ni as the sample size of the claims.

Step 4.

Generate the accident arrival time {τki,k=1,,ni} from the uniform distribution U(0,ti) and the claim sizes {Xki,k=1,,ni} from (4.1) and (4.2).

Step 5.

Calculate the expression D below for each ti and denote them as {Di,i=1,,m}: Di=k=1niXkie-rτki-0tie-rsC(ds),i=1,,m, where r and C(t) have been defined and their values have also been assigned.

Step 6.

Select the maximum value from {Di,i=1,,m}, and denote it as D*, compare D* with x; if D*>x, then the value of l increases 1.

Step 7.

Repeat Step 2 through Step 6, N=109 times.

Step 8.

Calculate the moment estimate of the finite-time ruin probability, l/N.

Step 9.

Repeat Step 1 through Step 8 ten times and get ten estimates. Then, choose the median of the ten estimates as the analog value of the finite-time ruin probability.

For different value of x, the analog value and the theoretical result of the finite-time ruin probability are presented in Table 1, and the percentage of the error relative to the theoretical result is also presented in the bracket behind the analog value. It can be found that from Table 1, the larger x becomes, the smaller the difference between the analog value and the theoretical result is. Therefore, the approximate relationship in Theorem 2.2 is reasonable.

Comparison between the analog value and the theoretical result in Theorem 2.2.

x  (×103) Theoretical result Analog value
0.5 3.2846e-6    3.8120e-6  (16.1%)
1   8.2270e-7  9.1100e-7  (10.7%)
2   2.0586e-7    2.2300e-7  (8.3%)
5   3.2956e-8    3.5000e-8  (6.2%)
Acknowledgments

The authors would like to thank the two referees for their useful comments on an earlier version of this paper. The revision of this work was finished during a research visit of the first author to Vilnius University. He would like to thank the Faculty of Mathematics and Informatics for its excellent hospitality.  Research supported by National Natural Science Foundation of China (no. 11001052), China Postdoctoral Science Foundation (20100471365), National Science Foundation of Jiangsu Province of China (no. BK2010480), Natural Science Foundation of the Jiangsu Higher Education Institutions of China (no. 09KJD110003), Postdoctoral Research Program of Jiangsu Province of China (no. 0901029C), and Jiangsu Government Scholarship for Overseas Studies, Qing Lan Project.

WangD.Finite-time ruin probability with heavy-tailed claims and constant interest rateStochastic Models2008241415710.1080/15326340701826898238468910.1080/15326340701826898ZBL1132.91502ClineD. B. H.Intermediate regular and Π variationProceedings of the London Mathematical Society199468359461610.1112/plms/s3-68.3.594126231010.1112/plms/s3-68.3.594ZBL0793.26004EmbrechtsP.OmeyE.A property of longtailed distributionsJournal of Applied Probability1984211808773267310.2307/3213666ZBL0534.60015BinghamN. H.GoldieC. M.TeugelsJ. L.Regular Variation1987Cambridge, UKCambridge University Press898871TangQ.TsitsiashviliG.Precise estimates for the ruin probability in finite horizon in a discrete-time model with heavy-tailed insurance and financial risksStochastic Processes and their Applications2003108229932510.1016/j.spa.2003.07.0012019056ZBL1075.91563KlüppelbergC.StadtmüllerU.Ruin probabilities in the presence of heavy-tails and interest ratesScandinavian Actuarial Journal1998149581626664ZBL1022.60083AsmussenS.Subexponential asymptotics for stochastic processes: extremal behavior, stationary distributions and first passage probabilitiesThe Annals of Applied Probability19988235437410.1214/aoap/1028903531162493310.1214/aoap/1028903531ZBL0942.60034KalashnikovV.KonstantinidesD.Ruin under interest force and subexponential claims: a simple treatmentInsurance: Mathematics and Economics200027114514910.1016/S0167-6687(00)00045-7179697610.1016/S0167-6687(00)00045-7ZBL1056.60501KonstantinidesD.TangQ.TsitsiashviliG.Estimates for the ruin probability in the classical risk model with constant interest force in the presence of heavy tailsInsurance: Mathematics and Economics200231344746010.1016/S0167-6687(02)00189-0194554310.1016/S0167-6687(02)00189-0ZBL1074.91029TangQ.Asymptotic ruin probabilities of the renewal model with constant interest force and regular variationScandinavian Actuarial Journal20051152118521ZBL1144.91030TangQ.The finite-time ruin probability of the compound Poisson model with constant interest forceJournal of Applied Probability2005423608619215750810.1239/jap/1127322015ZBL1132.91500LiuL.Precise large deviations for dependent random variables with heavy tailsStatistics & Probability Letters20097991290129810.1016/j.spl.2009.02.001251901310.1016/j.spl.2009.02.001ZBL1163.60012EbrahimiN.GhoshM.Multivariate negative dependenceCommunications in Statistics198110430733761240010.1080/03610928108828041ZBL0506.62034BlockH. W.SavitsT. H.ShakedM.Some concepts of negative dependenceThe Annals of Probability198210376577265954510.1214/aop/1176993784ZBL0501.62037TangQ.TsitsiashviliG.Randomly weighted sums of subexponential random variables with application to ruin theoryExtremes20036317118810.1023/B:EXTR.0000031178.19509.572081850ZBL1049.62017ChenY.ChenA.NgK. W.The strong law of large numbers for extended negatively dependent random variablesJournal of Applied Probability2010474908922275290210.1239/jap/1294170508ZBL1213.60058TangQ.Insensitivity to negative dependence of the asymptotic behavior of precise large deviationsElectronic Journal of Probability2006111071202217811ZBL1109.60021ChenY.YuenK. C.Sums of pairwise quasi-asymptotically independent random variables with consistent variationStochastic Models2009251768910.1080/15326340802641006249461410.1080/15326340802641006ZBL1181.62011