The 4% rule is being debated — again —

[xray]

Retirement Hacks
The 4% rule is being debated — again — but here’s what you should do
Last Updated: Nov. 16, 2021 at 11:19 a.m. ET
First Published: Nov. 15, 2021 at 1:04 p.m. ET
By Alessandra Malito

This retirement savings withdrawal rule was meant just as a guideline, but some experts say it is outdated.
Withdrawal rates are crucial to protecting retirement savings throughout one’s lifetime, but not all experts agree on how much someone should take out of their accounts each year. 

The 4% rule — which suggests retirees withdraw 4% of their retirement savings every year for living expenses — may be too high, according to the latest analysis of the popular strategy.

With the 4% rule, retirees would withdraw no more than 4% of their retirement assets, adjusting each year thereafter for inflation. It’s a strategy for retirees to avoid outspending their retirement savings before they die.

But a new analysis from investment research firm Morningstar suggests the withdrawal rate may be off. In fact, researchers suggest the rate should be as low as 3.3% for people who want to ensure their retirement savings last their lifetimes. The 3.3% figure assumes a balanced portfolio and fixed withdrawals over the span of 30 years, an estimated length of retirement years, which leads to a 90% probability of not running out of money in retirement.

“Given current conditions, retirees will likely have to reconsider at least some aspects of how they define their ‘safe’ withdrawal rate to make their assets last,” the research note said. “Our research finds that retirees can take a higher starting withdrawal rate and higher lifetime withdrawals by being willing to adjust some of these variables — tolerating a lower success rate or forgoing complete inflation adjustments, for example.”

Take for example adjusting expectations for successfully outlasting their retirement savings — retirees who are willing to accept an 85% success rate could bump their withdrawal rate to 3.7%, and those with an 80% probability could use a 3.9% withdrawal rate. Of course, whether lower rates of withdrawal are feasible also depends on how much money has been saved, if it’s invested appropriately and what the retirees’ budget and income needs will be — immediately upon retiring, as well as in their older ages.

This is not the first time the 4% rule has been contested. Even the guideline’s creator, Bill Bengen, said it was used too simplistically, and meant as a way for retirees to protect their nest eggs in a “worst-case scenario,” such as October 1968 when the stock market soared and inflation had gotten out of hand. A newly retired person would have had their money last if they maintained a 4% withdrawal rate, he said when he created this rule of thumb in 1994. Bengen’s updated suggestion is in a direction opposite the Morningstar researchers. He said retirees’ withdrawal rates should be no more than 5%. One reason: persistently low inflation rates. Inflation was what was most dangerous for retirees in the 1970s, he noted.

Other analysts agree the safe withdrawal rate would be around 3% — or lower. “I think it is far too aggressive today and other advisers agree,” Allan Roth, founder of Wealth Logic, wrote on Barron’s. After modeling these rates, Roth said a 3% withdrawal rate would be good for 25 years, so a couple at the age of 65 who retired should try to aim for 2-2.5% if their budget is mostly non-discretionary. This rate also depends on the age at retirement — someone younger would need to spend less whereas someone older could have a higher rate.

---------

Live Long and Prosper....

[Mustang]

From another Post copied and pasted here:

I am hardly an expert but I would take that recommendation with a grain of salt. In another thread I pointed out some of the problems I've read with Monte Carlo analysis. Basically small changes in inputs can make a big difference in the results and Morningstar has been writing about the demise of the 4% Rule for some time now.

John Rekenthaler wrote an article in Morningstar (October 8, 2020) “The Math for Retirement Income Keeps Getting Worse: Revisiting the 4% Withdrawal Rule.” He said in 2013 he thought everything was OK. He now writes we are in changing times. His article compared a 2013 retirement to a 2020 retirement. He said in 2013 the yield on treasury bonds was 3.6% and inflation was 2.3%. That made the real return 1.3%. In 2020 he said that yield was 1.4% and inflation was 1.7%. That made the real return negative. He showed that a portfolio of 100% treasury bonds would run out after 24 years. That is not a surprise. He found nothing new. All of the studies (Bengen, Trinity, Pfua 2018 update) show that 100% bond portfolios usually fall short.

