Author Topic: Rich, Broke or Dead: Visualizing probabilities of outcomes in early retirement  (Read 69363 times)

boarder42

  • Walrus Stache
  • *******
  • Posts: 9332
Yup, radical life extension (that works and is available to many people) definitely falls into the that category of black swan/long tail events where historical data probably would lose most of its ability to forecast the future.

If the reality of a person's body or mind being too old to work went away, would the concept of regular retirement vanish entirely? Or would people continue to retire and the balance of workers to retirees would rapidly spiral out of control? (If timed perfectly this could work well with the accelerating pace we're seeing blue and white collar jobs being replaced by robotics and AI respectively.)

In the past increases in life expectancy in a given country tend to be correlated with declines in birth rates (usually with a lag time of a generation or so, so you still see a dramatic uptick in population). Would that relationship hold true if lifespans doubled or tripled?

google's futurist predicted the millenial generation will be the first to have to choose to die.  be really cool to see if that comes true.  Death parties.  unlike weddings and the other things you can waste money on to hurt you - a big blow out bash where you give away money and throw one helluva party followed by killing yourself in some crazy way.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9922
  • Registered member
Re-reading my unsolicited rant from last night :) I think I was trying to get around to asking for a 'Mustachian' button.  It would be really awesome to toggle between statistical average (current version) to an optimized outcome (college educated, married, cope well with stress, etc for longvity + able to reduce withdrawal 25% during bear market, AA annually re-balanced, etc.)...  Just for fun.  Maybe it would nudge people in a good direction toward financial education and being more confident to ER.

I think the easiest way to handle life expectancy is just to have the system auto-populate the number it's using now, and let the user change it to anything they want.  Go nuts with a third-party LE calculator or just guess.  It's the least work for the creator with maximum flexibility

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
google's futurist predicted the millenial generation will be the first to have to choose to die.  be really cool to see if that comes true.  Death parties.  unlike weddings and the other things you can waste money on to hurt you - a big blow out bash where you give away money and throw one helluva party followed by killing yourself in some crazy way.

I mean in all fairness I could choose to do die that way (complete with blowing all of my savings on a ridiculous party) today or tomorrow. ;-)

But yes, one way or another it's going to be fascinating to see what the next several decades have in store. And if google's futurist is right, perhaps see what the next several centuries have in store.

Rubic

  • Handlebar Stache
  • *****
  • Posts: 1130
My preference would be sentenced to death in the manner of Socrates for corrupting
the youth, but without blowing my savings on a farewell party.

boarder42

  • Walrus Stache
  • *******
  • Posts: 9332
google's futurist predicted the millenial generation will be the first to have to choose to die.  be really cool to see if that comes true.  Death parties.  unlike weddings and the other things you can waste money on to hurt you - a big blow out bash where you give away money and throw one helluva party followed by killing yourself in some crazy way.

I mean in all fairness I could choose to do die that way (complete with blowing all of my savings on a ridiculous party) today or tomorrow. ;-)

But yes, one way or another it's going to be fascinating to see what the next several decades have in store. And if google's futurist is right, perhaps see what the next several centuries have in store.

i dont know that its fair or right to stay alive for the sake of being alive. and depleting natural resources.

TVRodriguez

  • Pencil Stache
  • ****
  • Posts: 773

Thank you!  so, we're all gonna die, huh?

Hahaha! 

Seriously, though, I love this calculator.  The visual is so striking, and it's much easier for "regular" people to use and understand at a glance than others, like FIREcalc (which I also love).

boarder42

  • Walrus Stache
  • *******
  • Posts: 9332
My preference would be sentenced to death in the manner of Socrates for corrupting
the youth, but without blowing my savings on a farewell party.

i wasnt saying blow it all i was saying give it all away and have some to fund the party.  whether its to heirs or your favorite charities.  We're all going to die with mulitple millions thats the likely case here.

Roland of Gilead

  • Handlebar Stache
  • *****
  • Posts: 2454
What will likely happen, to prevent massive overpopulation if a cure for aging happens, is forced birth control.  Maybe to have a kid, you have to agree to terminate yourself at some future time.

Otherwise in a few centuries the earth would just be a shithole.

