This paper investigates the two-sided first exit problem for a jump process having jumps with rational Laplace transform. The corresponding boundary value problem is solved to obtain an explicit formula for the first passage functional. Also, we derive the distribution of the first passage time to two-sided barriers and the value at the first passage time.
1. Introduction
One-sided and two-sided exit problems for the compound Poisson processes and jump-diffusion processes with two-sided jumps have been applied widely in a variety of fields. For example, in actuarial mathematics, the problem of first exit from a half-line is of fundamental interest with regard to the classical ruin problem and the expected discount penalty function or the Gerber-Shiu function as well as the expected total; see, for example, Mordecki [1], Xing et al. [2], Zhang et al. [3], Lewis and Mordecki [4], and Avram et al. [5]. In mathematical finance, the first passage time plays a crucial role for the pricing of many path-dependent options and American type options; see, for example, Geman and Yor [6], Bertoin [7], Kyprianou [8], Rogers [9], Avram et al. [10], and Alili and Kyprianou [11]. Recently, Cai [12] investigated the first passage time of hyperexponential jump-diffusion process. Cai and Kou [13] proposed a mixed-exponential jump-diffusion process to model the asset return and found an expression for the joint distribution of the first exit problem for a jump and overshoot for a mixed-exponential jump-diffusion process. Chen et al. [14] and Yin et al. [15] discussed the first passage functional for hyperexponential jump-diffusion process.
Motivated by works mentioned above, the main objective of this paper is to study the first exit time of the two-sided first exit problem for jump-diffusion process having jump with rational Laplace transform proposed by Lewis and Mordecki [4]; see also Kuznetsov [16]. This extends recent results obtained in Chen et al. [14, Theorem 2.5] on the hyperexponential jump-diffusion process.
The rest of the paper is organized as follows. In Section 2, we introduce the jump-diffusion process having jumps with rational Laplace transform. Section 3 concentrates on deriving the joint distribution of first exit time and a nonnegative bounded measurable function of the process value at the first exit time to two flat barriers. Section 4 presents the analytical solution to the pricing problem of standard double-barrier options.
2. The Model
A Lévy jump-diffusion process X={Xt,t≥0} is defined as(1)Xt=X0+μt+σWt-∑i=1NtYi,where μ∈R and σ>0 represent the drift and volatility of the diffusion part, respectively, W={Wt,t≥0} is a (standard) Brownian motion, N={Nt,t≥0} is a homogeneous Poisson process with rate λ, and {Yi,i=1,2,…} are independent and identically distributed random variables supported in R∖{0}; moreover, {Wt,t≥0}, {Nt,t≥0} and {Yi,i=1,2,…} are mutually independent; finally, the probability density function (pdf) of Y1 is given by(2)fy=∑j=1m∑i=1mjpijηjiyi-1i-1!e-ηjy1y≥0+∑j=1n∑i=1njqijθji-yi-1i-1!eθjy1y<0,where pij,qij≥0, such that ∑j=1m∑i=1mjpij+∑j=1n∑i=1njqi,j=1, Re(ηj) and Re(θj)>0 and that ηi≠ηj and θi≠θj for all i≠j.
Another important tool to establish the key result of the article is the infinitesimal generator of Xt. Note that Xt is a Markovian process and its infinitesimal generator is given by(3)Lhx≔limt↘0EhXt∣X0=x-hxt=μh′x+σ22h′′x+λ∫-∞+∞hx-y-hxfydy,for any bounded and twice continuously differentiable function h.
Throughout the rest of the paper, the law of X such that X0=x is denoted by Px and the corresponding expectation by Ex; we write P and E when x=0. The Lévy exponent of X is given by (4)Gζ=lnEexpζXtt=μζ+σ22ζ2+λEe-ζY1-1=μζ+σ22ζ2+λ∑j=1m∑i=1mjpijηjiηj+ζi+∑j=1n∑i=1nj-qijθjiζ-θji-1.Accordingly, G is a rational function on C. The equation G(ζ)-α=0 with >0,σ>0, and μ∈R yields S=M+N+2 zeros with M=∑i=1mmi and N=∑j=1nmi; see Kuznetsov [16].
