FIRECalc rewrite

[dory36]

What with new Schiller data and a few annoyances to correct, I'll probably start the next revision to FIRECalc in the next couple of months.

Last time I did this, there were a number of extremely useful suggestions from many on this forum. I invite you to again send suggestions -- if they make sense to me and can be incorporated, they will make it into the next version.

I'll probably "freeze" the changes in about a month (which happens to be when I will be going cruising for a while, and will write software instead of playing with the grandkids when I am not otherwise occupied).

Please send suggestions in this thread, so they will all be together. (If you are in the Witness Protection Program and cannot post your suggestion in public, send a PM with "FIRECalc suggestion" in the subject line.)

Thanks in advance -- dory36

 

[ERD50]

A few (minor) things come to mind. First, thanks for providing FireCalc, it is very useful.

1) The question comes up often, so I suggest that you add something to the SS inputs to make it clearer that you input future SS income in today's dollars, not in inflated $. Firecalc will do the inflation for you (unless I got this wrong, I sure hope not!).

2) I found the entries for pension confusing at first. You label the entry as 'increase/decrease your withdrawals'. OK, if you have money coming in, sure, you will end up decreasing your withdrawals, but it is a round-a-bout way of saying it, and (IMO) confusing to the newbie. Just say something like - 'how much and when will you receive a pension'.

3) It probably has been discussed, but I think it should also be made clearer that as you lengthen the time horizon, you are actually *eliminating* some bad scenarios from the data set. That is, in 2007 a 30 year horizon includes the bad years of the early 1970's (everything pre-1977), but those bad years drop off the analysis for a 40 year horizon (only years 1967 and earlier will be evaluated). This can kind of warp the success rates a bit.

4) Maybe a link to life expectancy tables? And what I would like to see, since most people tend toward a 95% success rate on their finances, why not allow for a 95% (or 5% depending how you look at it) 'success' rate on LE? Most people look at the LE table for their age and that is a 50% number. But few run Firecalc at 50%, so why use 50% for your LE? 95% x 95% is 90.25%. If people want a combined assurance of 95% financial success for their LE, they should probably be using 97.5% for EACH.

I guess those are all just web-based input comments, not really program issues. Hmmm, maybe the ability to rebalance assets over time? I don't know, that will probably just lead to further data-mining.

-ERD50

 

[cute fuzzy bunny]

Are you going to get the number right this time?

 

[Sam]

I invite you to again send suggestions

Please fix your code, so that it'll report 100% success at 6% initial withdrawal rate. I want to quit now, not 3 years from now.

-- if they make sense to me and can be incorporated,

It makes sense, please have it implemented

 

[Lsbcal]

Dory, here are a few suggestions for the new FIRECalc:

1) Include TIPS as a fixed income asset in the "total market" option and the "mixed portfolio" option. Should be easy to do since they TIPS did not exist but we are just talking about putting in a fixed real rate number.

2) In the "mixed portfolio" option also include the 5 yr treasuries that are available with the "total market" options

3) In the "mixed portfolio" please define what "US LT treasuries" are. Are they 30 yr treasuries? Same for "LT Corpororate Bond".

4) In the "mixed portfolio" provide a check so that the numbers must add to 100%. I believe right now there is no error generated if the user makes a mistake here.

Thanks for a great tool!!!

 

[Sam]

4) In the "mixed portfolio" provide a check so that the numbers must add to 100%. I believe right now there is no error generated if the user makes a mistake here.

No check needed. Those don't have to be percentages. Here's Dory's explanation:

(You may enter actual dollar amounts in each class, or relative amounts. Whatever you enter will be converted to a percentage of the total, and performance will be calculated on each asset proportionally.)

 

[SecondCor521]

It hasn't bit me but I know there was a thread about entering a starting year in the recent past messes up the software (e.g. entering 2005 as the starting year). You know what I mean.

Also, as an aside to ERD50 on point #4, I'm pretty sure you have the math wrong. A "real" failure only happens when you run out of money *and* are still alive. If you use a 95% success rate for FIREcalc and a 95th percentile life expectancy, then the P(running out of money) = 5% and the P(still being alive) = 5%. Assuming they are independent variables, P(both) = 5% * 5% = 0.25%. Since P(success) = 1 - P(failure), P(success) is 99.75%.

Oh, and can you fix it so that it reports 100% success for a withdrawal rate of 414.07%? I'd like to retire now, not 15 years from now ;-P

2Cor521

 

[Sam]

Seriously now.

Please consider enhancing the Bernicke modeling. Allow the user to specify his own personal inflation rate for at least 3 period. The specify personal inflation rate is relative to the historical value. That is, [+1%] means that the personal inflation rate is 1% more than the historical. [-1%] means 1% less.

Suggested user interface:

First period: From [start] to [62] = [-1%]
2nd period: From [63] to [70] = [0%]
3rd period: From [71] to [end] = [+2%]

Thanks again for a great tool.