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming
Sorry, I've been (and still am) on vacation and haven't had time to check on the thread.  But I did find a few minutes to do a few things:


1- I updated the tool so that you can now toggle between nominal and inflation adjusted values.  You can see that the inflation adjusted (real) metrics for success look worse than the nominal metrics (i.e. > than initial balance or 2x initial balance). 
2 - I checked twitter and noticed that Jacob (the guy from ERE) tweeted my tool out to his followers and posted it to the ERE forums.  Nice to see some acknowledgement from one of the founders of the FIRE movement:  https://twitter.com/extremejacob/status/1010301383034245126
3- I updated the site to https.  I may have broken some things for a few hours, but hopefully it's all working now.


I'll try to add more items from the list as I find a few min/hours on my vacation.

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
Congrats, CCCA! It's exciting to see your tool getting name dropped by one of the two people who (at least from my personal perspective) really kicked off the modern FIRE movement.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9922
  • Registered member
Thanks for adding inflation, cool tool.  Ur in the Bay Area I’ll buy you a beer

HAPPYINAZ

  • Bristles
  • ***
  • Posts: 382
This is really cool, thank you!


PhilB

  • Walrus Stache
  • *******
  • Posts: 5700
Freakin' awesome!  You sir are an absolute star!

Huffduf41

  • 5 O'Clock Shadow
  • *
  • Posts: 63
Trying to wrap my head around how the likelihood of me being broke increases when I toggle off the death curve.  Is this just so everything adds up to 100%?

It's like the performance of the stock market all depends on me being alive....definitely not true.

DreamFIRE

  • Handlebar Stache
  • *****
  • Posts: 1593
Trying to wrap my head around how the likelihood of me being broke increases when I toggle off the death curve.  Is this just so everything adds up to 100%?

It's like the performance of the stock market all depends on me being alive....definitely not true.

When you uncheck death, that takes death out of the equation.  If you're not dead, it's down to one of the other results.

Huffduf41

  • 5 O'Clock Shadow
  • *
  • Posts: 63
But why would me being alive impact the 4% rule (what I plugged in for spend and portfolio) failing?

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
But why would me being alive impact the 4% rule (what I plugged in for spend and portfolio) failing?

The percent chance of being broke is the chance that you run out of money either before you die or before you reach the maximum age in the simulation.

If you knew with absolute certainty you were going to die in 6 months, you could spend 150% of your net worth per year and have no risk of running out of money.

FIRE 20/20

  • Pencil Stache
  • ****
  • Posts: 758
But why would me being alive impact the 4% rule (what I plugged in for spend and portfolio) failing?

Let's say at some point you have the following percentages at age 85:
2x - 10%
>start and <2x - 10%
less than start - 10%
broke - 10%
dead - 60%

If you know for certain that you have invented a miracle anti-death medicine and an Ironman suit that prevents death from accidents, then you know you won't die by then.  What are the chances of the other situations?  They should each be 25% now.  It's a little like rolling a die.  If you have a normal 6 sided die, there's a 1/6 chance of rolling each number.  If you want a 1/4 chance, how can you get that from a standard die?  You could just say that you'll completely ignore and re-roll any 5s and 6s that you roll.  Now the odds of rolling a 1, 2, 3, and 4 are 1/4.  In that example you have to completely ignore the 5s and 6s - if they come up you need to just pretend they didn't happen. 

EscapeVelocity2020

  • Magnum Stache
  • ******
  • Posts: 4815
  • Age: 50
  • Location: Houston
    • EscapeVelocity2020
Sorry, I've been (and still am) on vacation and haven't had time to check on the thread.  But I did find a few minutes to do a few things:
1- I updated the tool so that you can now toggle between nominal and inflation adjusted values.  You can see that the inflation adjusted (real) metrics for success look worse than the nominal metrics (i.e. > than initial balance or 2x initial balance). 

Please give some info on the inflation toggle, i.e. is this in line with a historical inflation of the last 'x' years (x being retirement time-frame selected), or simply fixed current inflation (~2%), historical CPI, historically estimated chained-CPI?.

Is ER healthcare accounted for, because that is definitely higher than 2% inflation.

Thanks for the work you are putting in to this!

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming
Sorry, I've been (and still am) on vacation and haven't had time to check on the thread.  But I did find a few minutes to do a few things:
1- I updated the tool so that you can now toggle between nominal and inflation adjusted values.  You can see that the inflation adjusted (real) metrics for success look worse than the nominal metrics (i.e. > than initial balance or 2x initial balance). 