Let us denote the zeros of G(ζ)-α in the half-plane Re(ζ)>0Re(ζ)<0 as ρ1,ρ2,…,ρM+1ρM+2,ρM+3,…,ρM+N+2. Also, we assume that all zeros of G(ζ)-α are simple.
3. Distribution of the First Passage Time to Two Flat Barriers
Define τ to be the first passage time of Xt to two flat barriers h and H with h<H; that is,(5)τ≔inft≥0:Xt≥horXt≤H.And let(6)ϕx=Exe-ατgXτ,where α>0 and g is nonnegative bounded measurable function.
Now, by Feynman-Kac formula (see, e.g., Theorem 4.4.2, Karatzas and Shreve [17]) we have that ϕ(x) must satisfy(7)L-αϕx=0inh,H,ϕx=gxon-∞,h∪H,+∞.Our goal in this section is to solve the boundary problem (7) and find explicit formulae for ϕ(x). We first show that ϕ satisfies an integrodifferential equation and then derive an ordinary differential equation for ϕ. Based on the ODE, we show that ϕ can be written as a linear combination of known exponential functions. As a consequence, its explicit form is obtained, for instance, choosing g(x) to be eγx; it is easy to derive the joint distribution of the first passage time of X to two flat barriers and the process value at the first passage time.
Now, let P0(ζ)=∏j=1m∏i=1mj(ζ+ηj)i∏j=1n∏i=1nj(ζ-θj)i; then P1(ζ)=P0(ζ)(G(ζ)-α) is a polynomial whose zero coincides with those of G(ζ)-α. Also, denote by D the differential operator such that its characteristic polynomial is P1(ζ).
The following lemma will be needed for our proof of Proposition 2.
Lemma 1.
Let d(k) indicate the kth derivative with respect to x of any differentiable function and define(8)Fi,ηj,x=ddx+ηjie-ηjx∫-∞xϕyx-yi-1eηjydywith d/dx+ηj(i) being the generalized Leibniz operator such that(9)ddx+ηji≔∑k=0iikηji-kdk.Then, for all i≥1,(10)Fi,ηj,x=i-1!ϕx.
Proof.
We proceed by induction on i. For i=1, we have (11)F1,ηj,x=ddx+ηje-ηjx∫-∞xϕyeηjydy=-ηje-ηjx∫-∞xϕyeηjydy+ϕx+ηje-ηjx∫-∞xϕyeηjydy=ϕx.Moreover, (12)ddx+ηje-ηjx∫-∞xϕyx-yi-1eηjydy=-ηje-ηjx∫-∞xϕyx-yi-1eηjydy+i-1e-ηjx∫-∞xϕyx-yi-2eηjydy+ηje-ηjx∫-∞xϕyx-yi-1eηjydy=i-1e-ηjx∫-∞xϕyx-yi-2eηjydy.It follows inductively that(13)Fi,ηj,x=ddx+ηjie-ηjx∫-∞xϕyx-yi-1eηjydy=ddx+ηji-1ddx+ηje-ηjx∫-∞xϕyx-yi-1eηjydy=i-1ddx+ηji-1e-ηjx∫-∞xϕyx-yi-2eηjydy=i-1Fi-1,ηj,x=i-1!F1,ηj,x=i-1!ϕx,which is the desired result.
We may now state the following.
Proposition 2.
Suppose a bounded solution ϕ defined on R to the boundary value problem (7) exists. Then on R∖{h,H}, ϕ is infinitely differentiable and satisfies the ODE,(14)Dϕ≡0,onh,H.Hence, on (h,H), ϕ(x)=∑k=1SQkeρkx for some constants Qk.
Proof.
