Withdrawal strategies

Withdrawal strategies are decumulation methods that generate retirement income from an investment portfolio. People generally spend many years saving for retirement and preparing to retire. What is often lacking is a strategy once they reach retirement and it comes to spend their savings.

On one hand, the retiree likely wants to ensure their yearly income supports their desired lifestyle. He/she might want to protect this income from the effects of inflation, and likely want this income to be as steady as possible.

On the other hand, if portfolio withdrawals are too aggressive, the portfolio might be prematurely depleted, i.e. the retiree might run out of money, so the withdrawal method must minimize or eliminate the odds of that happening.

Therefore, the retiree seeks a withdrawal strategy that tries to reconcile these goals; numerous strategies have been proposed. This article provides an overview of the topic, with links to other articles presenting specific strategies in more detail.

Portfolio composition assumptions

The investment portfolio is assumed to consist mostly of equities (stocks) and fixed income (bonds, Guaranteed Investment Certificates or GICs), perhaps with a cash and cash equivalents component, in proportions defined by the asset allocation.

There is generally no room for annuities for the strategies considered in this article; decumulation methods that involve annuities are considered under Safety-first retirement planning and liability matching.

Risks

Before exploring withdrawal strategies, certain risks must be mentioned that affect retirement plans. Future investment returns are unknown, future inflation is unknown, and the lifespan of the retiree is unknown, so finding an acceptable withdrawal strategy is not trivial task.

Investment risk

Investment returns might turn out to be too low over the retirement period, on average, to support the desired spending and bequests.

Sequence of returns risk

Future average returns affect outcomes, but also the order in which these returns happen. Even ‘balanced’ (e.g., 60% equities, 40% fixed income and cash) portfolios have time-fluctuating (volatile) values, and making withdrawals during a prolonged bear market can hurt the long-term prospects of a decumulation plan. A generalization of this idea is called “sequence of returns risk”: if the bad returns happen early in retirement, the odds of success go down.

Sequence of returns risk is more of a problem for strategies that do not take realized returns (or the remaining portfolio balance) into account, such as constant (fixed) dollar withdrawals, whereas adapting withdrawals to portfolio performance, i.e. flexible spending, is a generally safer approach in terms of avoiding depleting the portfolio, and might also allow higher initial withdrawals.

Longevity risk

Longevity risk is the risk of living longer than your life expectancy, or longer than the planning horizon. The longer you live, the more expensive the retirement, and the greater the odds of exhausting the portfolio. Common ways to address this problem, in a total return framework which tends to reject annuities, include a longer planning horizon (e.g., to age 95 or 100), increasing the expected return through a higher equity allocation, and/or a smaller withdrawal rate.

Inflation

Inflation decreases your purchasing power every year. In general, retirees will want to rise their income along with inflation, but future inflation is unknown.

Spending shocks

Spending shocks can derail retirement plans. Contingency planning is best addressed separately from planning for regular retirement living expenses, the topic of the current article.

Classification of strategies

So the retiree or future retiree has a portfolio and wants to convert this financial wealth into retirement income, progressively over time. A number of withdrawals strategies have been developed.

Whether the spending from the portfolio is fixed or variable is one way to classify withdrawals strategies.

Another classification, proposed by Pfau (2015), distinguishes decision rules from actuarial methods. The main differences are summarized in the following table (adapted for Canada):

	Decision rules	Actuarial methods
Examples	* Constant inflation-adjusted spending (e.g., 4% rule) * Constant percentage of remaining portfolio * Four-percent rule with guardrails * Floor-and-ceiling withdrawals * Guyton and Klinger rules	* Required minimum distributions, such as RRIF withdrawal rules * Variable percentage withdrawal (Bogleheads/FWF) * Amortization based withdrawal (PMT formula in Excel) * Annuitize the floor & invest for discretionary (hydrid strategies)
Initial spending	Higher than justified by the bond yield curve (relies on future stock growth)	May start at a lower level but spending increases if portfolio returns allow it
Spending over time	Attempt to keep steady or minimize variations	Frequent adjustments
Planning horizon	Beyond life expectancy (e.g., 30-40 years) to be conservative	Dynamic adjustments to time horizon
Investment returns hypotheses	Often based on historical data	Frequently updated in some methods
Remaining wealth at death	Wide range of outcomes	Generally smaller range of outcomes, and lower median remaining wealth (the portfolio has been spent more efficiently)

Several of these methods are presented below. Literature reviews presenting an even wider range of withdrawal methods include MacDonald et al. (2013) and Pfau (2015).

