The Effect of Income-Dependent Mortality on Withdrawal Strategies

Abstract

Recent literature documented significant dependency between individual lifetime earnings and life expectancy. In light of this phenomenon, we examine the effect of using salary-fitted mortality tables on optimized withdrawal strategies. We find that although earnings rank meaningfully affects expected retirement length, the impact of modeling salary-fitted mortality is more nuanced and depends on the investor’s risk-aversion parameter. In the base case, a risk-neutral investor is willing to adjust her optimal withdrawal strategy roughly in proportion to the change in expected retirement length and can therefore experience a substantially larger misallocation of withdrawals. In contrast, for a risk-averse investor, a 10% income-related difference in expected retirement length does not necessitate a sizable adjustment of an optimized withdrawal strategy; the unconditional mortality assumption causes annual withdrawal misallocation of only about 1% to 3% of pre-retirement consumption. Given this difference in results, we stress the need for a comprehensive and cohesive framework when analyzing the full implications of a retirement horizon assumption.

TOPICS: Retirement, statistical methods