Applying the infinitesimal generator L to the function ϕ, we obtain (15)Lϕx=σ22ϕ′′x+μϕ′x+λ∑i=1m∑j=1mipijηjii-1!e-ηjx∫-∞xϕyx-yi-1eηjydy+λ∑j=1m∑i=1mjqijθjii-1!eθjx∫x+∞ϕyy-xi-1e-θjydy-λϕx.Next, ϕ will be shown to satisfy an ODE. Using Lemma 1, we get, for j=1,2,…,m, i=1,2,…,mj,(16)ddx+ηjie-ηjx∫-∞xx-yi-1ϕyeηjydy=i-1!ϕx.The same computation will yield, for j=1,2,…,n, i=1,2,…,nj, (17)ddx-θjieθjx∫x+∞y-xi-1ϕye-θjydy=-i-1!ϕx.Now, since σ>0 and (L-α)ϕ≡0 then, thanks to Proposition 3.3 in the work of Chen et al. [18], ϕ is infinitely differentiable on (h,H) and(18)0=∏j=1m∏i=1mjddx+ηji∏j=1n∏i=1njddx-θjiL-αϕx=∏j=1m∏i=1mjddx+ηji∏j=1n∏i=1njddx-θjiσ22d2dx2+μddx-λ-αϕx+λ∑j=1m∑i=1mj∏k=1,k≠jm∏i=1mjddx+ηkipijηjii-1!i-1!ϕx-λ∑j=1n∑i=1nj∏k=1,k≠jn∏i=1njddx-θkiqijθjii-1!i-1!ϕx.Hence, (18) transforms the integrodifferential equation (L-α)ϕ≡0 into an ODE:D^ϕ≡0, where D^ is high order differential operator.
To complete the proof, D^ must be shown to coincide with D. To show that the characteristic polynomials of D and D^ will suffice, write P^(ζ) as the characteristic polynomial of D^. Then, by (18), P^ is given by (19)P^ζ=∏j=1m∏i=1mjζ+ηji∏j=1n∏i=1njζ-θjiμζ+σ22ζ2+λ∑j=1m∑i=1mjpijηjiζ+ηji+∑j=1n∑i=1nj-qijθjiζ-θii-1-α=P0ζGζ-α.This equation reveals that the characteristic polynomial P1(ζ) of D equals that, P^(ζ), of D^, which completes the proof.
Proposition 3.
Suppose that ϕ is a bounded solution to the boundary value problem (7) and, on (h,H),ϕ(x)=∑k=1SQkeρkx for some constants Qk. Then the constant vector Q satisfies the equation(20)AQ=Vg,where A is S×S nonsingular matrix given by(21)A=Z1Z2,(22)Z1=η1ρ1+η1eρ1h⋯η1ρS+η1eρSh⋮⋱⋮η1ρ1+η1m1eρ1h⋯η1ρS+η1m1eρSh⋮⋱⋮ηmρ1+ηmeρ1h⋯ηmρS+ηmeρSh⋮⋱⋮ηmρ1+ηmmmeρ1h⋯ηmρS+ηmmmeρSheρ1h⋯eρSh(23)Z2=θ1ρ1-θ1eρ1H⋯θ1ρS-θ1eρSH⋮⋱⋮θ1ρ1-θ1n1eρ1H⋯θ1ρS-θ1n1eρSH⋮⋱⋮θnρ1-θneρ1H⋯θnρS-θneρSH⋮⋱⋮θnρ1-θnnneρ1H⋯θnρS-θnnneρSHeρ1H⋯eρSHand Vg=Vg,1(j,i),j=1,2…,m,i=1,2…,mj,g(h),Vg,2(j,i),j=1,2…,n,i=1,2…,nj,g(H),(24)Vg,1j,i=∫-∞0gy+hηji-yi-1eηjyi-1!dy,(25)Vg,2j,i=∫0+∞gy+Hθjiyi-1e-θjyi-1!dy.
