Randomization in the first hitting time problem (Q1038440): Difference between revisions

Let \((W_t)\) be a standard Wiener process and let \(a\) be a random variable such that \(a\geq 0\) a.s. Consider the process \(X_t= a+ W_t\) \((t\geq 0)\) starting at the random initial state \(a\), and let \(a\) linear boundary be given by \(b(t)= \mu t\), where \(\mu\geq 0\). The corresponding first hitting time \(\tau^{(\mu)}\) for \((X_t)\) is then defined by \[ \tau^{(\mu)}= \begin{cases}\text{inf}\{t\geq 0: X_t< b(t)\},\quad &\text{if }X_t< b(t)\text{ for some }t> 0,\\ \infty,\quad &\text{otherwise}.\end{cases} \] In this paper the following inverse problem for the first hitting time distribution is studied: Let \(F\) be a given distribution function. Find a distribution for \(a\) such that \[ P(\tau^{(\mu)}\leq t)= F(t),\qquad t\geq 0. \] The authors obtain a sufficient condition under which use inverse problem has a closed form solution. This problem has imnortant applications in credit risk modeling where the process \((X_t)\) represents the so-called distance to default of an obligor, \(\tau^{(\mu)}\) represents a default event, and the boundary separates the healthy states of the obligor from the default state [see, e,g., \textit{M. Avellaneda} and \textit{J. Zhu} [Risk 14(12), 125--129 (2001)].