Flexible Retirement Planner Support Forum

Hello,

I like your program. The free service is a wonderful gift.
And better than what my broker is providing (so far).

I do have a question though.

I would like to have approximately zero "median ending portfolio" value.
In order to achieve that, I have to increase the withdrawal amount or would otherwise leave behind a large portfolio. (no heirs)

But as I increase the annual withdrawal, my probability of success declines.

Why is that? I suppose it is because as the value of the portfolio declines, the remaining balance is less likely to generate enough to cover the withdrawals over the remaining years.

Please explain or tell me what further reading I need to do. In my mind, I would like a 90% success factor but end with a small portfolio. Impossible?

Thanks

Thanks for writing about the planner. You've stumbled on one of many weaknesses of using a Monte Carlo approach to try to solve the retirement problem.

Because Monte Carlo simulation is all about probability, the plans most likely to succeed usually have lots of extra padding at the end.

The problem you're trying to solve is a little like cutting the engine on an airplane while you're several miles from the airport, then trying to land and stop the plane on a dime at the midpoint of the airport runway (without using the brakes). It's a really tough problem because there are both unknowable and uncontrollable factors that will influence the final result.

Another thing to consider is that the number that the planner reports for the ending portfolio value is actually a median value. That means half of the 10k simulation runs ended with a lower value and half ended with a higher value. So unlike the high probability of success, this value is only 50/50 bet.

The bogleheads investing forum (a group that's into index investing using Vanguard funds) recently had this interesting thread about the so-called 4% safe-withdrawal rate rule of thumb. They get to the heart of the problem of trying to predict something that's basically unpredictable.

I suppose in the end if you really want to a fixed withdrawal amount that you can't outlive, and immediate annuity is probably the only instrument that can provide this. Unfortunately, the opportunity cost of going for this extra safety is quite high.

Hope this explanation shed at least some light...

Regards,

Jim

So the method really has to leave a lot of money on the table to be statistically "reliable". It is ironic that one has to curtail spending to be sure the money will not run out. Yet, that means the heirs benefit more than the principal person.

Oddly, there seems to be no mention of odds when it comes to predicting age at death. Or, for that matter, if one has a fat Long Term Care package that will completely care for a person for the last 5 years of life. Obviously, all beyond the scope of the issue at hand except the death timing odds should be somehow taken into statistical account beyond just filling in a guesstimate number.

It is compelling to consider the need for funds for various periods. Say, from age 66 to age 80 and from age 66 to age 95. Intuitively, it seems that it would be less risky to draw more from the nest egg if needed for only the 14 year period vs the 29 year period.



It would be useful to know the lo/hi range. Is that possible?
I suppose its a frequency distribution. If so can I know the values by one and two standard deviations, perhaps?


I get that. I'm still hung on the idea of not being able to spend most of my nest egg because I may live a year longer, 2 years longer, etc.

In management consulting we usually advised use of multiple futures planning and it applies here also.


Yep. Thorny problem. Much more difficult than I first thought.


You have!!! Thanks for taking the time with this.

As far as the hi/lo for ending portfolio value.

If the probability of success is less than 100%, the low is $0.

The high for the portfolio value is probably something extreme, but I don't think I show it anywhere in the output.

I'll have to think about whether there's a way to better present this info. At the end of doing 10k simulation runs, I have a list of all 10k ending portfolio values. I could probably do anything I needed to with the list, it's just hard to decide what's important to distill out of it.

Anyhow, I made a note to take a look at this next time I'm in the code.

Regards,

Jim