Please give some info on the inflation toggle, i.e. is this in line with a historical inflation of the last 'x' years (x being retirement time-frame selected), or simply fixed current inflation (~2%), historical CPI, historically estimated chained-CPI?.

Is ER healthcare accounted for, because that is definitely higher than 2% inflation.

Thanks for the work you are putting in to this!

Yes, inflation is historically indexed and linked to the same annual data for stock and bond returns.  There is currently no ER healthcare accounted for. 

Trying to wrap my head around how the likelihood of me being broke increases when I toggle off the death curve.  Is this just so everything adds up to 100%?

It's like the performance of the stock market all depends on me being alive....definitely not true.

Others have answered and in skimming them, I think the answers are correct but I thought I'd add my own, hopefully, clear explanation.  The main thing is that the retirement balance probabilities (i.e. balance > start, balance< start, balance <0) is that they are all assuming that you are alive.

Conditional probability is the probability of event B occurring assuming that event A will occur (or has already occurred).  If you die with money left, you cannot by definition, become broke later (i.e. probability of going broke is 0%, if you die).  The probability of going broke, if you are alive until you are 80 to 100 is non-zero for many ER scenarios. 

Thus removing the death wedge is like looking at the conditional probabilities of various outcomes of your retirement balance assuming you are alive throughout the entire period of interest.  Including the death wedge, you are multiplying the conditional probabilities by the probability of being alive.  If you have a 50% chance of being alive in a given year, then your likelihood of becoming broke cannot be over 50%.

Not sure if that is clearer or not, but adding another explanation can hopefully help people who didn't get it after the other explanations.

EscapeVelocity2020

  • Magnum Stache
  • ******
  • Posts: 4815
  • Age: 50
  • Location: Houston
    • EscapeVelocity2020
Sorry, I've been (and still am) on vacation and haven't had time to check on the thread.  But I did find a few minutes to do a few things:
1- I updated the tool so that you can now toggle between nominal and inflation adjusted values.  You can see that the inflation adjusted (real) metrics for success look worse than the nominal metrics (i.e. > than initial balance or 2x initial balance). 
Please give some info on the inflation toggle, i.e. is this in line with a historical inflation of the last 'x' years (x being retirement time-frame selected), or simply fixed current inflation (~2%), historical CPI, historically estimated chained-CPI?.

Is ER healthcare accounted for, because that is definitely higher than 2% inflation.

Thanks for the work you are putting in to this!
Yes, inflation is historically indexed and linked to the same annual data for stock and bond returns.  There is currently no ER healthcare accounted for. 

Please toss out the reference you used for your historical inflation adjustment.  Thanks!

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming
I have a question for folks here, who might help me with some of the logic for the pension/social security addition to the calculator. 


If I want to add a pension/social security value of $20,000 per year (and adjusted for inflation annually) to the calc when the user is aged 67, how would I expect the user (who might retire at age 40) to enter the value of this pension/SS benefit?  Are they entering the nominal dollars 27 years in the future or are then entering a nominal amount now that I need to calculate 27 years of inflation for? 


Will most people be getting estimates of the 2018 value of these future benefits or the in-year estimate (say 2045) value?


thanks.




Systems101

  • Stubble
  • **
  • Posts: 218
If I want to add a pension/social security value of $20,000 per year (and adjusted for inflation annually) to the calc when the user is aged 67, how would I expect the user (who might retire at age 40) to enter the value of this pension/SS benefit?  Are they entering the nominal dollars 27 years in the future or are then entering a nominal amount now that I need to calculate 27 years of inflation for?

It's closer to present value, but technically, it's neither of the two options you provided.  Salary under age 60 is indexed (but by average age index not by inflation) up until age 60, above age 60, the index is always 1.0.  See the example and text on the left side of this page for more detail: https://www.ssa.gov/oact/progdata/retirebenefit1.html

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
Please toss out the reference you used for your historical inflation adjustment.  Thanks!

From CCCA's website:

Quote
Data source and Tools Historical Stock/Bond and Inflation data comes from Prof. Robert Shiller.

DreamFIRE

  • Handlebar Stache
  • *****
  • Posts: 1593
I have a question for folks here, who might help me with some of the logic for the pension/social security addition to the calculator. 


If I want to add a pension/social security value of $20,000 per year (and adjusted for inflation annually) to the calc when the user is aged 67, how would I expect the user (who might retire at age 40) to enter the value of this pension/SS benefit?  Are they entering the nominal dollars 27 years in the future or are then entering a nominal amount now that I need to calculate 27 years of inflation for? 


