failure years

[Htown Harry]

Bengen is still writing. Here's his current article:
http://www.fa-mag.com/news/is-4-5---...tml?section=47

Quote:

About four years ago, I authored an article for Financial Advisor in which I examined risks to the “4.5% safe withdrawal rule,” which I had developed over many years of research. At that time, my conclusion was that the rule was still viable. But when you consider the case of the person who retired on January 1, 2000 (someone who had to contend with not just one but an unprecedented two huge stock market declines within 10 years), the rule was under slight duress. I opined that the rule’s survival depended on the avoidance of a sustained bout of double-digit inflation in the near future, the kind of inflation the U.S. endured in the 1970s, which was a key element giving rise to the 4.5% rule in the first place. In this article, we’ll examine the evolution of “SAFEMAX” (the historically lowest safe withdrawal rate) from the late 1920s to the present, to establish a frame of reference. We will then revisit the “sustainability” prospects for people who retired in either 2000 or 2008.

[mathjak107]

although it is said over and over that the 4% safe withdrawal rate is based on the worst of times , unless you go 100% success rate it really isn't .

it failed the 4 or 5 worst of the worst case scenario's at the acceptable 95% level . to me that is not really safe at all .

if i was building a house i would build it to withstand hurricane sandy being i live in nyc .

95% would be like not only building it to withstand sandy but 3 or 4 less powerful hurricanes too . kind of self defeating in a way doing that .

don't you want to at least start out being able to withstand the worst of the past ?

[Totoro]

Quote:

Originally Posted by mathjak107 if i was building a house i would build it to withstand hurricane sandy being i live in nyc . 95% would be like not only building it to withstand sandy but 3 or 4 less powerful hurricanes too . kind of self defeating in a way doing that . don't you want to at least start out being able to withstand the worst of the past ?

Maybe. Fortifying your house costs money, or in our context more time in the workforce. That last 5% can be quite costly relatively speaking. In addition, you can't protect your house from all dangers.

E.g. It's 95% over 30 years, while a 55 year old single male only has a 38% chance of making it that far (vanguard longevity calculator). So it's really 98% adjusted for life expectancy in his case.

Similar arguments on the upside can be made for surprise inheritances, tax breaks, lifestyle adjustments. Downside of course is long term illness and tax hikes.

In the end it's risk appetite vs. costs. 95% is quite reasonable to me for the purpose stated, but different scenarios should be tested and adjusted for individual circumstances.

[mathjak107]

while i agree with what you said the last 5% does not always require more time in the work force . a drop in withdrawal rate from 4% to 3.50% shows 100% for 30 years . it can just mean it should really be called the 3.50% rule .

having survived the worst of the worst can give you a lot more slack in the plan then removing the really bad ones . from the surveys i have seen many retirees just seem to naturally fall out in that 3-3.50% range once they take ss .

i know we are higher now delaying ss but we will be mid to low 3% range once we file .

[Senator]

Quote:

Originally Posted by RunningBum Yes, that's my point. It is counterintuitive that a 30 year time frame would have a lower success rate than a 50 year time frame, but the reason is that a 30 year time frame includes 1966, but the 50 year time frame does not yet include that bad start.

Rather than use a 50-year time frame, use a 25-year time frame with only half of the assets. That should pickup all of the years from about 1990 forward.

If you cannot survive the 25 years, with 50% of the assets, you are not financially prepared to retire with the numbers you have input.

[mystang52]

Sorry for thread creep but I'm confused. If I input my assets, pensions, and yearly expenses etc. what does 4% rule - or any other percentage rate - have to do with it? Wouldn't the projected expenses be the only thing that matters?

[audreyh1]

I always assumed that it was understood that 95% meant you had a 5% chance of running out of money within 30 years. But that also assumed you withdrew inflation-adjusted amounts blindly and made no adjustments due to a rapidly shrinking portfolio. It stresses the need to have wriggle room - discretionary expenses that can be foregone in the face of a big portfolio drop.

Bernstein philosophized that there wasn't much point in among for better than lime 80% because he thought some super disruptive financial event/disaster that couldn't be planned for had a 20% chance of happening in our lifetimes.

Me, I prefer a slightly lower withdrawal and no inflation adjusting, and using remaining portfolio to calculate withdrawal (which is also an aggressive approach after good market years). And a large cash cushion outside that portfolio. And probably almost half our spending is discretionary.

But I think one of the folks, Bengen?, really did find like 4.2% survived all historical cases. He might have used a slightly different set of asset classes.

[audreyh1]

Quote:

Originally Posted by mystang52 Sorry for thread creep but I'm confused. If I input my assets, pensions, and yearly expenses etc. what does 4% rule - or any other percentage rate - have to do with it? Wouldn't the projected expenses be the only thing that matters?

You take your projected expenses (including projected taxes), you divide by 4%, i.e. multiply by 25, and that tells you the minimum you need to have in your retirement nest egg before you can quit working. It then assumes you take this amount out each year, inflation adjusted.

