It has become more and more evident to me that most investors have almost no real sense of what they need to save for retirement—an issue that is closely linked to the problem of determining how much income an investor can safely draw from his/her portfolio to reliably support a long-term income stream. This is obviously one of the most crucial issues in retirement planning.
The standard approach to this problem has traditionally been calculations of income draw vs. longevity risk. There is a simple “rule of thumb” that a retiree can plan future income by taking out 4% of the balance in his/her account at retirement and then drawing that amount each year thereafter, indexed upwards with inflation. Such calculations are derived using a generic portfolio that is 60% in the S&P500 and 40% in a broad bond index. The 4% draw rate is often referred to as a Safe Withdrawal Rate [SWR], but there are a range of alternative SWR calculations.
One of the ways to move beyond the very simple SWR calculations is to look at a wider asset allocation (i.e. beyond the S&P500 and domestic bonds). By adding asset classes like commodities, REITs, TIPS, international equities, and specific sectors (like utilities), we have found that you can get a substantially higher SWR than 4% (as discussed in this article called Choosing Your Portfolio Risk Tolerance.
Another area in which the “4% Rule” and related calculations are too limiting is in the assumption that investors cannot adjust their plans over time. This is patently silly. As time passes, you know exactly how much your portfolio has grown—and this information is very important in planning for subsequent years.
The standard concept of a safe withdrawal rate [SWR] from a retirement portfolio is based on the idea that an investor will draw a flat amount of real income in every year of retirement—and that this amount is determined on the day or retirement and is never revisited. This is not a very realistic assumption (I hope). A person retiring today, and trying to look forward through a 30-year retirement has limited information about the future. That person can run a Monte Carlo simulation tool like QPP and come up with an asset allocation and income draw that is projected to be sustainable.
Five years into retirement, our investor has information about how his portfolio has performed during the first five years of retirement, and he only has to plan for a remaining 25 years. If he sits down and calculates how much money he can draw and what the right asset allocation is to maximize his ability to provide income for the rest of his life, he will almost undoubtedly come out with different results for safe income draws and the choice of asset allocation than he did when he performed the calculations five years previously.
As an investor progresses through his/her life, it makes sense to periodically perform the projections of future income draws that can be safely sustained every few years (at least). As time passes, returns for years that were in the future are actually realized—so you have more information and the time horizons for which you are planning get shorter.
The idea of setting a static portfolio and a static SWR when you retire and never re-evaluating is like a ship’s captain who sets the course for a trans-oceanic journey and then leaves the bridge for the duration of the trip. This may work out, but the ship is far more likely to reach its destination on time if the captain periodically checks in and adjusts the ship’s course [note: this is a metaphor—I know that current autopilots on ships are very good]. In almost every situation imaginable, having the ability to check in on your progress towards some goal and adjust the “course of action” makes sense—and this is certainly true for investing.
SUNY Emeritus professor of economics John J. Spitzer has just published Retirement Withdrawals: an Analysis of the Benefits of Periodic Midcourse Adjustments, available here in PDF format.
Dr. Spitzer’s approach was originally proposed by Ben Stein and Phil DeMuth in their 2005 book called Yes, You Can Still Retire Comfortably. Dr. Spitzer’s analysis reveals a key result: periodically re-evaluating asset allocation and safe withdrawal rates increases the income that retirees can safely draw.
To help to motivate how this process works, I have developed an example case. Let’s imagine that we have Jane Doe, who has saved $1,000,000 and is retiring today at age 60. Jane looks at the portfolios described in Choosing Your Portfolio Risk Tolerance and decides that the P50 portfolio is to her liking:
She determines that she will draw $45,000 per year against this portfolio for the next five years, adjusting her income draw upwards to keep pace with inflation (assumed to be 3% per year). At that point, she will re-evaluate her strategy. Using QPP, we estimate that Jane can draw this income level and have only a 20% chance of running out of money by age 89, and Jane (hypothetically) is happy with this.
At age 65, Jane is drawing $52,000 per year (because of inflation). Using Monte Carlo Simulation in QPP, we can estimate the range of possible portfolio values at this time. Jane’s Median Portfolio Value is $1.1M. Jane’s 80th percentile portfolio value is $1.3M, and her 20th percentile portfolio value is $930K.
Median case after five years
If Jane’s portfolio ends up at the median value and she continues to escalate her draw at 3% a year for inflation, she now has a projected 20% chance of running out of money by age 92. In the median case, her outcomes are now brighter. If Jane wants to get her survival rate back to the original at age 60 (20% chance of running out of money by age 89), she can increase her draw to $56,000 per year (increasing with inflation). This is equivalent to drawing 5.1% per year (inflation adjusted). In the median case, Jane can increase her draw going forward because her investment horizon is shorter. We are assuming that Jane’s asset allocation remains the same.
Low Case after five years
If Jane’s portfolio ends up at the 20th percentile outcome, she will have $930K at age 65. If she continues to draw at here original projection, Jane now has a 20% chance of running out of money by age 85. Because the portfolio has performed on the low range of possible outcomes, Jane’s future income is less secure. If Jane is willing to lower her income draw from $52,000 to $48,000 per year, she raises here 20th percentile failure year to 89 again. This is equivalent to a forward-going draw of 5.1% ($48,000 divided by $930,000). Because Jane is now five years older, she can draw a larger fraction of the remaining balance in her portfolio—just as in the Median case.