Will most people be getting estimates of the 2018 value of these future benefits or the in-year estimate (say 2045) value?

I always calculate my SS benefits in "today's dollars".  That is the default, although there are options on some of the ssa.gov calculators for "future (inflated) dollars".  In addition to the wage indexing mentioned, beginning at age 62, the benefit will increase by COLA each year, which is based on CPI-W, whether you have started taking benefits or not, not to be confused with the delayed retirement credit.  So, if there was an option, I would pref to plug in my benefit for the year/age I expect to start taking benefits using the benefit calculated in today's dollars that I got from ssa.gov or the offline calculator and have your graph handle the compensation to future dollars for my FIRE date and beyond.
« Last Edit: June 30, 2018, 12:25:22 PM by DreamFIRE »

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming
Here is a beta (not official), early preview version of the updated tool, (this link doesn't work anymore, go to the original link for the latest version) which now includes investment fees, taxes and social security/pension.  I made some slight modifications to the code so things won't exactly line up with the old model with taxes, investment fees, and social security set to 0.

https://engaging-data.com/will-money-last-retire-early/

Social security is assumed to be in present dollars and inflated by the cumulative inflation rate until the age you take it.  However, I realize there's no year assumed here.  Year 1 of your retirement is assumed to be today, even if your starting retirement age is in the future.

As always, any suggestions, comments, etc are appreciated!

« Last Edit: July 30, 2018, 01:37:19 AM by CCCA »

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9922
  • Registered member
You are doin a lot of work to accommodate options, which is great, but here’s yet another suggestion:

Hide some of the options behind an “advanced options” menu or something to keep the basic tool unintimidating

Or I guess you could just publish two separate tools, basic and advanced

But if it gets too cluttered, more basic users might get scared away

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming
You are doin a lot of work to accommodate options, which is great, but here’s yet another suggestion:

Hide some of the options behind an “advanced options” menu or something to keep the basic tool unintimidating

Or I guess you could just publish two separate tools, basic and advanced

But if it gets too cluttered, more basic users might get scared away
Yeah, I’m trying to figure out how to not make this turn into firecalc or cfiresim. I thought of the idea of hiding things but maybe making a simple version is better.

But I do want it to be useful so I’d like to add things that can help people without scaring away others.

DreamFIRE

  • Handlebar Stache
  • *****
  • Posts: 1593
There something I noticed playing around with SS.  I set my retirement horizon to age 90.  I got 100% initially regardless of what age I would take SS benefits, but when I scaled down my stash by $250K to drop my success rate to the high 90's% just for testing, I noticed my success rate using age 62 SS benefits gave me a slightly higher success rate than using my age 67 SS benefits (based on SS calculator and present day dollars.)  This was unexpected for an age 90 time horizon, which is well past the break-even for delayed SS benefits.  When I use those same figures in cFIREsim, I get slightly better success percentage by delaying SS to 67.  I just confirmed that again now using the same age ranges, retirement length, allocation, and such comparing age 62 vs. age 67 benefits.
« Last Edit: July 02, 2018, 08:17:54 PM by DreamFIRE »

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
Are you including the death curve?

If so, it makes sense that you'd get different results because in cFiresim, you're surviving until 90 100% of the time, and with CCCA's website, sometimes (fairly frequently) you die in 60s or 70s, and in that subset of scenarios taking social security early will either prove to be the better choice, or you'll have accumulated enough money through compounding that regardless of which choice you make, you won't run out of money.

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming
There something I noticed playing around with SS.  I set my retirement horizon to age 90.  I got 100% initially regardless of what age I would take SS benefits, but when I scaled down my stash by $250K to drop my success rate to the high 90's% just for testing, I noticed my success rate using age 62 SS benefits gave me a slightly higher success rate than using my age 67 SS benefits (based on SS calculator and present day dollars.)  This was unexpected for an age 90 time horizon, which is well past the break-even for delayed SS benefits.  When I use those same figures in cFIREsim, I get slightly better success percentage by delaying SS to 67.  I just confirmed that again now using the same age ranges, retirement length, allocation, and such comparing age 62 vs. age 67 benefits.
It could be because of how I calculate SS benefits in a given year. I can’t remember how much more you get each year you delay SS but it is substantial as I understand it. However since inflation has been high in the past, maybe in a lot of cases taking it at 62 gets you as much (or almost as much) money when you are 67 as the SS calculator tells you. But you also get the money from 62 to 67.