[audreyh1]

Quote:

Originally Posted by MBSC The FireCalc portfolio tab defaults to long-term corporate bonds. Source: https://www.bogleheads.org/wiki/Trinity_study_update Source: Safe Withdrawal Rates for Retirement and the Trinity Study

There you go!

Not surprising that relying on corporate bonds during times of market stress yields a lower safe withdrawal rate, as govt bonds are less correlated to stocks and behave better during equity downturns.

Super key point!

I thought the Trninty study yielded better succes rates.

[audreyh1]

Quote:

Originally Posted by mathjak107 although it is said over and over that the 4% safe withdrawal rate is based on the worst of times , unless you go 100% success rate it really isn't . it failed the 4 or 5 worst of the worst case scenario's at the acceptable 95% level . to me that is not really safe at all . if i was building a house i would build it to withstand hurricane sandy being i live in nyc . 95% would be like not only building it to withstand sandy but 3 or 4 less powerful hurricanes too . kind of self defeating in a way doing that . don't you want to at least start out being able to withstand the worst of the past ?

Before concluding that the claims are completely wrong based on Firecalc, you need to review the Trinity study and Bengen's papers and then figure out what to adjust in the Firecalc model: the asset classes used.

[bingybear]

Quote:

Originally Posted by ERD50 Unless you have a proven Crystal Ball, what other options do you have? I would not go so far as to say it's a 'guess' - anyone can guess. It is a number derived from past cycles. Look at the description on the FIRECalc page, with their analogy to temperatures - we don't guess that it will be hot in the summer and cold in the winter here in the Midwest, it's based on historical cycles. Yes, we can get lower lows and higher highs, records are made to be broken. And the 4% 'rule' isn't based on the 'growth of the last 116 years', it is based on the worst cycles in those 116 years. In most other cycles, the 4% is way conservative. IIRC, even 6% survives 50% of the time. -ERD50

While the 4% estimate is based on the worst historical cycles, it does depend on the the growth that followed. However, in the most likely outcome, the person who used the 4% WR (95% case) will end up so flush with $ because the likely case will be much better since the method is based on high levels of certainty.

Quote:

Originally Posted by mathjak107 although it is said over and over that the 4% safe withdrawal rate is based on the worst of times , unless you go 100% success rate it really isn't .

Even if you go 100% based on historical data, it does not really provide 100% guarantee that it will work out for you. We could always have a worse downturn in the future. But reducing the WR will increase likelihood of success.
As others have noted, errors in expenses or longevity are more likely to cause failure. It is hard to determine things like medical expenses. Even if you use max out of pocket, that could change over the years by far more than one would expect.

All these calculation are warm fuzzies in my mind. It gives you a sense of how well you are positioned. It is likely more important to have some flexibility in your spending plan. Can you reduce during down markets. What if you planned for 30 years and live 50 years? How do you handle health care max out of pocket costs or premiums that increase faster than you planned?
This is really a hope for the best, expect the worst situation. When you plan for the worst, you will likely need to find ways to spend the excess later in life.

[bingybear]

Quote:

Originally Posted by mystang52 Sorry for thread creep but I'm confused. If I input my assets, pensions, and yearly expenses etc. what does 4% rule - or any other percentage rate - have to do with it? Wouldn't the projected expenses be the only thing that matters?

The 4% rule is a withdraw rate that assumes you are increasing with inflation and for a retirement of 30 years. If you have pensions, this is added money that can be spent.

Really, just put your numbers and estimated life into firecalc and see the % of success. Or use other good monte carlo retirement analyzers. These all estimate likelihood of success. Remember... garbage in ... garbage out. There are many who have questioned if the 4% rule is valid for the new normal.
If you us a greater than 95% success rate as you "good to go" metric, you will likely have extra to spend.

[REWahoo]

Quote:

Originally Posted by bingybear Really, just put your numbers and estimated life into firecalc and see the % of success. Or use other good monte carlo retirement analyzers.

Not trying to nitpick, but for clarification FIRECalc uses history, not random ups and downs as the basis for calculations. Although you can select a monte carlo option in FIRECalc, it was designed to use actual investment return history - and the associated sequence of returns - and that is what separates FIRECalc from most other retirement calculators.

[bingybear]

Quote:

Originally Posted by REWahoo Not trying to nitpick, but for clarification FIRECalc uses history, not random ups and downs as the basis for calculations. Although you can select a monte carlo option in FIRECalc, it was designed to use actual investment return history - and the associated sequence of returns - and that is what separates FIRECalc from most other retirement calculators.

I did not say that FireCalc used monte carlo. I may have said it used historical data in another post in this tread... if not, Senator made that very clear. I'm not sure if the historical data method makes it better or worst than well constructed monte carlo methods/programs. I have run several tools to get my warm fuzzy on my likelihood of retirement success... including Firecalc. I was just suggesting that mustang52 could use a monte carlo tool either in place or in addition to firecalc.

