A little flexibility goes a long way when deciding how much you can withdraw from investments every year for retirement income.

Tall buildings flex when the wind blows. A 100-storey building can shift by one meter in response to high winds. Dampers can used to reduce the swaying and provide occupants with a more comfortable experience.

Similarly, should retirees be flexible and adjust their spending when market winds blow?

We investigate how being flexible impacts total spending in retirement.

A constant withdrawal seems ideal – income is predictable and makes budgeting easier. Fixing the size of the withdrawal is a challenge: too high and the retiree runs out of money, too low and some retirement aspirations may never be fulfilled. A familiar example of a fixed spending rule is the 4% rule whereby 4% of the initial capital is withdrawn annually. An investor with $1 million who applied the 4% rule would withdraw $40,000 every year[1]. The risk of running out of money or underspending can be adjusted, but never eliminated, by tinkering with the withdrawal level. The risk is inherent in demanding a fixed income from investments that fluctuate unpredictably.

The other approach is to have a spending rule that flexes with changing market conditions: if markets do well, then spend more; if markets do badly, then spend less. This leads to the question of how to compare a constant and variable spending rule and decide which is better?

With the help of our friends Alice and Bob we consider the impact of a constant versus a variable spending rule. Alice and Bob are both 63 and are starting retirement with a portfolio of $1.5 million. We assume both live to age 95. They invest in a well diversified 60% equity, 40% bond portfolio that has an expected return of 5.26% and a volatility of 7.04%. To be realistic, we subtract advisor and portfolio fees of 1.4% from the portfolio return.

For our variable withdrawal strategy, we will use ARVA. ARVA is a variable spending rule that has been described elsewhere. In the simplified case of zero real interest rates, ARVA divides the remaining capital by the number of years left. This calculation is repeated every year so the amount to be withdrawn depends on the current market value, the retirement period left, and changes in real interest rates. Important features of ARVA are:

  • all the portfolio is converted to income and
  • volatility in the portfolio is reflected in income volatility.

To compare a constant withdrawal strategy with ARVA we started with the same withdrawal, as calculated by ARVA, as $51,227 in the first year.  The constant withdrawal case continued with the same annual withdrawal, adjusted for inflation. To model market fluctuations we assumed that market returns have a normal distribution and we sampled returns at random for the 31 years of retirement. We ran 5,000 simulations and aggregated the results, as illustrated below.

Constant Spending Variable Spending (ARVA)
Total average income $1,609,420 $2,097,481
Average Annual Income $50,294 $65,546
Standard Deviation of Average Income 0 $6,511
Probability of Ruin 15.18% 0%
Average Bequest $605,834 $0
Source: PWL Capital

On average, variable spending generated 30% more income than constant spending during the retirement period.

The average annual income using constant withdrawal is $50,294, which is less than the initial withdrawal of $51,227. This is because a sequence of poor market returns can lead to running out of money using a constant withdrawal. By age 95, the portfolio has depleted to zero on 15% of occasions using constant spending. The variable spending never runs out of money but it also never leaves a bequest.  If leaving a bequest is more important than generating retirement income then funds should be set aside separate from retirement investments. Many retirees take the view that bequests are secondary, and contingent on retirement needs being met.  In the event that withdrawals from the portfolio exceeds spending needs then gifting during retirement, rather then leaving a bequest, may be more attractive and more tax effective.

The constant spending produces a steady income (except when it falls to zero), while the variability of spending with ARVA is about +/-10%, as indicated by the standard deviation. Like the movement of tall buildings, the income variations can be damped by adding a mass, this time of bonds, into the portfolio, as explained here.

Whether a constant or variable spending is preferable depends on which risks are of most concern to investors. Many retirees are very averse to running out of money and are flexible about whether any money is left over when they die. Also, the idea of adjusting spending according to portfolio performance is not only intuitive but responsible, and most retirees are able to separate essential spending needs from desirable spending that is more flexible.

A little flexibility can go a long way.

[1] We will keep the conversation in today’s dollars, so every year the amount withdrawn would be increased by inflation to preserve real spending power.