Proof.
Since L-αϕ=0 and ϕ(x)=∑k=1SQkeρkx on (h,H), which entails(26)0=L-αϕx=σ22ϕ′′x+μϕ′x+λ∫-∞+∞ϕx-yfydy-λ+αϕx=∑k=1SQkeρkxσ22ρk2+μρk-λ+α+λ∫-∞+∞ϕx-yfydy,furthermore(27)∫-∞+∞ϕx-yfydy=∫-∞h+∫H+∞gyfx-ydy+∫x-Hx-hϕx-yfydy=∑j=1m∑i=1mjpije-ηjx-h∫-∞0ηjii-1!gy+hx-h-yi-1eηjydy+∑j=1n∑i=1njqijeθjx-H∫0+∞θjii-1!gy+Hy+H-xi-1e-θjydy+∑j=1m∑i=1mjpij∑k=1SQkeρkxηjii-1!∫0x-hyi-1e-ηj+ρkydy+∑j=1m∑i=1mjqij∑k=1SQkeρkxθjii-1!∫x-H0yi-1eθj-ρkydy.Now, since (a±y)i-1=∑l=0i-1i-1l(±y)l(a)i-1-l,(28)∫0ayi-1e-βydy=β-iΓi,aβ=β-ii-1!1-e-aβ∑l=0i-1aβll!with Γi,x being the incomplete gamma function (see Gradshteyn and Ryzhik [19, page 342]).
Consequently, by combining (26) and (27) and taking into account the fact that G(ρk)-α=0 for all k=1,2,…,S, we obtain(29)0=∑j=1m∑i=1mjpije-ηjx-h∫-∞0ηjii-1!gy+h∑l=0i-1i-1lx-hl-yi-1-leηjydy+∑j=1n∑i=1njqijeθjx-H∫0+∞θjii-1!gy+H∑l=0i-1i-1lH-xlyi-1-le-θjydy+∑k=1SQkeρkh∑j=1m∑i=1mj-pije-ηjx-hηjiηj+ρki∑l=0i-1ηj+ρkx-hll!+∑k=1SQkeρkH∑j=1m∑i=1mj-qijeθjx-Hθjiρk-θji∑l=0i-1θj-ρkx-Hll!.Comparing e-ηj(x-h) and eθj(x-H) yields (20), which entails the desired result.
Lemma 4.
For a given value of α>0 the matrix A given by (21) is invertible.
Proof.
Assume that AC=0 for some vector C=C1,C2,…,CST. Consider the function V(x)=∑k=1SCkeρkx for x∈(h,H) and V(x)=0, otherwise, with ρ1,…,ρS to be the distinct zeros of the equation G(x)-α=0. Since AC=0 and V(x) is a solution to the boundary value problem,(30)L-αϕx=0,inh,H,ϕx=0,on-∞,h∪H,+∞.From the uniqueness of solutions to the boundary value problem (30), V(x)≡0 in (h,H). Now, since {eρkx,1≤k≤S} are linearly independent then C=0 and A is invertible.
In the following, y·z is written for the usual inner product of the vectors y and z in CS and for every real value x, eαρ(x)=[eρ1x,eρ2x,…,eρSx], where ρ1,ρ2,…,ρS are the S=N+M+2 roots of the equation G(ζ)=α. Our main result is the following.
Theorem 5.
For any α≥0 and a nonnegative bounded measurable function g on (h,H)c, the following assertions are equivalent:
ϕ(x)=Ex[e-ατg(Xτ)], where τ≔inft≥0:Xt≥horXt≤H.
The function ϕ(x) solves the boundary problem (7).
The function ϕ(x) is given by the formula (31)ϕx=Qg·eαρx,ifx∈h,H,gx,ifx∉h,H,with Q(g)=A-1Vg and A and Vg are given by formulas (21) and (24), respectively.
Proof.