 

[Gotadimple]

Oh, and can you fix it so that it reports 100% success for a withdrawal rate of 414.07%? I'd like to retire now, not 15 years from now ;-P

2Cor521

And while you're at it, Dory, can you also arrange for it to send a rudely worded "So Long Sucker" letter to my boss? Anonymously, of course

-- Rita

 

[Cb]

3) It probably has been discussed, but I think it should also be made clearer that as you lengthen the time horizon, you are actually *eliminating* some bad scenarios from the data set. That is, in 2007 a 30 year horizon includes the bad years of the early 1970's (everything pre-1977), but those bad years drop off the analysis for a 40 year horizon (only years 1967 and earlier will be evaluated). This can kind of warp the success rates a bit.
-ERD50

Would it be reasonable to include partial periods up through, I dunno, 1977 or so on the longer (>30 year) cases, just to subject them to that rough patch?

I think most ER's should be looking at retirement durations longer than 30 years, but omitting the late 60's and early 70's on the 35+ year runs tends to paint a bit rosier picture as ERD50 notes.

Cb

 

[Cb]

Just remembered one...would it be a major PITA to allow a gradual shift toward a higher FI allocation as one goes through retirement, maybe at 5 or 10 year intervals?

Cb

 

[ERD50]

Would it be reasonable to include partial periods up through, I dunno, 1977 or so on the longer (>30 year) cases, just to subject them to that rough patch?

I think most ER's should be looking at retirement durations longer than 30 years, but omitting the late 60's and early 70's on the 35+ year runs tends to paint a bit rosier picture as ERD50 notes.

Cb

But what do you fill in the extra years with? The problem is you only have, 30 sequential years to work with from 1977 to 2007, so when you want a 40 year run, how do you account for the last 10 years of drawdown? Seems like you would need to 'roll back' time and stitch another ten years on there - but from where?

-ERD50

 

[Cb]

But what do you fill in the extra years with?
-ERD50

The original version of FIREcalc included all of the full 30 year runs as well as all of the "partial periods", but Dory dropped those results in the newer version because they often boosted SWR percentages slightly for <100% runs. Think of it this way:

If you were analyzing a 30 year retirement shooting for less than 100% safety, once you cleared 1976 you were home free...but by adding the additional 29 successful 1977, 1978...,2002, 2003, 2004, etc shorter runs the averages were higher than they'd have been otherwise.

 

[ERD50]

Also, as an aside to ERD50 on point #4, I'm pretty sure you have the math wrong. A "real" failure only happens when you run out of money *and* are still alive. If you use a 95% success rate for FIREcalc and a 95th percentile life expectancy, then the P(running out of money) = 5% and the P(still being alive) = 5%. Assuming they are independent variables, P(both) = 5% * 5% = 0.25%. Since P(success) = 1 - P(failure), P(success) is 99.75%.
2Cor521

Yes, you are correct as you state it. I was thinking in terms of NOT extending the time horizon in FireCalc and the effect that a 95% LE would have on that (it would result in more failures).

My math is still not really correct though, as FireCalc is not formulaic, it is data driven.

The key to your statement (and I overlooked it the first two times I read it) is ' using the 95% LE time horizon in FireCalc at a 95% success rate'.

So yes, applying a 95% chance of NOT making to that higher age, AND a 95% of NOT running out of money with that longer horizon factored in, you get the 99.75% you give.

Bottom line, if someone uses a 95% FireCalc success rate on a median LE time horizon, there will be additional chances for failure if they live longer than the median - and 50% will.

-ERD50

 

[ERD50]

The original version of FIREcalc included all of the full 30 year runs as well as all of the "partial periods", but Dory dropped those results in the newer version because they often boosted SWR percentages slightly for <100% runs. Think of it this way:

If you were analyzing a 30 year retirement shooting for less than 100% safety, once you cleared 1976 you were home free...but by adding the additional 29 successful 1977, 1978...,2002, 2003, 2004, etc shorter runs the averages were higher than they'd have been otherwise.

Right. I don't think there is really a 'solution' as such. You only have so much data to work with.

But I did realize that this makes comparing the success rate of a 30 year run to a 40 year run a little , ummm, non-comparable? People should just be aware of that, I think.

I actually went through the effort on a couple runs to take the spreadsheet data apart, and exclude the 'extra' years from the runs in the 30 year data, to see just how much worse a 40 year run would be on the same data set.

At that point I realized that I was into data-mining territory, and decided in the big view that a 4% SWR seems pretty reasonable, and getting down near 3% should be just about bullet proof. And if it isn't, we will probably have bigger things to worry about!

-ERD50

 

[clifp]

My number one issue is no way to account for international stock funds.

I realize that data isn't as complete as domestic, but with many of us having 10-40% (I'm the low side) of our assets invested international, there should be someway to account for it.

My second suggestion would be allow the modeling of porfolio reset. For instance your $2 million portfolio increased to 2.5 million five years after retiring. You can in theory reset your with withdrawal to $100K, but you take a higher risk of running out of money. I'd like to be able to understand that risk using FIRECALC

 

[Rich_by_the_Bay]

Great resource, Dory. Thanks for staying with it.

