Inconsistent investment and consumption problems (Q2355306)

Motivated by mean-variance control problems in portfolio optimization, the article considers a class of investment-consumption problems that is not amenable to dynamic programming due to the time-inconsistency of optimal policies. By viewing it as a game among instantaneous reincarnations of the optimizing agent at each time, a notion of an equilibrium strategy is defined and an abstract characterization thereof is given in terms of a system of partial differential equations that replaces the Hamilton-Jacobi-Bellman equation. This is then specialized to mean-variance control with constant risk aversion, mean-variance control with time and state dependent risk aversion, and mean-standard deviation control without pre-commitment, in order to derive more explicit results concerning optimal policies in each case.