As for stocks, Rekenthaler said in 2013 the price/earnings ratio of the S&P 500 was 17. In 2020 it was 27. (The norm is around 20.) According to the Shiller CAPE (Cyclically Adjusted Price Earnings) ratio he was using that meant the market is overvalued, that prices are too high and a correction is coming. Using Buffet’s forecasting formula he tested a portfolio of 50% S&P 500 stocks and 50% treasury bonds and showed that it survived a 30-year payout period. I found that interesting. In spite of the eye catching headline, he proved it could still be successful. He just didn't like the ending balance.

Another Morningstar analyst (Amy Arnott “Will the Real Retirement Income Number Please Stand Up, March 3, 2021) using Monte Carlo analysis and a different set of assumptions found that the 4% Rule had an 86% chance of success. Not bad but she seemed to be shooting for 90% even though a lot of analysts think that when using Monte Carlo analysis a 70% probably of success is acceptable. I read one article that said 50% might be acceptable. But he phrased it differently. Its not a 50% chance that the portfolio will fail. It a 50% chance that adjustments might be needed. www.kitces.com/blog/monte-carlo-retirement-projection-probability-success-adjustment-minimum-odds/

Inputs determine outputs. Here are their inputs: equity returns 6-11%, fixed income returns 2-3.5%, inflation 2.1% and a 30-year payout period. Here is a comparison of initial withdrawal rates to Pfau’s historical data findings. The top percentage the rate for a portfolio that is 75% stock. The bottom is the rate for a portfolio that is 50% stock:

Payout Period       Their Computer       Pfau's Update
15-years               6.0%/6.4%               6%/6%
20-years               4.7%/4.9%               5%/5%
30-years               3.2%/3.3%               4%/4%
40-years              2.7%/2.8%                4%/3%

Not a big difference and they didn't report the margin of error. But, I found this to be very interesting. For the input bonds has a return that is 4 to 7% less than stocks. Yet for every single payout period the portfolio with more bonds allowed for a larger initial withdrawal. That pretty much defies logic.

They also said that the initial withdrawal rate could be increased from 3.3% to 3.9% by reducing the probability of success to 80%. (Arnott’s said a 4% withdrawal rate had an 86% probability of success.) Inputs determine outputs and no one can see the future.

[Capital]

It's been a subject of debate since William Bengen first pronounced it's existence in 1994. This one is probably good for debate for another 50-60 generations at least.

[Mustang]

Since the 4% Rule has become a standard there seem to be hundreds of authors trying to disprove it. Most are trying to sell annuities. Some, like Morningstar, are using Monte Carlo analysis with exceptionally high probabilities of success. Many are comparing apples to oranges. For example to prove that 4% (the rate for 30 years) no longer works they use a 35 year payout. Others want to include fees or taxes on top of the 4% initial withdrawal. If the advisor takes 1% in fees and the retiree takes 4% then the total is a 5% initial withdrawal. Just like a regular paycheck taxes are not paid out of the portfolio on top of the 4% initial withdrawal. If the retiree pays 25% in state and local taxes that also run it up to around a 5% initial withdrawal. If the retiree pays both fees and taxes out of the portfolio then its closes to 6%. Of course its not going to last 30 years. An initial withdrawal of 5% if good for a 20 year payout. An initial withdrawal of 6% is good for a 15 year payout.

I am planning on using a modified version of the 4% Rule. I have been researching problems that will shorten the longevity of my portfolio. I have yet to find research that actually compares apples to apples that shows it will not work. But, challenging the 4% Rule makes for good headlines.

[yogibearbull]

If one skips auto-COLA & is willing to make some periodic adjustments, higher than 4% is possible.