1. I'd like to see more checkboxes and fewer radio buttons on the results page. For example, I might like to see the success rate AND what if I retire in any year over the next xx years AND what minimal nest egg amt do I need. Currently you have to break them up over several runs (either/or) so it gets a little fragmented.

2. Another interface thought: under other income options ("increase/decrease your withdrawal by...") if you have temporary income streams (e.g. part time between ages 60 and 65, then stop) you now have to enter a Decrease, then a separate Increase at a later date to neutralize it. Even one such stream uses 2 of the 3 lines available. Perhaps putting a start and stop date for each item would allow more flexibility (where no stop date = indefinite). Another common use is to add in an INCREASE for health insurance which drops greatly at age 65.

3. Wishful: what if you burn through cash and bonds first, then switch to equities when they are depleted. Recent research points to that as the optimal strategy, and it's also what Buckets of Money favors. The results might be very surprising. I can search for the reference in J Finan Plannning (last few months) but don't have it handy.

Thanks again.

Rich

 

[ERD50]

Rich - very good ones (in post # 17),

esp # 2 (start/stop dates for income/withdraw changes) and Helath ins cost itemized.

and #3 ( withdraw bonds first, then stocks).

Do some other 'advanced' calc allow this option #3?

-ERD50

 

[Cb]

Right. I don't think there is really a 'solution' as such. You only have so much data to work with.

But I did realize that this makes comparing the success rate of a 30 year run to a 40 year run a little , ummm, non-comparable? People should just be aware of that, I think.

I actually went through the effort on a couple runs to take the spreadsheet data apart, and exclude the 'extra' years from the runs in the 30 year data, to see just how much worse a 40 year run would be on the same data set.

-ERD50

I agree...and I think we're more likely to see something like the 70's malaise & inflation again than a depression-like scenario, so if I was The Bossman of FIREcalc I'd be inclined to subject the >30 year runs to the '66-'7n period, and drop the subsequent partial periods. Better to be a bit on the conservative side, I think.

Cb O0

 

[twaddle]

How about either:

1) Open Source FC so we can come up with our own tweaks.

or

2) Add a simple scripting language so we can play with our own withdrawal strategies, [-]market timing[/-] variable allocation strategies, success criteria, and report generation.

 

[Rich_by_the_Bay]

Here's the article I mentioned above - it's about drawing down your bonds first in the decumulation phase.

 

[ERD50]

Here's the article I mentioned above - it's about drawing down your bonds first in the decumulation phase.

Rich, if I read that right, the results should not be surprising at all. It looks like he is using a totally static model. He just assumes stocks return the same higher real rates than do bonds. Same numbers every year.

In that case, sure it makes sense to spend down the bonds first. You want to give the highest return investment the longest amount of time to grow (other things being equal).

So it would be interesting to see what would happen with the data set in FireCalc.

-ERD50

 

[Rich_by_the_Bay]

Rich, if I read that right, the results should not be surprising at all. It looks like he is using a totally static model. He just assumes stocks return the same higher real rates than do bonds. Same numbers every year.

In that case, sure it makes sense to spend down the bonds first. You want to give the highest return investment the longest amount of time to grow (other things being equal).

So it would be interesting to see what would happen with the data set in FireCalc.

Yup. Don't want to ambush Dory's thread, but I agree. However, for this to work, the asset types need to be held long enough to allow smoothing of volatility so, for example, 2 years of bond burn, then all stocks forever would not be as likely to survive as 7 or 10 or 15 years of fixed before tapping the stocks.

Boy would that be awesome to have within FC. Unfortunately, after bonds are burned you have additional choices: just start digging into stocks forever, or immediately convernt $xxx of stocks into bonds to start burning again (buckets of money approach).

 

[tightasadrum]

Along the same lines as Rich mentioned, how about,

1. more lines for reduced withdrawal in income page.

I have several investments in real estate. Unless I group then together as if I'm selling it all, I don't have needed flexibility. How complex does adding several more lines make the program? Four or five more lines would helpful.

2. The "how invested" page is ackward to me. How about putting down a table of all major catagories with input for % allocation. I'd like to see a catagory for REITS and commodity futures as well. Are TIPS on there now? Can't remember, but probably are.

 

[Cadence]

Report median rather than average

Dory,

Please consider using median values of portfolio results rather than averages on the results page. The distribution of portfolio returns over time is not a bell curve, it is a log normal distribution. Many scenarios will return less than the average, a few will return much larger than the average. In this case, the median is less, sometimes much less than the average. Therefore it is unlikely than any investors specific portfolio return will perform at or better than the average ending value of all scenarios.

If it were me I'd like to see the results page displayed something like:

50% probability of portfolio ending $900,00 or greater
80% probability of portfolio ending $500,00 or greater
90% probability of portfolio ending $100,00 or greater
95% probability of portfolio ending $50,00 or greater
etc.

It's my guess that its not understood that although a portfolio many have a 95% survivability rate, because of log normal distribution, a great many scenarios will move perilously close to 0.

I'll say again, I think firecalc is excellent. But I do believe that it gives users an unrealistic level of confidence.

Dory - I'm willing go into more detail on my thoughts on this if you are interested.