Dynamic Retirement Withdrawal Planning
R. Gene Stout, Professor,
Department of Finance and Law
Central Michigan University
334 Sloan Hall
Mt. Pleasant, MI 48859
Ph: 989-774-6459, Fax: 989-774-6456
R.Gene.Stout @ cmich.edu
Corresponding author
John B. Mitchell, Professor,
Department of Finance and Law
Central Michigan University
328 Sloan Hall
Mt. Pleasant, MI 48859
Ph: 989-774-3651, Fax: 989-774-6456
mitch1jb @ cmich.edu
As baby boomers retire, financial planners are pressured
for more meaningful and accurate withdrawal strategies for
retirement savings. The ultimate goal of these strategies
is to provide relatively stable income over the retiree's
remaining lifespan, while ensuring that the retiree does not
exhaust their savings (a situation known as portfolio ruin).
Current retirement planning literature centers upon the notion
of a 4.5% real withdrawal rate over a 30-year time horizon.
However, this strategy unnecessarily constrains affordable
withdrawal rate increases from over-performing portfolios,
and permits the continuation of withdrawals form under-performing
portfolios. Past researchers have made minor tweaks to this
basic formula, such as adjusting withdrawals with floor and
ceiling limits and different amortization lengths, resulting
in incremental decreases in probability of portfolio ruin.
Our work explores considering the retiree's remaining lifespan
when choosing a withdrawal strategy.
To do this, we constructed three different GoldSim models:
the first simulated the performance of a portfolio with a
4.5% real withdrawal rate over a 30-year time horizon; the
second considered effect of mortality on the probability of
ruin for a fixed withdrawal rate; and the third considered
mortality and dynamically adjusted the withdrawal rate. All
models were run with a portfolio of 65% large-cap stocks and
35% intermediate-term U.S. bonds, based on historic rates
of return from 1926 to 2004.
The first model demonstrated that a fixed 4.5% withdrawal
rate has a 13.44% probability of portfolio ruin over a 30-year
horizon. In the second model, if we use the same withdrawal
rate with a 60 year old retiree, the probability of ruin before
death is only 7.16%.
In the last model, the withdrawal rate varies dynamically
between an upper and a lower withdrawal rate limit. The rate
is adjusted yearly and is controlled by the current value
of the portfolio relative to the amount required to sustain
the prior rate of withdrawal over the retiree's expected remaining
lifespan. When this ratio becomes large, withdrawals can be
increased; when the ratio drops too low, the withdrawal rate
is adjusted downward. If the rate needs to be adjusted upwards,
the adjustment is based on a proportion of the difference
between the current rate of withdrawal and the rate of withdrawal
that would exhaust the portfolio over the retiree's expected
remaining lifespan. This is done to prevent an overreaction
to an unusually good year of returns.
Using conservative controls on the withdrawal rate, this
active management approach produces a 6.63% average withdrawal
rate with a 4.43% probability of portfolio ruin before death.
Comparing the three GoldSim models made it clear that active
management of withdrawal rates, based upon portfolio performance
and expected remaining lifespan, is likely to permit substantially
higher withdrawal rates while significantly reducing the probability
of portfolio ruin.
|