Constant real dollar

Main article: Sustainable withdrawal

The decision rule to which all other methods are generally compared, as a baseline, is the constant real dollar method. In this context, 'real' means inflation-adjusted. The idea is to take a percentage of the initial portfolio balance at retirement, like 3% or 4%, and withdraw this dollar amount of the portfolio during the first year. For example, $4k would initially be extracted out of a $100k portfolio based on a 4% rate. During the second year, the dollar amount would be increased by the rate of inflation, no matter the performance of financial markets (e.g. ).

If the initial withdrawal rate is too high, the portfolio might go to zero, so the method aims to pick an initial percentage that is deemed "safe" or "sustainable" to minimize the odds of failure. This strategy might initially seem attractive because of the stable income it provides, but if followed blindly, it can lead to ruin if investment returns turn out to be low (especially early on, see "sequence of returns risk" above). At the other extreme, if investment returns are very good (especially early on) and the withdrawal amount turns out to be too low in retrospect, this strategy might leave an overly large legacy, i.e. retirement spending has not been optimized.

This is shown by the following figure, which assumes a 4% rule, the percentage of equities shown, and the balance in fixed income:

With 80% equities (20% bonds), in the worst case-scenario, the money runs out after 20 years of retirement, but in the best-case scenario, the legacy amount is much larger than the portfolio size upon retirement.

Constant percentage

The constant percentage method, a.k.a. the "endowment approach", withdraws a fixed percentage of the remaining portfolio value each year. For example, suppose that the chosen rate is 5%: in the first year, $5k is taken out of a $100k portfolio. If the balance at the end of the year is $90k because of negative returns, the second withdrawal will be 5% of that, so $4.5k. Since the portfolio value fluctuates over time, the withdrawal amounts are very variable, and unpredictable. Bengen (2001) shows the case of a hypothetical 1955 retiree, starting with $5k withdrawals, which then would have declined to about $2700 in real (inflation-adjusted) terms by the early 1980's, due to the difficult economic environment of the 1970s. On the positive side, the portfolio can never be depleted with the constant percentage method (although there is no guarantee that the final withdrawal will be enough to live on).

The constant percentage rule is an extreme case of a variable spending rule: it is completely safe in terms of portfolio depletion, but very unpredictable in terms of meeting the retirement budget. The constant dollar (e.g., “4%”) rule is another extreme end-member: the withdrawals are completely smooth, but the portfolio balance is not taken into account, and financial ruin may ensue. Most decision rules therefore attempt to find a compromise, based on a variable spending to reflect investment performance, but with smoother withdrawals than with the constant percentage rule.

Guardrails

Klinger (2016) starts with the four percent rule (constant dollar) method but adds guardrails to ‘rescue’ portfolios in danger of failing, based on "early warning signs".

For the first method, the retiree must monitor the withdrawal rate ratio (WRR), defined as the ratio between the current withdrawal rate and the initial withdrawal rate. For example, if the initial withdrawal is 4% of the original portfolio and the current withdrawal is 5% (of current portfolio value), WRR = 5%/4% = 1.25. An example of a guardrail is to cut the dollar amount to be withdrawn by 10% when the WWR has increased to 1.2 (a 20% increase). This will bring back the WRR below the threshold. Further cuts can be made if the WRR rises too much again. Klinger (2016) only applies these adjustments to the first 15 years, i.e. the first half of a 30 year retirement.

Another proposed guardrail relates to portfolio value. The withdrawal amount is cut by 10% when the remaining portfolio value decreases to a certain level, such as 70% or 80% of the original portfolio (adjusted for inflation).

These guardrails minimize the odds of failure. More rules can be added to adjust the withdrawals upwards in good times and react to years of negative returns (Guyton and Klinger rules).

RRIF minimum withdrawals

Certain types of registered accounts have mandated minimum withdrawals that vary with age. The minimum withdrawals for Registered Retirement Income Funds (RRIFs) can be used as a simple withdrawal strategy. Each year, the withdrawal amount is calculated as a percentage of the current portfolio balance, and this percentage increases with age, based on a table: 4.00% at age 65, 4.17% at age 66, and so on. At age 95 and beyond, the percentage reaches 20%, since the objective is to deplete the registered account, so that the government gets their taxes. Note that in a RIFF context, the government mandates minimum withdrawals out of the account, but that does not mean that the retiree actually has to spend it all; the unspent portion can go into a TFSA if there is contribution room available.