High Case after five years
If Jane’s portfolio ends up on the high end of possible outcomes (the 80th percentile), Jane will have $1.3M at age 65. Whereas she would be drawing $52,000 per year (under her original plan), Jane now has both a shorter time horizon and some very good returns to bolster her portfolio value. She can increase her draw to $66,000 (which equates to 5.1% draws going forward).
Whatever situation Jane finds herself in, she is maintaining the same portfolio survival rates by adjusting her income draws over time. As she gets older, she can naturally draw a higher fraction of the remaining balance because she does not need to plan for such a long time horizon. Hopefully, this seems fairly intuitive.
Now, imagine that we step forward five years to age 70 for Jane in the median case. If she ends up with the median case again, she now has $1.15M at her higher draw rate. She is planning to draw $65,000 per year (the $56,000 adjusted up with inflation). She can increase this to $68,000 per year and still maintain her 20th percentile probability of running out of money by age 89.
There are a wide range of permutations. An 80th percentile five year period could be followed by a 20th percentile five year period in terms of performance. In this case, Jane stepped up her income draw at age 65 to $66,000, but has now experienced 20th percentile performance between age 65 and 60. She now has $1.16M, so she can draw about $68,000 going forward (as in the case in the previous paragraph).
We could certainly go through a wide range of possible future returns and income draw choices, but the few examples here make the point: an investor will, on average, be able to safely draw more income if he or she revisits the calculations from scratch every few years.
Changing the Asset Allocation
For these examples, we have kept Jane’s asset allocation static. We can easily expand the example into additional scenarios by adjusting the asset allocation at age 65. A number of factors may change in the five year period—and these factors will impact the projected risk and return of different asset classes, thereby impacting the total portfolio. This factor is quite important and we could explore it with a variety of additional examples, but the main thing to consider is that forward-looking estimates of portfolio risk and return change over time—and this will alter the projected future outcomes and safe income draws. As new information becomes available, rational expectations for the returns on a range of asset classes evolve. When Jane re-examines her portfolio and safe withdrawal rates every few years, it is important to examine how it may make sense to alter the asset allocation.
Forward-looking portfolio projections do evolve over time. After an asset class has substantially out-performed, QPP projects that it is likely to revert-to-mean over some time horizon.
There is no question that the price at which you buy into an asset makes a difference to long-term returns, and this poses one of the ongoing challenges to adjusting desired asset allocations. That said, historical performance suggests that less activity tends to be better than more activity in this regard. In other words, it made sense to reduce inflows to real estate as the real estate bubble expanded—but timing such a thing is hard and should be undertaken with considerable caution. If you have some faith in the asset allocation models and approach used to plan your retirement portfolio at age 60, it certainly makes sense to revisit these models when you are 65, 70, etc.
In earlier work on safe withdrawal rates, I showed that Monte Carlo simulations suggested that:
(1) A four percent safe draw rate is consistent with a portfolio allocated 60% to the S&P500 and 40% to a bond index fund.
(2) A more thoughtful asset allocation to a broader set of asset classes could increase this ‘safe’ draw rate.
In the current analysis, I have examined the impacts of changing the income draw with periodic evaluations through time and I have found that this approach can substantially increase safe withdrawal rates through time—consistent with Dr. Spitzer’s analysis:
(3) Periodically re-evaluating the forward safe withdrawal rate increases the effective ‘safe’ draw rate (on average).
(4) Periodically re-evaluating asset allocation also makes sense.
All Monte Carlo simulations rely on assumptions about future rates of return on asset classes and the correlations between them. There is considerable uncertainty in these numbers, so all results that project forward decades into the future are to be treated as indicative. Further, even if the projected average rates of return were perfect, an investor only gets to experience one realization of the future—not all of them. A strategy that looks good on average may be very painful if you happen to fall into the lowest 10% or 20% of possible outcomes. With these caveats, statistical models provide the best guidance available to help shape how we plan for the future. The results discussed here and in my earlier work on this issue are consistent with current standards in portfolio planning that are practiced at the largest firms: it is crucial to diversify effectively and to periodically evaluate the course you are one. These points are also consistent with common sense—always an important litmus test.
In financial theory, the value of making a decision to alter a course of action is referred to as a real option. For investors, the real options of changing draw rates and asset allocation through time clearly has value. A range of analysis has suggested that the 4% SWR is an inefficient use of capital because the average portfolio tends to end up with a large ending balance—the 4% SWR is calculated by minimizing risk, but the average portfolio ends up quite large. The point of the calculations is that no individual should bet retirement on getting average returns—he/she must plan for less rosy scenarios. By making good use of the embedded real options to make mid-course adjustments in income draws and asset allocation, the average income draw increases and becomes a more efficient use of capital. These points are nicely articulated, with academic references in Dr. Spitzer’s analysis.
As always, the ultimate test of the output from a model is whether it makes sense. This is patently true here. Revisiting your plans for income through retirement is just common sense—and the models confirm that doing so should increase both portfolio survival rates and average income draw.
To make effective use of the effective real options to change course, it is crucial to have a consistent and objective framework in which to plan and make decisions. QPP was designed to provide these functions. Without such a tool, there is the danger than investors will end up acting in ways that are detrimental to their future well-being. Historically, retail investors tend to re-allocate their portfolios in ways that take away value rather than add value by chasing performance. For this reason, a considerable amount of the value of mid-course adjustments in income draws and asset allocation will depend on having good tools, a well-defined process, and the discipline to follow the process.