Last week, Morningstar published some new research entitled “The State of Retirement Income: Safe Withdrawal Rates” that claims the old “4% rule” should be changed to 3.3% given the current state of low bond yields and relatively high stock valuations.  Limited and flawed research that makes sweeping generalizations like this do nothing to help retirees make the best of their retirement.  Unfortunately, some retirees and financial advisors do not possess the expertise to know this research’s flaws and limitations.  Research like this is meaningless to an individual for the purpose of retirement planning.  After I got over my initial irritation, I became inspired to write this Insight to argue just the opposite, that an initial withdrawal rate should be higher than the old 4% rule would suggest.  I would further argue that a majority of retirement plans are overly conservative.   I base these arguments on the following ten reasons.

1). Advisors/People Anchor to Rules of Thumb

Rules of thumb, such as the “4% rule” that stemmed from William Bengen’s 1994 research study “Determining Withdrawal Rates Using Historical Data,” have to make too many simplifying assumptions to make the research work.  These simplifying assumptions make the research less meaningful to any one individual.   The three biggest simplifying assumptions made in Bengen’s study were 1). a 50% stock and 50% bond asset allocation, 2). using only the S&P 500 to represent stocks and intermediate-term U.S. government bonds to represent bonds, and 3). the same amount of income had to be inflation-adjusted and taken from the portfolio for 30 years regardless of how markets performed (i.e., no changes could be made).  

Morningstar’s own research finds that retirees can take a higher starting withdrawal rate and higher lifetime withdrawals by being willing to adjust some of these assumed variables, such as forgoing complete inflation adjustments or implementing some flexible withdrawal systems. Morningstar’s research finds retirees who employ variable withdrawal systems based on portfolio performance--taking less in down markets and more in good ones--can significantly enlarge their starting and lifetime withdrawals and would support a nearly 5% starting withdrawal rate.  See our Insight entitled “Dynamic Withdrawal Rules – A Significant Opportunity for Retirement Plan Optimization” for more.

Too many retirees and financial advisors anchor to these rules of thumb because they seem scientific and easy to understand.  Unfortunately, all they do is rob a retiree of higher potential quality of life while living only to leave money on the table upon passing beyond what was planned.   

2). Limitations of Monte Carlo Simulation Analysis

Monte Carlo simulation analysis is the tool most financial planning software systems utilize to calculate the probabilities of success for a given retirement plan.  While Monte Carlo simulation analysis is an incredibly powerful and useful tool, it is not without flaw.  Monte Carlo simulations are used to model the probability of different outcomes in a process that cannot easily be predicted due to the intervention of random variables. The basis of a Monte Carlo simulation is that the probability of varying outcomes cannot be determined because of random variable interference. Therefore, a Monte Carlo simulation focuses on constantly repeating random samples to achieve certain results.  A Monte Carlo simulation takes the variable that has uncertainty and assigns it a random value. The model is then run, and a result is provided. This process is repeated again and again while assigning the variable in question with many different values. Once the simulation is complete, the results are averaged together to provide statistical estimates.

 The disconnect is that investment markets are not entirely random.  If the S&P 500 declined 37% as it did in 2008 following the great recession, there is not an equal chance of the market suffering a 22% decline the following year like what was experienced at the end of the Internet bubble bursting at the end of 2002.  Yet, this is precisely the type of random scenario that Monte Carlo analysis can pick up as one of its simulated observations.  While markets are not predictable, they are also not completely random either.  Markets that are down are typically followed by a period of positive performance and vice-versa.  This leads Monte Carlo simulation analysis to capture some extreme positive and extreme negative simulated observations that are highly unlikely to occur.  In retirement planning, we only care about those downside scenarios.  By testing a retirement plan based on the bottom 10% of the worst-case scenarios, Monte Carlo analysis likely overestimates the downside possibilities.

(A side note: Monte Carlo simulation analysis is very sensitive to the market return assumptions used.   Some financial advisors set the assumptions to tempered forward-looking return expectations, which could be considered a further layer of conservatism.  Some financial advisors set the assumptions to actual historical performance, which may be too aggressive given current stock valuations, bond yields, and economic growth prospects.)  

3). Planning for 90th Percentlie Success Rates

Most retirement plans are modeled based on a 90th percentile success rate because many misinterpret just what this means.  Most interpret this as meaning you have a 10% chance of failing in retirement, which would be very hard to stomach indeed.  If this were true, then I would not want to be more aggressive either.  But it is not.  Having a 90th percentile success rate means there is a 10% probability that an adjustment will need to be made somewhere along the way during a 30-year retirement to avoid failure based on the underlying assumptions and modeling.  Planning for a 90th percentile success rate may be too conservative for someone with a long retirement time horizon and the ability to be a little flexible if needed.

4). Probabilities of Success Fail to Capture the Magnitude of Potential Failure

Somewhat connected to #3 above, what exactly does “failure” mean?  Planning based on probabilities of success and failure rates neglects to quantify what failure means.  Monte Carlo probability-based modeling and measurement makes no differentiation between failing by $1 million or $1.00 or running out of money with ten years to go or one month to go.  Looking at “failure” as an absolute term rather than a relative term leads many to be more conservative than they might be otherwise.

5). Modeling Ignores That Retirees Can Make Changes

Fortunately, most retirees are not forced to take an unchanging amount of income that must be adjusted for inflation every year from the portfolio for 30 years regardless of how markets perform, as the simplifying assumptions made in most research would suggest.  Being able to forgo an inflation adjustment or reduce spending during adverse investment market conditions can substantially improve one’s range of outcomes.  See our Insight entitled “Dynamic Withdrawal Rules – Higher Withdrawals With Less Risk” for more.