The fact that (b) implies (c) is straightforward consequence of Proposition 3. Conversely, consider the function V(x)=∑k=1SQkeρkx for x∈(h,H) and V(x)=g(x) otherwise, where g is a bounded function on (h,H)c and Qk’s are given constants. Then the same reasoning as in Proposition 3 shows that(32)L-αVx=∑j=1m∑i=1mjpije-ηjx-h∫-∞0ηjii-1!gy+hx-h-yi-1eηjydy+∑j=1n∑i=1njqijeθjx-H∫0+∞θjii-1!gy+Hy+H-xi-1e-θjydy+∑j=1m∑i=1mjpij∑k=1SQkeρkxηjii-1!∫0x-hyi-1e-ηj+ρkydy+∑j=1m∑i=1mjqij∑k=1SQkeρkxθjii-1!∫x-H0yi-1eθj-ρkydy.Thanks to (20), we conclude that (c) implies (b).
Let us finally assume that (a) holds. Then by Feynman-Kac formula, the function ϕ(x) solves the boundary problem (7); hence (b) holds. Conversely, thanks to Proposition 4.1 in the work of Chen et al. [18], the boundary problem (7) has a unique solution; consequently (b) implies (a). The proof is complete.
As an illustration of the main result of Theorem 5, we can obtain closed-form expressions for the expectations of a variety of functions with respect to τ and Xτ. For instance, choosing g(x)=eγx in the above theorem, we can derive the joint Laplace transform of τ,Xτ, which is presented in the following corollary.
Corollary 6.
For any α>0, γ>0,(33)Exe-ατ+γXτ=Q·eαρx,ifx∈h,H,eγx,ifx∉h,H,where Q=Q1,Q2,…,QST=A-1V(γ), A is given by formula (21), and V(γ) is given by(34)Vγ=eγhγ+ηji,j=1,2…,m,i=1,2…,mj,eγh,eγHγ-θji,j=1,2…,n,i=1,2…,nj,eγH.
As another consequence of Theorem 5 and Lebesgue’s dominated convergence theorem, we get the following for the asymptotic case when h→-∞ and H→+∞, respectively.
Corollary 7.
For two flat barriers h and H(h<H), define the first downwards passage time under h and the first upwards passage time over H by (35)τh+≔inft≥0:Xt≥h,τH-≔inft≥0:Xt≤H.Then for α>0, one has the following:(36)Exe-ατh+gXτh+=Q1·e1αρx,ifx≥h,gx,ifx<h,Exe-ατH-gXτH-=Q2·e2αρx,ifx≤H,gx,ifx>H,with (37)Q1=Q1,Q2,…,QM+1T=A+-1V+,Q2=QM+2,QM+3,…,QST=A--1V-,e1αρx=eρ1x,eρ2x,…,eρM+1x,e2αρx=eρM+2x,eρM+3x,…,eρSx,V+=Vg,+j,i,j=1,2…,m,i=1,2…,mj,gh,V-=Vg,-j,i,j=1,2…,n,i=1,2…,nj,gH.Vg,+j,i=∫-∞0gy+hηji-yi-1eηjyi-1!dy,Vg,-j,i=∫0+∞gy+Hθjiyi-1e-θjyi-1!dy,A+=η1ρ1+η1eρ1h⋯η1ρM+1+η1eρM+1h⋮⋱⋮η1ρ1+η1m1eρ1h⋯η1ρM+1+η1m1eρM+1h⋮⋱⋮ηmρ1+ηmeρ1h⋯ηmρM+1+ηmeρM+1h⋮⋱⋮ηmρ1+ηmmmeρ1h⋯ηmρM+1+ηmmmeρM+1heρ1h⋯eρM+1h,A-=θ1ρM+2-θ1eρM+2H⋯θ1ρS+θ1eρSH⋮⋱⋮θ1ρM+2-θ1n1eρM+2H⋯θ1ρS-θ1n1eρSH⋮⋱⋮θnρM+2-θneρM+2H⋯θnρS-θneρSH⋮⋱⋮θnρM+2-θnnneρM+2H⋯θnρS-θnnneρSHeρM+2H⋯eρSH.