Just speculating here but that could be why taking at 62 is better than 67 in this historical cycle context. If we think inflation won’t ever be higher than 2-3% going forward that  may not hold in the future.

DreamFIRE

  • Handlebar Stache
  • *****
  • Posts: 1593
Are you including the death curve?

If so, it makes sense that you'd get different results because in cFiresim, you're surviving until 90 100% of the time, and with CCCA's website, sometimes (fairly frequently) you die in 60s or 70s, and in that subset of scenarios taking social security early will either prove to be the better choice, or you'll have accumulated enough money through compounding that regardless of which choice you make, you won't run out of money.

I had the death curve turned off during my earlier testing which had given those results.  But I just tested it again with the death curve enabled, and while the graph looks different with that increasing amount of grey, the death curve didn't change my success rate percentage (where it shows at the top above the graph, but enabling death does change the broke and other percentages where I hover the mouse pointer over the graph on the far right).
« Last Edit: July 02, 2018, 10:51:57 PM by DreamFIRE »

DreamFIRE

  • Handlebar Stache
  • *****
  • Posts: 1593
There something I noticed playing around with SS.  I set my retirement horizon to age 90.  I got 100% initially regardless of what age I would take SS benefits, but when I scaled down my stash by $250K to drop my success rate to the high 90's% just for testing, I noticed my success rate using age 62 SS benefits gave me a slightly higher success rate than using my age 67 SS benefits (based on SS calculator and present day dollars.)  This was unexpected for an age 90 time horizon, which is well past the break-even for delayed SS benefits.  When I use those same figures in cFIREsim, I get slightly better success percentage by delaying SS to 67.  I just confirmed that again now using the same age ranges, retirement length, allocation, and such comparing age 62 vs. age 67 benefits.
It could be because of how I calculate SS benefits in a given year. I can’t remember how much more you get each year you delay SS but it is substantial as I understand it. However since inflation has been high in the past, maybe in a lot of cases taking it at 62 gets you as much (or almost as much) money when you are 67 as the SS calculator tells you. But you also get the money from 62 to 67.

Just speculating here but that could be why taking at 62 is better than 67 in this historical cycle context. If we think inflation won’t ever be higher than 2-3% going forward that  may not hold in the future.

I normally hear 8% per year.  In my case the age 67 benefit was 42% higher than the age 62 benefit in today's dollars based on the ssa.gov offline calculator.

Hvillian

  • Stubble
  • **
  • Posts: 150
Here is a beta (not official), early preview version of the updated tool, which now includes investment fees, taxes and social security/pension.  I made some slight modifications to the code so things won't exactly line up with the old model with taxes, investment fees, and social security set to 0.

[ . . .]

CCCA - I love playing with this updated version and have shared it with a few friends and family.  I have been using the social security feature to model having a part time salary, since it will let you use whatever age you want as a start age.  I think it is really fun to see how well the stash holds up if you continue to work to replace various portions of your spending.  My only requested enhancement would be a separate input line similar to the social security for other income with inputs for Starting age and Ending age.   

I've also shared the Market Timing game with a few coworkers as they debate pulling out of the market every week or two.  Engaging Data indeed.

letsdoit

  • Bristles
  • ***
  • Posts: 405
isn't it always better to start it at age 67?

Mississippi Mudstache

  • Handlebar Stache
  • *****
  • Posts: 2170
  • Age: 40
  • Location: Danielsville, GA
    • A Riving Home - Ramblings of a Recusant Woodworker
isn't it always better to start it at age 67?

No. What if you die when you're 66?

Telecaster

  • Magnum Stache
  • ******
  • Posts: 3551
  • Location: Seattle, WA
isn't it always better to start it at age 67?

Hotly debated question.  Bottom line it comes down to how long you live. 

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
isn't it always better to start it at age 67?

More specifically the government has structures SS so that for the average person the total lifetime payout is about equal regardless of when you take it. But you have lots of inside information on your own health/family history/genetic risk factors. So if, based on your insider info you're likely to live longer than the average person who makes it to retirement age (which is already longer than the average person at birth), it probably makes sense to delay. If you're likely to live less long than the average person who makes it to retirement age, you can maximize your expected total payout by taking SS early.

There's a second level of reasoning where delaying SS acts as longevity insurance if you live a lot longer than you had expected, which will be more or less important based on the rest of your assets/income/retirement expenses.

