Random walks with non-convolution equivalent increments and their applications (Q601305)

Let \(X_{n}, n\geq 1\), be i.i.d. random variables with \(EX_1<0\) and common distribution function \(F\), let \(S_0=0, S_{n}=X_1+...+X_{n}, n\geq 1\), be the associated random walk, and let \(G\) denote the distribution function of \(\sup_{n\geq 0}S_{n}\). For \(x\geq 0\), let \(\gamma(x)=\inf{n\geq 1:S_{n}>x}\) define the first passage time over \(x\), and let \(S_{\gamma(x)}-x\) be the overshoot of the random walk at the level \(x\). Under various conditions on \(F\) and its Laplace transform, the authors first investigate the asymptotics, as \(x\rightarrow \infty\), of \(G(x), 1-G(x)\) and the probability \(P(S_{\gamma(x)}-x\in I;\gamma(x)<\infty\) for several intervals I. Then the obtained results are applied to renewal risk models and infinitely divisible laws.