Efficient Retirement Financial Plans: An Inverse Optimization and Parameterization of Intertemporal Discounted Habit Formation Utility

摘要

Over the past decade, retirement systems have undergone significant changes shifting from employer-sponsored pension plans to defined contribution plans, commonly referred to as 401(k) or individual retirement accounts (IRA). A critical aspect of these plans is that the individual, as opposed to the employer, is responsible for managing the account and its associated investments. Demographic data indicates that the proportion of the American population older than 55 is projected to increase considerably through 2050. In the very near future, millions of Americans will require sound advice regarding myriad retirement financial decisions. Retirement strategies currently employed by financial planners are based on rules of thumb and have been shown to be inefficient and poorly matched with retiree preference. This thesis demonstrates feasibility of applying inverse optimization and utility maximization as a means of developing efficient retirement portfolios based on individual investment preferences. We administer a survey to collect investment preference data. Next, we implement a habit formation utility model and develop a bi-level inverse optimization technique to quantify, estimate and parameterize retiree preference. Using our estimate, we generate preference- based optimal investment portfolios.