CCCA

  • Pencil Stache
  • ****
  • Posts: 631
  • Location: Bay Area, California
  • born before the 80's
    • FI programming

CCCA - I love playing with this updated version and have shared it with a few friends and family.  I have been using the social security feature to model having a part time salary, since it will let you use whatever age you want as a start age.  I think it is really fun to see how well the stash holds up if you continue to work to replace various portions of your spending.  My only requested enhancement would be a separate input line similar to the social security for other income with inputs for Starting age and Ending age.   

I've also shared the Market Timing game with a few coworkers as they debate pulling out of the market every week or two.  Engaging Data indeed.


Thanks!  Having one or two additional income sources is fine and I can easily code that in.  I think my challenge is that I'm trying to keep it from blowing up to an indeterminate number of income and spending sources like cFIREsim.  And it's not a coding challenge but more of an interface challenge*. 


*both from a visual clutter perspective as well as a web design issue (where adding more things makes the calculator bigger to the point where it doesn't fit on the screen anymore). 


I have been thinking about other updates.  I started to code up a spousal survival probability combination (probability of you dying and your spouse dying is where you don't need money anymore), but I kinda got into a bit of a complicated mess.  The challenge is that both spending and income (i.e. SS) can change depending on if you or your spouse die.  And since the outputs are probability based I guess I'd need to gather all the inputs (spending and income) and then calculate probabilities of running out of money for each of the possibilities (both live, you live, spouse lives).  It's definitely doable but I have to re-do the structure of the calc alot.


Anyway, this is just me thinking out loud mostly.  I should have time over the next few weeks to look at some of these issues and can probably do some of the easy things quickly.  I think I'm not going to make this a full-fledged cFIREsim replacement, but I can replicate some of the main elements without too much more effort.


By the way, I just got back from a month-long vacation roadtrip with the family.  It was nice to be out and about in the Western US, lots of beautiful mountain and deserts landscapes.
« Last Edit: July 27, 2018, 11:38:51 PM by CCCA »

cerat0n1a

  • Handlebar Stache
  • *****
  • Posts: 2319
  • Location: England
Re-reading my unsolicited rant from last night :) I think I was trying to get around to asking for a 'Mustachian' button.  It would be really awesome to toggle between statistical average (current version) to an optimized outcome (college educated, married, cope well with stress, etc for longvity + able to reduce withdrawal 25% during bear market, AA annually re-balanced, etc.)...  Just for fun.  Maybe it would nudge people in a good direction toward financial education and being more confident to ER.

I think the easiest way to handle life expectancy is just to have the system auto-populate the number it's using now, and let the user change it to anything they want.  Go nuts with a third-party LE calculator or just guess.  It's the least work for the creator with maximum flexibility

Another (big) thank you and a request for this ^. (Those of us in developed countries outside the US will typically have higher life expectancy.)

DreamFIRE

  • Handlebar Stache
  • *****
  • Posts: 1593

Don't forget to remove taxes from your spending if you're using the average tax rate option.

With cFireSim, I use the option to increase my spending at age 65 due to the higher cost of Medicare Parts/supplemental.

Much Fishing to Do

  • Handlebar Stache
  • *****
  • Posts: 1140

CCCA - I love playing with this updated version and have shared it with a few friends and family.  I have been using the social security feature to model having a part time salary, since it will let you use whatever age you want as a start age.  I think it is really fun to see how well the stash holds up if you continue to work to replace various portions of your spending.  My only requested enhancement would be a separate input line similar to the social security for other income with inputs for Starting age and Ending age.   

I've also shared the Market Timing game with a few coworkers as they debate pulling out of the market every week or two.  Engaging Data indeed.


Thanks!  Having one or two additional income sources is fine and I can easily code that in.  I think my challenge is that I'm trying to keep it from blowing up to an indeterminate number of income and spending sources like cFIREsim.  And it's not a coding challenge but more of an interface challenge*. 


*both from a visual clutter perspective as well as a web design issue (where adding more things makes the calculator bigger to the point where it doesn't fit on the screen anymore). 


I have been thinking about other updates.  I started to code up a spousal survival probability combination (probability of you dying and your spouse dying is where you don't need money anymore), but I kinda got into a bit of a complicated mess.  The challenge is that both spending and income (i.e. SS) can change depending on if you or your spouse die.  And since the outputs are probability based I guess I'd need to gather all the inputs (spending and income) and then calculate probabilities of running out of money for each of the possibilities (both live, you live, spouse lives).  It's definitely doable but I have to re-do the structure of the calc alot.