4. Pricing Double-Barrier Options
We now show how our theoretical results can be easily applied to derive pricing formulae for standard double-barrier options. We assume the asset price process St:t≥0 under the risk-neutral probability measure P is defined as St≔eXt. The log-return process Xt:t≥0 is given by (1) where X0=log(S0) and μ≔r-σ2/2-λE[eY1-1] (i.e., E[e-r-TST]=S0), where r>0 is the risk-free rate. More recently, Cai et al. [20] presented the following.
The payoff of a standard double-barrier option is activated (knocked in) or extinguished (knocked out) when the price of the underlying asset crosses barriers. For example, a knock-out call option will not give the holder the payoff of a European call option unless the underlying price remains within a prespecified range before the option matures. More precisely, consider an interval (L,U) and the initial asset price S0 is in it. The holder will receive ST-K+1{τ>T} at maturity T, where (38)τ=inft≥0:St≤UorSt≥L.Under the risk-neutral probability measure, the price of such option with maturity T and strike K is given by (39)e-rTEST-K+1τ>T∣S0.Make a change variable κ≔-logK. Then, the expectation can be represented as(40)Cκ,T≔e-rTExS0eXT-e-κ+1τ>T.Define C^(α,γ) and Δ^(α,γ) as the double Laplace transforms of the price C(κ,T) in (40) and the delta Δ(α,γ)=∂C(κ,T)/∂S0 with respect to T and κ, respectively; that is,(41)C^α,γ=∫0∞∫-∞∞e-γκ-αTCκ,TdκdT,Δ^α,γ=∫0∞∫-∞∞e-γκ-αTΔα,γdκdT.
Theorem 8.
For any γ>0 and α>maxG(γ+1)-r,0(42)C^α,γ=S0γ+1γγ+1Gγ+1-α+rQ·eα+rρx-eγ+1x1logL/S0,logU/S0x,(43)Δ^α,γ=S0γγGγ+1-α+rQ·eα+rρx-eγ+1x1logL/S0,logU/S0x,where Q=A-1V(γ+1) and A associated with r+a is defined as in Theorem 5 and V is given by formula (34).
Proof.
Equation (43) is an easy consequence of (42). To prove (43), using an idea of Kou et al. [21] (see also Cai et al. [20]) along with a change of the order of integration and the integral with respect to κ, we obtain(44)C^α,γ=∫0∞dT∫-∞∞e-γκ-αTCκ,Tdκ=Ex∫0τe-α+rTdT∫-logS0+XT∞e-γκS0eXT-e-κdκ=S0γ+1γγ+1Ex∫0τe-α+rT+γ+1XTdT.Now, we suppose that G(γ+1)<α+r and applying Itô’s formula to the process e-(α+r)t+(γ+1)Xt,t≥0, we obtain (45)Mt≔e-α+rt∧τ+γ+1Xt∧τ-eγ+1X0-∫0t∧τe-α+rT-α+reγ+1Xs+Leγ+1XTdT=e-α+rt∧τ+γ+1Xt∧τ-eγ+1X0-Gγ+1-α+r∫0t∧τe-αT+γ+1XTdTis a local martingale starting from M0=0. Since G(γ+1)<α+r, it follows from Fubini’s theorem that (46)E∫0te-α+rT+γ+1XTdT=∫0te-α+rTEeγ+1XTdT=∫0te-α+r+Gγ+1TdT=1-α+r+Gγ+1e-α+r+Gγ+1t-1∀t≥0.It follows from the dominated convergence theorem that {Mt;t≥0} is actually a martingale. In particular(47)Exe-α+rτ+γ+1Xτ-eγ+1x=Gγ+1-α+rEx∫0τe-α+rT+γ+1XTdT.Combining (44) and (47) and applying Corollary 6 we can therefore conclude that(48)C^α,γ=S0γ+1γγ+1Gγ+1-α+rExe-α+rτ+γ+1Xτ-eγ+1x=S0γ+1γγ+1Gγ+1-α+rQ·eα+rρx-eγ+1x1logL/S0,logU/S0x,where Q=A-1V(γ+1) and A associated with r+α is defined as in Theorem 5 and V is given by formula (34), which ends the proof.
