Long Retirement

[fchelp]

How does one balance the desire to plan for a potentially long retirement period (say 45 years) with the resulting reduction in the number of cycles used?

As an example, if I enter 45 years FC checks fewer cycles and gives me 100% success because it is not using some of the "ugly" cycles. If I enter 30 years I only get 96%, even though the money does not have to last as long, because it is using more uglies.

I understand the phenomenon ... I'm just wondering if I'm missing some good ideas on how to model the longer retirement period without losing the reality that comes with more cycles.

 

[REWahoo]

Not sure if it helps but here is an old post from the creator of FIRECalc commenting on early retirees and going beyond the 30 year time frame:

First, a 3.98% withdrawal works, using the default settings in FIRECalc, 95% of the time for 30 year withdrawals, while the rate drops to 3.73% for 50 years, and 3.56% for 75 years.

What makes the safe rate as low as it is is almost always what happens in the first 5-7 years of a retirement. If you are lucky enough, or your crystal ball is good enough, to retire into a rising market, then by the time your portfolio has been growing for those 5-7 years, it can withstand almost anything, and your problem will be what to do with all the money, not worrying about the withdrawal rate.

http://www.early-retirement.org/for...rate-for-young-retirees-23422.html#post434788

 

[MasterBlaster]

How does one balance the desire to plan for a potentially long retirement period (say 45 years) with the resulting reduction in the number of cycles used?

As an example, if I enter 45 years FC checks fewer cycles and gives me 100% success because it is not using some of the "ugly" cycles. If I enter 30 years I only get 96%, even though the money does not have to last as long, because it is using more uglies.

I understand the phenomenon ... I'm just wondering if I'm missing some good ideas on how to model the longer retirement period without losing the reality that comes with more cycles.

The other issue is that FIRECALC, using historical data cycles, can only give you a ballpark idea of how likely you are to not run out of money. The answers shouldn't be inspected to the 4th decimal place. I suspect that what you ran into was a sparsely (or sparser) computed answer on the 45 year retirement than the 30 year retirement. In either case FIRECALC is telling you that you most probably will be successful.

In this case the 100% over 45 years and the 96% success over 30 years are approximately equal.

So what if FIRECALC says you'll have a 100% success rate and then during your retirement the Earth is hit by a giant meteor or a new great plague causes worldwide human extinction or at least a collapse of civilization. Wouldn't that be a scream and just mess up all those calculations !

 

[fchelp]

Okay, so I had almost ended my original post with something like "or am i guilty of "measuring something with a micrometer when you'll be cutting it with an axe". Just wondered if there was a "work around" or if I was just working it too hard. Thanks for the replies.

 

[Onward]

There's a pretty good argument for treating all calculated success rates above a certain number, maybe 80%, as meaningless:

The Retirement Calculator from Hell, Part III

 

[fchelp]

So 80% success in firecalc is good enough to pull the plug on, huh?

But is that on a 45 year plan, or a 30 year plan?

 

[REWahoo]

So 80% success in firecalc is good enough to pull the plug on, huh?

But is that on a 45 year plan, or a 30 year plan?

Yes.

I will add, it helps to have a sharp axe....

 

[ERD50]

There's a pretty good argument for treating all calculated success rates above a certain number, maybe 80%, as meaningless:

The Retirement Calculator from Hell, Part III

So 80% success in firecalc is good enough to pull the plug on, huh?

But is that on a 45 year plan, or a 30 year plan?

fchelp - I think you're right to be cynical of that.

I've always argued that that efficient frontier analysis is a little misleading with regards to FIRECALC. Clearly, the 80% and 100% numbers from FC are not just noise level stuff. Of course, it doesn't mean that 80% will be 80% in the future, or that 100% will be 100%.

But FC is not Monte-Carlo, it is reporting on history. So while the numbers might be considered SWAGS as far as looking forward, it is a fact that 4.55% led to 80% success, and 3.59% led to 100% historically. And I certainly expect the lower WR to provide higher relative success going forward, though it may not be 100%. I see that as meaningful.

I look at it as the difference between absolute measurement and relative measurement. I can use a piece of old string to very reliably tell me that this pencil is longer than that pencil. If you want to know to some level of precision what the absolute length is, that gets tougher.


