Finding the Optimal Withdrawal Rate on a Retirement Portfolio

Lukas Ehrenberger Novotny & Andreas MÃ¥nsson

Student thesis: Master thesis

Abstract

This thesis seeks to illuminate how investors may withdraw optimally from their retirement portfolio, considering personal attributes, to maximize their lifetime utility. It thereby challenges the popular existing research conducted by Cooley et al. (1999)which investigates a sustainable withdrawal rate, mainly by using the overlapping period method.
The main finding of the thesis is, that the personal attributes of the investors heavily impacts their recommended payment plans, suggesting that the generic sustainable withdrawal rate presented by Cooley et al. (1999) hardly can be used as a basis for planning retirement.
To do this, this study specifies a utility framework, which is being used as a tool for assessing investor utility by given payout plans, considering a series of personal traits, mainly including subjective discount rate, constant relative risk aversion, time horizon, habit consumption and the loss aversion of the investor. The thesis then assesses optimal investor payout plans with an exogenously given rate of return, thus excluding risk.
Subsequently the study introduces uncertainty on the investment assets, making the optimization problem more complex. To assess risk, the thesis considers two assets with their simulated returns being provided by Monte Carlo simulation and bootstrapping simulation, over a 30-year time horizon. These are then indexed and used for providing optimal payment plans considering lower returns given by the utilization of tail risk measures such as Value at Risk. The thesis finally attempts to show the diversity in optimal withdrawals by applying the utility framework to the fictious case examples, and by using the data from the simulations. Despite only being a fictious case, the stereotypical investors each represent qualities which are relatable to real life investors.

EducationsMSc in Applied Economics and Finance, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2020
Number of pages125