Using FireCalc after retirement

[Fatcat]

Hi all,

Apologies if this is a repeat and/or dumb question, I couldn't immediately see what I was after knowing in the thread list.

So, scenario: if I use Firecalc and come up with a 'number' for my retirement, and then live that financial life for 1 year. The stock market just happens to dip by 15%.

I come back to Firecalc 1 year later with my reduced portfolio and my number of retirement years 1 year less. I input those numbers.

Should/would I expect to generate the same result i.e. same number of successful cycles? I'm not sure if the tool can really be used this way, but it'd certainly be handy/reassuring if it could be!

Thanks

 

[Dtail]

Hi all,

Apologies if this is a repeat and/or dumb question, I couldn't immediately see what I was after knowing in the thread list.

So, scenario: if I use Firecalc and come up with a 'number' for my retirement, and then live that financial life for 1 year. The stock market just happens to dip by 15%.

I come back to Firecalc 1 year later with my reduced portfolio and my number of retirement years 1 year less. I input those numbers.

Should/would I expect to generate the same result i.e. same number of successful cycles? I'm not sure if the tool can really be used this way, but it'd certainly be handy/reassuring if it could be!

Thanks

First of all welcome to our wonderful site.
Well, Firecalc doesn't "know" you used it previously and each year one uses it , it is a fresh start.
Thus, your results will be lower than your original input, but conceptually you are now re-retiring into a lower valued market and thus the market results going forward will have more likelihood of reaching the higher squiggly lines.

 

[TwoByFour]

The test is different than your original test, but it is hard to say whether the success rate will be different from the original. It is certainly not guaranteed to stay the same.

Running the test with a 1-year shorter life expectancy than the original test, and a different start value than the original test, is not just a simple subset of the original scenario. The whole process is stochastic so things change in unpredictable ways when initial conditions change.

 

[travelover]

This is the oft discussed conundrum. Many here hedge by taking say, 4%, of their portfolio value as of a yearly date. That way you are adjusting for lean years and fat years.

 

[braumeister]

It's often useful to occasionally look through FIRECalc's own documentation:

https://www.firecalc.com/intro.php

 

[Fatcat]

First of all welcome to our wonderful site.
Well, Firecalc doesn't "know" you used it previously and each year one uses it , it is a fresh start.
Thus, your results will be lower than your original input, but conceptually you are now re-retiring into a lower valued market and thus the market results going forward will have more likelihood of reaching the higher squiggly lines.

Thanks for the welcome! Yes, so that's kind of what I expected to be the case.

I'm just trying to clarify the implications of that in my mind. Assuming that it simply doesn't take into account the market level we're at currently, and just uses the data it has independently, wouldn't that be a fatal flaw if one tries to use it 'in motion' so to speak?

Taking it to an extreme, if after year 1, our imaginary retiree was unfortunate enough to experience a 1929 monumental recession, wouldn't FIREcalc after that year 1 still be modelling with data including the scenario of what just happened last year i.e. another 75% (or whatever it was) drop? Thereby blowing what appeared to be our retirees very sound plan to smithereens (assuming he/she was trying to achieve 0 cycle failure).

Likewise, sitting where we are now, doesn't FIREcalc assume as equally likely a four-fold market gain in the next 10 years than it did when were sitting in 2008? i.e. is it being too optimistic?

I'm sure I'm just missing a basic tenet of the calculator here, sorry for the disorganised thought!

 

[Fatcat]

It's often useful to occasionally look through FIRECalc's own documentation:

https://www.firecalc.com/intro.php

Thanks, although I can't see the answer to my question there unless I'm being blind - quite possible!

 

[Fatcat]

This is the oft discussed conundrum. Many here hedge by taking say, 4%, of their portfolio value as of a yearly date. That way you are adjusting for lean years and fat years.