-ERD50

 

[fchelp]

Agreed - all the cautions about using/abusing FIRECALC are understood.

As for the original question: "How does one balance the desire to plan for a potentially long retirement period (say 45 years) with the resulting reduction in the number of cycles used?" - I'm thinking the following might be an interesting exercise, if not a useful workaround - at least in my case.

Since 15 of the 45 years are before SS I can:
- guestimate the income needed above SS to meet expenses during the last 30 years,
- guestimate the cost buying a SPIA 15 years from now to cover that gap,
- cut the FIRECALC years back to 15 years,
- tell FIRECALC to leave enough to pay for the SPIA

There are plenty of potential gotcha's here, including but not limited to:
- the (lack of) accuracy of the "guestimates",
- trusting in SS,
- that I would probably not actually buy the SPIA in 15 years.

But this does give the benefits of lots of cycles, and the theory is if I'm comfortable with the success rate of that 15 year plan, I can be comfortable with the other 30 years. Comfortable, of course, being a relative term. I suppose I could even take the output of the 15 year plan and start a separate 30 year plan. Probably overkill - it wouldn't be the first time I was guilty of that.

I'm going to be away for a bit and unable to actually try this, but I thought I'd get my thoughts down before I forget them.

 

[fchelp]

FYI ... I tried to simulate this and got about the same success rate for the 15 year plan + 30 year SPIA as I did for a 30 year plan - both of which were less than the 45. I guess this is pretty much what I expected.

 

[ERD50]

FYI ... I tried to simulate this and got about the same success rate for the 15 year plan + 30 year SPIA as I did for a 30 year plan - both of which were less than the 45. I guess this is pretty much what I expected.

Stringing two runs together will probably give you a much bleaker outcome than what has happened historically.

The 'failures' in FIRECALC are due to the bad scenarios. When you string two runs together, you 'experience' the worst stuff that ever happened - twice! The 15 Year will not look at 1995 on; The 30 year will not look at 1980 on. Both the 30 year and the 15 year run will include the Great Depression, and the 70's. So assuming those are the worst periods, you get four really bad streaks in one 'strung together' 45 year run.

If you can limit the 15 year run to the years not covered by the 30 year run, that might look better.

-ERD50

 

[fchelp]

Yes, and things look bad enough without doing that.

That is why with my "15 year plan + 30 year SPIA" I did not string two FIRECALC runs together. I just ran a 15 year scenario in FIRECALC that left enough money at the end of the 15 years to buy a SPIA that combined with SS would approximate my spending needs for the next 30 years. There are plenty of SWAGs in there, but this is just one of many different models I'm running (in this and other tools) so I'm not overly concerned with it being anything more than directionally informative.

Thanks for the comments ... much appreciated.

 

[ERD50]

That is why with my "15 year plan + 30 year SPIA" I did not string two FIRECALC runs together. I just ran a 15 year scenario in FIRECALC that left enough money at the end of the 15 years to buy a SPIA that combined with SS would approximate my spending needs for the next 30 years.

Ahhh, gotcha. So a FIRECALC run to get to point A, and then the SPIA from there. Reasonable approach, though I guess we don't know what SPIA payouts will be at the time.

You're right, plenty of SWAG in all this, no matter how you do it. But I think you'll find they tend to converge on a similar range.

-ERD50

 

[Gumby]

But FC is not Monte-Carlo, it is reporting on history.

And that one fact makes a very big difference, as this paper suggests:

http://welton.com/files/resources/Welton-Tail_Risk_5x_Worse__Visual_Insight_Series.pdf

 

[haha]

And that one fact makes a very big difference, as this paper suggests:

http://welton.com/files/resources/Welton-Tail_Risk_5x_Worse__Visual_Insight_Series.pdf

From this paper:

These managers are still drawing much of their performance from underlying equity market return drivers as
noted by their high Betas and correlations, but they are doing so in a more refined and risk controlled fashion.
One reason for this ability is because these managers are not rigidly bound to broad equity market
benchmarks. This grants them useful discretion to both mitigate risk and capitalize on opportunities. For
example, long/short managers may choose to stay in cash during treacherous market conditions. They may​

even be able to profit from market declines due to their ability to go short as well as long.

My heavens, I think they are suggesting market timing. Get out the stakes and faggots, I think we have some burning to do.

Ha