[Mustang]

As I continue to research into problems with various systems I found this article. www.advisorperspectives.com/articles/2014/09/16/the-key-problem-with-monte-carlo-software-the-need-for-better-performance-metrics

As I skimmed through it I realized it was comparing Monte Carlo analysis to analysis using averages. Bengen pointed out that using averages resulted in a high number of portfolio failure in 1994 and he was looking for a better method when he did his study. When using averages half will succeed and half will fail. I stopped reading when I got to the Single Premium Immediate Annuity and dynamic withdrawal sections. I'm not looking into those right now.

But at the very beginning were some links to other articles that I was interested in. The first describes what Monte Carlo analysis does. Much of the details were over my head and I had never even heard of some of the terms, such as variance drain. Although I was aware of its impact. www.advisorperspectives.com/articles/2014/08/26/the-power-and-limitations-of-monte-carlo-simulations

The second link actually compared Monte Carlo analysis to the 4% Rule. www.advisorperspectives.com/articles/2014/09/02/can-retirees-still-use-a-4-withdrawal-rate-practical-applications-of-monte-carlo-analysis

Written in 2014 it sets the inputs for Monte Carlo to historical data in the 1998 Trinity Study. It pointed out that because the Trinity Study used long term corporate bonds instead of the less volatile intermediate-term government bonds the 4% Rule was only 95% successful (failing in 1965-1966) where in Bengen's 1994 study it was 100% successful.

Pfau and Blanchett point out that when using historical data the rolling retirement periods are not independent of each other (e.g. when looking at rolling 30-year retirements one year's data could be used 30 times)resulting in an over weighing of the middle periods. In Monte Carlo analysis each of the 10,000 scenarios is randomly selected and stands on its own. What was interesting is that when using the same data Monte Carlo shows a higher success rate for bond heavy portfolio than what actually occurred. A 100% bond portfolio had a 73% probability of success compared to the Trinity Study's 43%. The two seem to come together when the portfolio was 30% stock. This was attributed to the higher weighing of the middle rolling retirement periods.

The study then adds the uncharted territory of high stock valuations and low bond yields that we find ourselves in. Bond portfolios probability of success dropped from 73% to 10%. No longer using historical data but a dimmer view of the future a 50/50 portfolio's probability of success dropped from 95% to 66% but the greater the proportion of stock the greater the probability of success.

The authors then going into international probability of success looking at data starting in 1900. They specifically mention France's 36% probability of success without mentioning the two world wars that devastated many of the world economies without touching the continental US. Some of the countries with the lowest probability of success were France, Italy, Belgium, and Germany. We really shouldn't expect a high probability of success when their industry was bombed back to the stone age.  And, I think both historical returns and even Monte Carlo analysis would be useless if a nuclear war destroyed our industry.

I enjoyed reading these article. I am going to reread them One thing not mentioned is that the random independent scenarios generated by Monte Carlo simulations can come up with some highly unlikely retirement scenarios such as 8 years of losses and two years of breakeven in a ten year period. That is one of the reasons a 70% probability of success is considered acceptable by many advisors. (Note: in the last article portfolios with 60% or greater stock all showed probability of successes of 70% or more.)

I thought these article were very good if you want to learn how Monte Carlo simulations work.

[uncleharley]

There is a factor that most of these discussions leave out and that factor is that it is O K if one spends all of their money prior to death. It is also O K if one leaves a huge estate. Consequently if one feels they need $2 million or more to retire, they have a fudge factor of $2 million or whatever number to work with. I merelt wanted to point that out in an effort to keep anyone from getting to worked up over a long term plan that has an almost infinite number of variables that may or may not appear. Oh BTW, I read an article some weeks ago that stated in part that medical care as advanced to the point that the person that can live forever has probably already beeen born. That is about all I have.

[Mustang]

I have just completed a review of the Early Retirement Now Safe Withdrawal Series. big-bang-investors.proboards.com/thread/992/early-retirement

The author hates the 4% Rule so it is surprising how many times his simulations showed that it worked and was successful.

[galeno]

Using firecalc.com our 50/50 port has a 95% SWR = 4.0% over the next 31 years.

When we include our state/govt pensions which start on Jan 2028 our 95% SWR = 5.1%