Anyway, this is just me thinking out loud mostly.  I should have time over the next few weeks to look at some of these issues and can probably do some of the easy things quickly.  I think I'm not going to make this a full-fledged cFIREsim replacement, but I can replicate some of the main elements without too much more effort.


By the way, I just got back from a month-long vacation roadtrip with the family.  It was nice to be out and about in the Western US, lots of beautiful mountain and deserts landscapes.
[/size]
[/size]   

Very Interesting, I was thinking about requesting a button to be able to switch to death probablilites for a couple (i.e. 'both are dead' would be the death line) but didnt think about all these other changes that does occur.  Which is funny because I never liked when people said there is no reason for life insurance after retirement because you are no longer making an income...I'm not saying life insurance is reasonable at an advanced age, but its simply not true that no one depends on your income after your retire if you have a spouse, I've seen it be a big hit to a couple of people when one of those SS checks disappears.

Mgmny

  • Pencil Stache
  • ****
  • Posts: 996
  • Age: 33
  • Location: East Side of MSP
THIS IS AWESOME!!! Thank you!!!

SugarMountain

  • Pencil Stache
  • ****
  • Posts: 938

Very Interesting, I was thinking about requesting a button to be able to switch to death probablilites for a couple (i.e. 'both are dead' would be the death line) but didnt think about all these other changes that does occur.  Which is funny because I never liked when people said there is no reason for life insurance after retirement because you are no longer making an income...I'm not saying life insurance is reasonable at an advanced age, but its simply not true that no one depends on your income after your retire if you have a spouse, I've seen it be a big hit to a couple of people when one of those SS checks disappears.

I believe surviving spouse gets the higher of the two social security amounts.  Say Joe is 72 collecting $2200/mo and Jane is 70 collecting $1500/mo and Joe dies, Jane can now collect $2200 instead of $1500.

Mgmny

  • Pencil Stache
  • ****
  • Posts: 996
  • Age: 33
  • Location: East Side of MSP

Very Interesting, I was thinking about requesting a button to be able to switch to death probablilites for a couple (i.e. 'both are dead' would be the death line) but didnt think about all these other changes that does occur.  Which is funny because I never liked when people said there is no reason for life insurance after retirement because you are no longer making an income...I'm not saying life insurance is reasonable at an advanced age, but its simply not true that no one depends on your income after your retire if you have a spouse, I've seen it be a big hit to a couple of people when one of those SS checks disappears.

I believe surviving spouse gets the higher of the two social security amounts.  Say Joe is 72 collecting $2200/mo and Jane is 70 collecting $1500/mo and Joe dies, Jane can now collect $2200 instead of $1500.

This is correct: "If you are the widow or widower of a person who worked long enough under Social Security, you can receive full benefits at full retirement age for survivors or reduced benefits as early as age 60."

https://www.ssa.gov/planners/survivors/ifyou.html

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
Re-reading my unsolicited rant from last night :) I think I was trying to get around to asking for a 'Mustachian' button.  It would be really awesome to toggle between statistical average (current version) to an optimized outcome (college educated, married, cope well with stress, etc for longvity + able to reduce withdrawal 25% during bear market, AA annually re-balanced, etc.)...  Just for fun.  Maybe it would nudge people in a good direction toward financial education and being more confident to ER.

I think the easiest way to handle life expectancy is just to have the system auto-populate the number it's using now, and let the user change it to anything they want.  Go nuts with a third-party LE calculator or just guess.  It's the least work for the creator with maximum flexibility

Another (big) thank you and a request for this ^. (Those of us in developed countries outside the US will typically have higher life expectancy.)

My guess is that there would be a major challenge here because you don't just need the average remaining life expectancy, but also the shape of the distribution of outcomes, so it wouldn't be as simple as typing in a single number.

This is particularly true because increases in life expectancy are primarily driven by reduced early death rates rather shifting the whole curve.