The amounts withdrawn based on the RRIF table will be variable since they depend on portfolio returns and age, so this is a form of actuarial method in the Pfau (2015) classification. Vettese (2018) examined this method and found it produced "a pretty good pattern of income" in his simulations, when combined with CPP/QPP and OAS. But he thinks investors can do better.

Variable percentage withdrawal

Main article: Variable percentage withdrawal

Variable percentage withdrawal (VPW) is a method which adapts portfolio withdrawal amounts to the retiree's retirement horizon, asset allocation, and portfolio returns during retirement. Like the RRIF table method, the VPW method uses an increasing percentage of the current portfolio balance. Unlike the RIFF table method, VPW has different percentages as a function of asset allocation. The way the annual percentages were calculated in the VPW method involves the PMT function (see next section), but some assumptions were made, including living to age 100 (see ) to produce a table. Therefore, retirees don't have to actually use the PMT function every year. Beyond age 80, the method suggests looking at partial annuitization. There is also a recommendation to avoid going over 10% withdrawal rates even in old age. These two elements aim to reduce longevity risk.

Two spreadsheets are provided, and regularly updated: one to calculate withdrawals (and potentially, contributions), the other to perform backtests based on historical US and Canadian data. This method is often discussed on the Financial Wisdom Forum and the Bogleheads Forum.

Amortization based withdrawal (PMT formula)

The PMT function in Microsoft Excel is originally designed to calculate the payment amount for a fixed loan based on constant payments and a constant interest rate. For example, it could be used to calculate monthly mortgage payments. However, the function can also be used to calculate a yearly portfolio withdrawal amount in an actuarial approach. Another name for this group of strategies is "amortization based withdrawal".

The PMT formula syntax is:

    =PMT(rate, nper, pv, [fv], [type])

where rate is an expected return for investments (probably in real, inflation-adjusted terms to allow rising withdrawals), nper (number of periods) is the planning horizon, pv (present value) is the current portfolio value, fv (future value) is a desired legacy amount (cash balance at the end of the planning period), if any, and type depends on payments being made at the beginning (1) or the end (0 or omitted) of each period. Variables in brackets can be omitted. The calculation would be repeated every year, with changes made to rate, nper, and pv to reflect changes in expected returns, remaining life expectancy and portfolio values. This allows withdrawals to continually adapt to new information and mitigates the impact of past incorrect assumptions.

Difficulties with this method include finding the suitable expected return every year, and deciding on a suitable planning horizon (how to deal with longevity risk). If the planning horizon turns out to have been too short, the method will deplete the portfolio.

To illustrate the range of possible outcomes, simulations can be conducted, and percentiles plotted. Assuming an initial portfolio of one millon dollars at age 60, the following figure compares the 4% rule with a PMT formula-type approach, using the same balanced asset allocation, and a number of assumptions including using probabilities for the planning horizon based on a life expectancy table; withdrawing only 68% of the suggested amount; and a lower bound of $40k of inflation-adjusted income (the same income produced by the 4% rule):

Such a "PMT" approach produces a tear drop-shape pattern of possible outcomes on wealth versus age graphs, as opposed to the fan or cone shape produced by the 4% rule. The portfolio is spent more efficiently, with a median wealth at age 90 smaller than the original amount at age 60 (compare with the 50th percentile curve for the 4% rule). The initial income with the PMT method in these particular models is about $50k (compared with $40k using the 4% rule). The modeling approach used by Frank and Brayman to prepare the graphs is complex, and not expected to be implemented by a typical retiree, but the graphs are included here to illustrate the effectiveness of actuarial approaches taking into account age and portfolio performance to calculate yearly withdrawals.

See also

• Canada's retirement income system
• Conventional retirement planning
• Safety-first retirement planning
• Retirement budget models
• Annuities

Further reading

• Financial Wisdom Forum topic: "Strategies and Tools for Minimizing Taxation During the Withdrawal Phase"
• Financial Wisdom Forum topic: "RRIF Withdrawal Methodologies"
• Financial Wisdom Forum topic: "Retirement Withdrawal Strategies"

External links

• Bogleheads wiki: Withdrawal methods
• Bogleheads wiki: Amortization based withdrawal
• Bogleheads wiki: Amortization based withdrawa