On the concavity of the infinitesimal renewal function (Q805078)

Let \(\{\) X(t), \(t\geq 0\}\) be a process with stationary independent non- negative increment such that \(X(0)=0\). The expected time of the first passage above the level x is \(H(x)=\int^{\infty}_{0}\delta_ 0(x- X(t))dt,\) where \(\delta_ 0(x)=1\) if \(x\geq 0\), \(=0\) otherwise. H(x) is called infinitesimal renewal function. The author gives some sufficient conditions on \(\{X(t),\quad t\geq 0\}\) for H(x) to be concave. Two main results are: 1) If \(\{X(t),\quad t\geq 0\}\) is a compound Poisson process generated by a DFR distribution function, then H(x) is concave; 2) fix arbitrary \(\epsilon >0\), if \(F(x,t)=\Pr (X(t)\leq x)\) are DFR for \(t\in (0,\epsilon)\), then H(x) is concave. Some connections with \textit{M. Brown}'s conjecture [Ann. Probab. 9, 891-895 (1981; Zbl 0527.60079)] and shock models are pointed out in this paper.