The problem is that only a few financial planning software systems dedicated to retirement planning can model the inclusion of flexible spending or dynamic withdrawal rules.   The two predominant financial planning software systems used in the industry, Money Guide Pro and eMoney, cannot model flexible spending or dynamic withdrawal rules.  Based on a survey conducted in 2018 by T3 and Adviser Perspectives, approximately 65% of the industry relies on one of these two software systems to conduct their retirement planning.  We have already learned how limiting this assumption can be to possible retirement outcomes.        

6). Modeling Limitations

While, thankfully, there are at least a couple of financial planning software systems that can model flexible spending or dynamic withdrawal rules, there is no system that I am aware of that can even come close to modeling the effect of going to savings or other reserve assets, such as a home equity line of credit, to fund retirement expenditures when markets are down substantially for an extended time.  Sequence-of-returns risk is the greatest risk a retiree faces, and the ability to manage this risk has tremendous implications on the initial withdrawal rate and portfolio’s longevity.  This is why most good financial advisors recommend some level of cash savings or reserve source of asset that is not correlated with investment markets.

Another limitation of financial planning software systems is their inability to model the benefits of tax efficient withdrawal order sequencing, asset location, tax loss harvesting, and gain harvesting if these tax management strategies are followed.  In their well-known Advisor’s Alpha research, Vanguard estimates that up to 1.85% of annual return equivalent value can be created using these tax management strategies.  Even if only half of this value were realized, it would amount to a significant difference in a retirement plan if it could be modeled.

7). Conservative Longevity Assumptions

Don’t forget about the conservative longevity assumption most retirement plans make also.  Thrive Retirement Specialists’ retirement planning uses U.S. Social Security Cohort Life Tables for longevity analysis. The cohort life table is based on age-specific probabilities of death, which are calculated using observed deaths (mortality) data from the cohort. Based on this data, we use 98 years old for the vast majority of our clients.  Based on the data, there is a 10% probability of one spouse being alive at 98.  Please note, the longevity assumption is a critical driver in retirement planning, and outcomes can be very sensitive to changes in this assumption.

If we are planning based on a 90% probability of success, we are planning based on what could happen in the bottom 10% of Monte Carlo scenarios.  Add to this now that we are also planning for the bottom 10% of longevity scenarios.   So the odds of experiencing both a bottom 10th percentile level of returns and a bottom 10th percentile level of longevity is 0.01% (0.10% x 0.10%).  Should we really plan based on the bottom 1% of scenarios, especially given all the other points we’ve discussed?

8). Consumption Tends to Decline in Real Terms

After considering the effects of inflation, most retirees will spend less as they further progress into retirement.  J.P. Morgan did a study using the Bureau of Labor Statistics Consumer Expenditures Surveys and found that household spending tends to peak at the age of 45, after which spending declines in all categories but health care and charitable contributions and gifts.  They found that housing remained the largest expense, even at older ages.  SmartAsset performed a similar study using the same source data and found that spending only increased in 3 of 14 spending categories: Healthcare, cash contributions, and reading.  Further, overall spending declined.  

9). Modeling Ignores Reserve Assets

Most retirement plans are based only on investment-based assets and incomes (from Social Security, pensions, annuities, etc.). They ignore other assets such as home equity, valuable collections, or those old U.S. savings bonds you might have in your drawer.  The greater these other assets, the more aggressive you can be in your retirement planning.  See our Insight entitled “Using ALL Retirement Assets to Maximize your Outcome” for more.

10). Failure to Regularly Update the Retirement Plan

Of all the points made, this one may be the most detrimental.  Many retirees and financial advisors treat retirement planning as a one-time, transactional, static exercise and then anchor to the outcome suggested by the retirement plan for a long time.  Recall that in retirement planning, assumptions are made using the bottom 10th percentile of possible outcomes.  What if a retiree is now 12 years into a 30-year retirement, and none of the bottom 10th percentiles of possible outcomes materialized?   This retiree’s situation has changed dramatically as asset values are now not as low as the plan predicted, and this retiree has a shorter planning time horizon.  This retiree can most likely substantially increase spending or gifting.   

It is good and necessary to plan for downside scenarios when retirement planning, but if those downside scenarios don’t occur, the retirement plan should be updated so that a retiree’s plan can stay optimized throughout retirement.  Retirement planning should be a dynamic, ongoing exercise, especially for those executing dynamic withdrawal rules and tax management strategies.   During a 30-year retirement, variables are bound to change.  Some you will control, but many you will not, such as actual market returns or changes in tax policy.   As soon as these variables change and what once was assumed becomes actual, the retirement plan should be updated so small adjustments can be made along the way, and spending can be increased if those downside scenarios fail to materialize.

The Goal of Retirement Planning

The goal of retirement planning is to develop an optimal strategy to convert retirement assets into sustainable income for however long the retiree lives while not overspending or underspending so the retiree can live a life of maximum contentment and happiness and be free from regret and fear.  Too much conservatism in retirement planning can lead to the unfortunate unintended consequence of sacrificing quality of life while healthy and living only to pass, leaving a much larger than planned legacy.  A proper balance must be struck so a reasonably conservative retirement plan can be created.  Unfortunately, research like what Morningstar put out only does a disservice to retirees and less-well-educated financial advisors. 

If you would like to learn more about how we can help you craft your optimal retirement plan, you can schedule a date and time that is convenient for you here at this link: https://calendly.com/thriveretire/thriveretire-call or contact us here at any time.

CONTACT