As a practical matter, the factors that determine how long our savings will last after we retire fall into two categories: those we can control and those we cannot. Major factors that are largely beyond our control include, of course, capital markets returns and how long we live. Factors that are, at least to some degree, within our control include:
- Annual withdrawal amount
- Annual withdrawal adjustment (cost of living increase)
- Asset allocation (how money is invested)
- Investment expenses
- Withdrawal strategy (which assets to withdraw, and when)
While each of these has been has been explored in academic research, a review of popular financial planning applications finds that the influence of withdrawal strategy has been largely ignored at the consumer/advisor level. In fact, nearly all programs assume that retiree portfolios maintain a constant allocation with annual portfolio rebalancing throughout retirement. Under these models, withdrawals from the various asset classes are typically made in proportion to their weight in the portfolio, followed by rebalancing to get the asset classes back to their designated percentages of the portfolio. The unanimity with which this methodology has been applied has led to the implicit assumption that constant allocation with annual rebalancing is an efficient withdrawal strategy for real world portfolios. But is this truly the case?
To illustrate the influence that withdrawal strategy may have on retirement income sustainability, consider the following reasonably realistic retirement planning scenario:
- 30 year retirement horizon
- $1 million initial retirement portfolio value
- 5% ($50,000) initial withdrawal rate
- 3% annual cost of living increase
- 60%-to-40% allocation between stocks and bonds, with equities divided 50% large cap, 30% small and mid-cap and 20% international; bonds are assumed to generate a constant 2% rate of return (roughly equivalent to the current 10 year Treasury yield)
- 1% annual investment expenses
With these conditions, I tested four different withdrawal strategies using Nest Egg Guru’s Retirement Spending Calculator, a tool I developed. The results of 5,000 simulations for each of the four strategies tested are presented in the tables below. The left column shows performance percentiles for the portfolios in the 5,000 simulations — the performance for the portfolio at the 80th percentile, the 60th percentile and so on, down to the worst performing portfolio in any simulation. The amounts in the grid are the remaining portfolio balances in five-year increments at the given percentile. Each table also shows the median portfolio — where half did better and half did worse — as well as the percentage of the 5,000 simulations that were successful, meaning they did not run out of money in the 30-year time horizon.
Table 1: Constant allocation with annual rebalancing
Successful portfolios out of 5,000 simulations = 33%
Table 2: Spend stocks first, then bonds (decreasing equity glidepath)
Successful portfolios out of 5,000 simulations = 28%
Table 3: Spend bonds first, then stocks (increasing equity glidepath)
Successful portfolios out of 5,000 simulations = 84%
Table 4: Guardrail Strategy (constant allocation, don’t spend stocks after down years)*
Successful portfolios out of 5,000 simulations = 60%
Note: The guardrail strategy involves converting stock and bond gains, if any, to cash and then spending down the cash, followed by bonds and then stocks. Under this approach, equity allocations are not reduced following negative-return years until all of the cash and bond allocations have been exhausted.
The results in these four tables clearly show that the choice of withdrawal strategy can have a dramatic effect on retirement income sustainability. The results also show how the widely adopted constant allocation strategy may be less than optimal.
Though the difference between the constant allocation and “bonds-first” withdrawal strategy is surprisingly dramatic, it is generally consistent with published research. For example, in a June 2007 paper in the Journal of Financial Planning, SUNY professors John Spitzer and Sandeep Singh demonstrated the superiority of a similar bonds-first withdrawal strategy. They concluded:
“While the wisdom of rebalancing in the accumulation phase of the life cycle is widely accepted, the wisdom does not appear to extend to the withdrawal phase. … Rebalancing during the withdrawal phase provides no significant protection on portfolio longevity.”
More recently, in a 2014 article in the Journal of Financial Planning, noted retirement researchers Wade Pfau of American University and Michael Kitces of the blog Nerd’s Eye View (not affiliated with NerdWallet) reached a similar conclusion about the value of avoiding equity withdrawals early in retirement.
While this post is intended to draw attention to the fact that withdrawal strategy is an overlooked and critically important consideration in determining income and portfolio sustainability, it is not necessarily intended to suggest that the bonds-first approach is the optimal solution. Indeed, a number of papers have been published recently proposing a range of dynamic withdrawal strategies that are likely superior the ones presented in this paper. Whether these strategies are practically implementable by consumers and their financial planners is a matter for further discussion.
Additionally, all readers should understand that the output from retirement planning calculators can vary widely from one application to the next, depending upon the calculators’ inputs and methodologies. Critical thinking is required by the user in evaluating the merits and limitations of all retirement planning calculators.