Competing Interests
The author declares that there is no conflict of interests regarding the publication of this paper.
MordeckiE.Ruin probabilities for Lévy processes with mixed-exponential negative jumps200348117017610.1137/S0040585X980178XingX.ZhangW.JiangY.On the time to ruin and the deficit at ruin in a risk model with double-sided jumps200878162692269910.1016/j.spl.2008.03.034MR25424652-s2.0-53249086549ZhangZ. M.YangY.LiS. M.The perturbed compound poisson risk model with two-sided jumps201023317731784LewisA. L.MordeckiE.WienerHopf factorization for Lévy processes having positive jumps with rational transforms200845111813410.1017/s0021900200004009AvramF.PalmowskiZ.PistoriusM. R.On Gerber-Shiu functions and optimal dividend distribution for a Lévy risk process in the presence of a penalty function20152541868193510.1214/14-aap1038MR33489972-s2.0-84936797225GemanH.YorM.Pricing and hedging double-barrier options: a probabilistic approach19966436537810.1111/j.1467-9965.1996.tb00122.x2-s2.0-0030305089BertoinJ.1998Cambridge University PressKyprianouA. E.20142ndBerlin, GermanySpringer10.1007/978-3-642-37632-0MR3155252RogersL. C.Evaluating first-passage probabilities for spectrally one-sided Lévy processes20003741173118010.1017/s0021900200018386MR18089242-s2.0-0034336883AvramF.KyprianouA. E.PistoriusM. R.Exit problems for spectrally negative Lévy processes and applications to (Canadized) Russian options200414121523810.1214/aoap/1075828052MR20230212-s2.0-4043110188AliliL.KyprianouA. E.Some remarks on first passage of Lévy processes, the American put and pasting principles20051532062208010.1214/105051605000000377MR21522532-s2.0-21244456162CaiN.On first passage times of a hyper-exponential jump diffusion process200937212713410.1016/j.orl.2009.01.002MR25020422-s2.0-61349161034CaiN.KouS. G.Option pricing under a mixed-exponential jump diffusion model201157112067208110.1287/mnsc.1110.13932-s2.0-81255178443ChenY. T.SheuY. C.ChangM. C.A note on first passage functionals for Hyper-Exponential jump-diffusion process20131818YinC.WenY.ZongZ.ShenY.The first passage time problem for mixed-exponential jump processes with applications in insurance and finance20142014957172410.1155/2014/5717242-s2.0-84904612497KuznetsovA.On the distribution of exponential functionals for Lévy processes with jumps of rational transform2012122265466310.1016/j.spa.2011.09.007MR28689342-s2.0-84155171252KaratzasI.ShreveS. E.19911132ndBerlin, GermanySpringerGraduate Texts in Mathematics10.1007/978-1-4612-0949-2MR1121940ChenY.-T.LeeC.-F.SheuY.-C.An ODE approach for the expected discounted penalty at ruin in a jump-diffusion model200711332335510.1007/s00780-007-0045-5MR23229162-s2.0-34547186065GradshteynI. S.RyzhikI. M.20006thSan Diego, Calif, USAAcademic PressMR1773820CaiN.ChenN.WanX.Pricing double-barrier options under a flexible jump diffusion model200937316316710.1016/j.orl.2009.02.006MR25283742-s2.0-64949202393KouS. G.PetrellaH.WangH.Pricing path-dependent options with jump risk via Laplace transforms200574123