[mathjak107]

fidelity uses monte carlo projections . firecalc and fidelity are pretty close . fidelity is a bit more conservative , but then again it inflates healthcare at 7% and long term care costs at 5.50% as well as the monte carlo scenario's try to find even worse outcomes

[mathjak107]

Quote:

Originally Posted by Htown Harry Bengen is still writing. Here's his current article: http://www.fa-mag.com/news/is-4-5---...tml?section=47

i thought safemax originally found the rate to be below 4% ?

[Senator]

Quote:

Originally Posted by mathjak107 fidelity uses monte carlo projections . firecalc and fidelity are pretty close . fidelity is a bit more conservative , but then again it inflates healthcare at 7% and long term care costs at 5.50% as well as the monte carlo scenario's try to find even worse outcomes

I find it a good exercise to use both (and every other calculator out there...). You do not want one to pass, and the others to fail.

With Fidelity, I even use the "under-performing market" in my projections. And add in extra money for travel. And LTC, even though I am self-insuring. And add in less rental income that I have been getting.

[Andre1969]

Quote:

Originally Posted by bingybear Consider the 30 something RE who might use 60 or 70 year retirements. I would not expect 1966 to fail with a 60 to 70 year retirement run on firecalc as there is not enough data to back test using a 1966 RE date and 60 or more years.

If you're planning a retirement of more than 30 years, it's still good to run FireCalc using 30 years, or even a shorter timeframe, to see how the results are.

In my case, for example, I'm 46 now, planning on retiring at age 50-51, and I'm plotting it out to age 100, just in case. So, essentially, 54 years. FireCalc gives me a 98.9% chance of success if I retire at 51, and want to live off of $60,000 per year. 91 or 92 cycles succeeded, meaning only one failed.

However, if I reduce it to 40 years, from 54, I now get a 96.2% success rate. This time, 102 out of 106 cycles succeeded, and 4 failed. If those cycles failed after 40 years, they're certainly not going to rebound by the 54 year mark. I'd either have to go back to work, reduce my spending, rob a bank, etc.

Tightening up even more, to 30 years, bumps my success rate to 100%...all 116 cycles passed. HOWEVER...4 of those cycles are about to run out of money at the 30 year mark. They won't survive another year or two at $60K per year. And there's a good number of other cycles that are coming a bit too close for comfort...probably wouldn't make it another 10 years at $60K per year.

Even knocking the timeline down to 5 years shows some interesting results. That would take it out to 2021, the year I want to retire. FireCalc has 141 five year cycles, and naturally they all succeed. However, according to FireCalc, my balance in 2021 could be anywhere between $767,868 to $3,221,134, with an average of $1,633,011. Well, if I only have $767K by the time I'm 51, I don't think I'm going to be too comfortable pulling the plug, and living off of $60K per year. I imagine that would end up being one of the cycles that proves to be a failure, in the long run.

[bingybear]

Quote:

Originally Posted by Andre1969 If you're planning a retirement of more than 30 years, it's still good to run FireCalc using 30 years, or even a shorter timeframe, to see how the results are. In my case, for example, I'm 46 now, planning on retiring at age 50-51, and I'm plotting it out to age 100, just in case. So, essentially, 54 years. FireCalc gives me a 98.9% chance of success if I retire at 51, and want to live off of $60,000 per year. 91 or 92 cycles succeeded, meaning only one failed. However, if I reduce it to 40 years, from 54, I now get a 96.2% success rate. This time, 102 out of 106 cycles succeeded, and 4 failed. If those cycles failed after 40 years, they're certainly not going to rebound by the 54 year mark. I'd either have to go back to work, reduce my spending, rob a bank, etc. Tightening up even more, to 30 years, bumps my success rate to 100%...all 116 cycles passed. HOWEVER...4 of those cycles are about to run out of money at the 30 year mark. They won't survive another year or two at $60K per year. And there's a good number of other cycles that are coming a bit too close for comfort...probably wouldn't make it another 10 years at $60K per year. Even knocking the timeline down to 5 years shows some interesting results. That would take it out to 2021, the year I want to retire. FireCalc has 141 five year cycles, and naturally they all succeed. However, according to FireCalc, my balance in 2021 could be anywhere between $767,868 to $3,221,134, with an average of $1,633,01. Well, if I only have $767K by the time I'm 51, I don't think I'm going to be too comfortable pulling the plug, and living off of $60K per year. I imagine that would end up being one of the cycles that proves to be a failure, in the long run.

is bold a typo?
Good points. Have you tried that on a monte carlo calculator like RIP? I would expect (or hope) you would not see the same anomalies using MC verse back test.

I have not run much shorter number of years as I would not expect such a systematic issue in many of the retirement calculators. I've been using RIP long before I ever heard of Firecalc.

[Andre1969]

Oops...yeah, that was a typo. I went back and fixed it. Should have been $1,633,011. Never tried a Monte Carlo, but I did have a Cutlass Supreme once Guess it wouldn't hurt to try other calculators though...