Unifying discrete and continuous time mixed optimal stopping-singular stochastic optimization for power plant investment
Date:
Minisymposium name: Recent Advances in Stochastic Control with Multiple Players, Incentives, and Financial Economics
Talk abstract: In this talk, we investigate a stochastic control problem in which a project manager invests money and effort into a power plant project to increase its value and reduce its variability. The manager continues to invest optimally until the project’s value becomes high enough to trigger acquisition by a parent firm or until the project is abandoned, incurring a terminal cost. We show that the continuous-time formulation of this problem reduces to a mixed control problem governed by a variational Hamilton–Jacobi–Bellman (HJB) equation with gradient constraints. We also examine the equivalent discrete-time Markov decision process and demonstrate the convergence of the corresponding value functions.