The financial question
Retirement planning involves a sequence of linked decisions. Spending today affects the resources available tomorrow, while changes in financial conditions can alter what a sensible future decision looks like.
My undergraduate research examined this problem through optimal control theory, with dynamic programming and reinforcement-learning ideas informing the approach to retirement spending and adjustment.
Why optimal control
Optimal control provides a way to express a decision problem through a system’s state, available actions, constraints, and an objective over time. Dynamic programming connects the value of a decision now to the consequences of decisions later.
The appeal of this perspective is the explicit connection between a local choice and a longer-term goal. In a financial application, the usefulness of an answer depends on the assumptions chosen for the system and the objective.
A continuing interest
This project helped shape my interest in sequential decisions under changing conditions. That interest carried into my master’s work on optimal transport and dynamic programming, and continues in my work on financial models and their limitations.