Singular stochastic control in the presence of a state-dependent yield structure (Q1411892)

The singular stochastic control model \[ dX(t)=\mu (X(t)) dt + \sigma (X(t)) dW(t)- dZ(t),\quad X(0)=x, \] \[ V(x)= \sup _{Z\in \Lambda } E_x\int _0^{\tau (0)} e^{-rs}\pi (X(s)) dZ(s), \] is studied, where \(\mu : R\to R\) and \(\sigma :R\to R\) are Lipschitz continuous mappings, \(\sigma >0\), \(W\) is a Wiener process defined on a complete filtered probability space \((\Omega , P, \{\mathcal F_t\}, \mathcal F)\) and the control strategy \(Z\) is admissible, that is, non-negative, non-decreasing, right-continuous and \(\{\mathcal F_t\}\)-adapted. In the definition of the cost functional \(V\), \(\Lambda \) denotes the set of all admissible strategies, \(\tau (0)\) is the first hitting time to 0 and the mapping \(\pi :R_+ \to R_+\) representing the instantaneous marginal yields accrued from exerting the control \(Z\), is continuous, non-increasing and satisfying \(0<\pi (0)<\infty \). By means of stochastic calculus, classical theory of diffusions and nonlinear programming the value of optimal strategy is derived. A set of conditions is given under which the optimal policy is to reflect the controlled process downwards at an optimal threshold satisfying an ordinary first-order necessary condition for the optimum. The comparative static properties of the value are also considered and sufficient conditions for its concavity are given. Several illustrative examples are included.