Static and Dynamic Tax Diversification of Withdrawals from Multiple Individual Retirement Accounts

Abstract

This article shows why diversified simultaneous retirement funding drawn from accounts with differing tax treatments will save more on taxes than sequential withdrawal. We utilize dynamic programming to quantify the optimal funding under the assumption that the goal is to maximize the total discounted after-tax consumption and bequest amounts. The dynamic programming setting uses the actual tax schedule, considers the required minimum distribution and life expectancy, and constrains consumption with upper and lower bounds. We also find the optimal funding through static diversification, a simple optimization schema. We compare the static and full dynamic programming solutions in two cases, one for a single filer and another for a married joint filer. The static and the full dynamic programming solution differ more in the single filer case than in the married joint filer case. The static optimal withdrawal is close to sequential withdrawal. The full dynamic programming’s optimal withdrawal is close the sequential withdrawal when both types of accounts have a substantial balance.

TOPICS: Retirement, statistical methods, legal/regulatory/public policy, wealth management