An explicit solution for an optimal stopping/optimal control problem which models an asset sale

DOI10.1214/07-AAP511MaRDI QIDQ957514zbMATH OpenOpenAlexFDO

Publication date 27 November 2008

Published in The Annals of Applied Probability (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/0806.4061

gambling incomplete market local time optimal stopping singular control utility maximization

Stopping times; optimal stopping problems; gambling theory (60G40) Martingales with continuous parameter (60G44) Probabilistic games; gambling (91A60) Optimal stochastic control (93E20)

Abstract: In this article we study an optimal stopping/optimal control problem which models the decision facing a risk-averse agent over when to sell an asset. The market is incomplete so that the asset exposure cannot be hedged. In addition to the decision over when to sell, the agent has to choose a control strategy which corresponds to a feasible wealth process. We formulate this problem as one involving the choice of a stopping time and a martingale. We conjecture the form of the solution and verify that the candidate solution is equal to the value function. The interesting features of the solution are that it is available in a very explicit form, that for some parameter values the optimal strategy is more sophisticated than might originally be expected, and that although the setup is based on continuous diffusions, the optimal martingale may involve a jump process. One interpretation of the solution is that it is optimal for the risk-averse agent to gamble.

Review text

The paper deals with an mixed optimal stopping and optimal control problem which models the decision facing a risk-averse agent over when to sell an asset. The market is incomplete so that the asset exposure cannot be hedged. In addition to the decision over when to sell, the agent has to choose a control strategy which corresponds to a feasible wealth process. The problem is formulated as one involving the choice of a stopping time and a martingale. Let \(U(\cdot,\cdot,\cdot)\) be payoff function in the optimization problem which form depends of utility function and it is a function of time, control process \(X\) and an exogeneous Markov process \(Y\). The set of feasible strategies is \(\mathcal X=\{X:{\mathbb E}X_0=x;\; X_u+Y_u\geq 0,\; u\leq \tau; X \text{ is a càdlàg martingale such that }[X,Y]_t\equiv 0\}\). The generic problem is to find \(V_\star=\sup_{\tau,\; X\in {\mathcal X}}{\mathbb E}[U(\tau,X_\tau,Y_\tau)]\) [see \textit{I.~Karatzas} and \textit{H.~Wang}, SIAM J. Control Optim.~39, 306--329 (2000; Zbl 0963.93079) for terminology of such optimization problem]. The form of the solution has been proposed and verified. The solution is available in a very explicit form, that for some parameter values the optimal strategy is more sophisticated than might originally be expected, and that although the setup is based on continuous diffusions, the optimal martingale may involve a jump process. One interpretation of the solution is that it is optimal for the risk-averse agent to gamble. The elements in the model which are necessary for the main conclusion that gambling can be beneficial are that the market is incomplete, that the real asset is indivisible, and that the asset sale is irreversible[see \textit{A. K.~Dixit} and \textit{R. S.~Pindyck}, Investment under Uncertainty. Princeton Univ. Press. (1994) for real option problems and \textit{F.T.~Bruss} and \textit{T. S.~Ferguson}, Ann. Appl. Probab. 12, No. 4, 1202--1226 (2002; Zbl 1005.60054), \textit{D.~Łebek} and \textit{K.~Szajowski}, Bull. Belg. Math. Soc. - Simon Stevin 14, No. 1, 143--155 (2007; Zbl 1138.60317)] for high risk investments].

Recommendations

Cites work

Cited in

(19)

This page was built for publication: An explicit solution for an optimal stopping/optimal control problem which models an asset sale

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q957514)