For example, look at the change in the distribution of age at death for men in the UK between the mid-1800s and the late 20th century:




Total life expectancy came close to doubling over that time frame, but if I just took the shape of the curve from the 1800s and either shifted it over 35 years, or multiplied all the ages by 2x, I'd do an awful job of modeling the mortality curve at the end of the 20th century.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9922
  • Registered member
Re-reading my unsolicited rant from last night :) I think I was trying to get around to asking for a 'Mustachian' button.  It would be really awesome to toggle between statistical average (current version) to an optimized outcome (college educated, married, cope well with stress, etc for longvity + able to reduce withdrawal 25% during bear market, AA annually re-balanced, etc.)...  Just for fun.  Maybe it would nudge people in a good direction toward financial education and being more confident to ER.

I think the easiest way to handle life expectancy is just to have the system auto-populate the number it's using now, and let the user change it to anything they want.  Go nuts with a third-party LE calculator or just guess.  It's the least work for the creator with maximum flexibility

Another (big) thank you and a request for this ^. (Those of us in developed countries outside the US will typically have higher life expectancy.)

My guess is that there would be a major challenge here because you don't just need the average remaining life expectancy, but also the shape of the distribution of outcomes, so it wouldn't be as simple as typing in a single number.

This is particularly true because increases in life expectancy are primarily driven by reduced early death rates rather shifting the whole curve.

For example, look at the change in the distribution of age at death for men in the UK between the mid-1800s and the late 20th century:




Total life expectancy came close to doubling over that time frame, but if I just took the shape of the curve from the 1800s and either shifted it over 35 years, or multiplied all the ages by 2x, I'd do an awful job of modeling the mortality curve at the end of the 20th century.

Luckily we live now so you can just use the actual distribution and adjust it left or right based on the input.  It won’t be perfect, but it’s not like life expectancy is a perfect predictor anyways.

Hell its close enough to a normal distribution, especially looking at a conditional distribution for people who have survived childhood.  just use a reasonable stdev

cerat0n1a

  • Handlebar Stache
  • *****
  • Posts: 2319
  • Location: England
Re-reading my unsolicited rant from last night :) I think I was trying to get around to asking for a 'Mustachian' button.  It would be really awesome to toggle between statistical average (current version) to an optimized outcome (college educated, married, cope well with stress, etc for longvity + able to reduce withdrawal 25% during bear market, AA annually re-balanced, etc.)...  Just for fun.  Maybe it would nudge people in a good direction toward financial education and being more confident to ER.

I think the easiest way to handle life expectancy is just to have the system auto-populate the number it's using now, and let the user change it to anything they want.  Go nuts with a third-party LE calculator or just guess.  It's the least work for the creator with maximum flexibility

Another (big) thank you and a request for this ^. (Those of us in developed countries outside the US will typically have higher life expectancy.)

My guess is that there would be a major challenge here because you don't just need the average remaining life expectancy, but also the shape of the distribution of outcomes, so it wouldn't be as simple as typing in a single number.

This is particularly true because increases in life expectancy are primarily driven by reduced early death rates rather shifting the whole curve.

All true - but I suppose it comes down to a single set of numbers - life expectancy at each age, per gender. I haven't looked at the data so this is purely speculation, but I guess the US figures being different to other western countries largely comes down to the fact that (uniquely) some subset of the population has no/limited access to healthcare which probably does change the shape of the curve relative to other places. That's something that does not affect most FIREees and so maybe using data for the whole population is not entirely accurate here? Just modelling it as a normal distribution around the expected mean value doesn't seem too bad?

Secondary effects like the increased incidence of obesity in the US maybe have some smaller effect but probably don't change the shape of the curve much, just shift it slightly. US life expectancy seems to have fallen 3 years running, I guess it's hard to extrapolate stuff like that for someone retiring early.

maizefolk

  • Walrus Stache
  • *******
  • Posts: 7400
Life expectancy isn't a normal distribution, the right tail is extremely foreshortened and since your risk is going to be disproportionately driven by percent survival to the highest ages that means modeling it as a normal distribution is going to produce misleading results.

Now what about your other idea of trying to fit a change in life expectancy by just shifting the distribution left or right?

If you don't like the cross century comparison (easier to see the pattern I was talking about) how about an ~25 year comparison?



Modeling one of those curves as the other one shifted left or right would be statistical malpractice, and provide significantly less accurate results than just using the old distribution to model the new one or vice versa.

dragoncar

  • Walrus Stache
  • *******
  • Posts: 9922
  • Registered member
Shifting looks good to me.  Think of it like significant figures, it’s already guesswork so make the best of what you have.  I’d rather use the shifted distribution than the original distribution, which is definitely wrong