Sure, and I've seen FIRECalc allows you to do that in the options. But, still, whatever margin of error you build in, as I see it after year 1 (if a disastrous year) the margin that you wanted will be significantly less according to the new results of the tool once you run it again (with 1 year less and your revised-down portfolio size)

Perhaps the answer just is " don't use the tool this way"?

 

[TwoByFour]

You can run a RIP calculation every year of your retirement if you want, and it is not an invalid thing to do, whatever "invalid" means. A RIP tool is not a prediction of the future, it is a what-if scenario of the past and in that context, one "what-if" is as valid as another.

 

[ERD50]

...

Taking it to an extreme, if after year 1, our imaginary retiree was unfortunate enough to experience a 1929 monumental recession, wouldn't FIREcalc after that year 1 still be modelling with data including the scenario of what just happened last year i.e. another 75% (or whatever it was) drop? Thereby blowing what appeared to be our retirees very sound plan to smithereens (assuming he/she was trying to achieve 0 cycle failure). ....

No. And for simplicity, let's just say that historically, FIRECalc says he's 100% fine retiring in 1928 with $X, but not in 1929 after the drop.

So remember that 1928 included surviving the 1929 drop. It's already baked in. If you recalculate in 1929, you are mostly double-counting.

It's a little hard to get your head around (it was for me). But remember that FIRECalc bases the numbers on the worst scenarios. Follow the squiggly lines, and you see very many of the rise and rise and rise. FC is being conservative. So the 1928 retire is told he can only take X% to survive. Now in 1929, his adjusted withdrawal % is much higher due to the drop, but history says he survives. The % withdraw will go up as the market dips, but that was already accounted for.

So the 1929 retiree running these numbers gets an overly conservative number, because now whatever $ he has are 'worth more', but FIRECalc doesn't try to account for this (there's no 100% sure method), it's all based on worst case.

-ERD50

 

[Fatcat]

No. And for simplicity, let's just say that historically, FIRECalc says he's 100% fine retiring in 1928 with $X, but not in 1929 after the drop.

So remember that 1928 included surviving the 1929 drop. It's already baked in. If you recalculate in 1929, you are mostly double-counting.

It's a little hard to get your head around (it was for me). But remember that FIRECalc bases the numbers on the worst scenarios. Follow the squiggly lines, and you see very many of the rise and rise and rise. FC is being conservative. So the 1928 retire is told he can only take X% to survive. Now in 1929, his adjusted withdrawal % is much higher due to the drop, but history says he survives. The % withdraw will go up as the market dips, but that was already accounted for.

So the 1929 retiree running these numbers gets an overly conservative number, because now whatever $ he has are 'worth more', but FIRECalc doesn't try to account for this (there's no 100% sure method), it's all based on worst case.

-ERD50

Ok, that makes sense to me. I guess the tool has to be taken with a pinch of salt, and you also have to be accepting of the fact that you're probably not going to end up with 100% cycle success in bad times unless you're very wealthy and frugal. Likewise, in today's heady times, we should probably not dream too much of the very highest lines of the outcome and instead focus on the overall trend. Thanks.

 

[Sunset]

I just deducted 25% of our savings and when it said we would still be 100% , I was happy.

But the reality is if the market tanked more than 15% I'd start getting worried.

 

[Dtail]

Thanks for the welcome! Yes, so that's kind of what I expected to be the case.

I'm just trying to clarify the implications of that in my mind. Assuming that it simply doesn't take into account the market level we're at currently, and just uses the data it has independently, wouldn't that be a fatal flaw if one tries to use it 'in motion' so to speak?

Taking it to an extreme, if after year 1, our imaginary retiree was unfortunate enough to experience a 1929 monumental recession, wouldn't FIREcalc after that year 1 still be modelling with data including the scenario of what just happened last year i.e. another 75% (or whatever it was) drop? Thereby blowing what appeared to be our retirees very sound plan to smithereens (assuming he/she was trying to achieve 0 cycle failure).

Likewise, sitting where we are now, doesn't FIREcalc assume as equally likely a four-fold market gain in the next 10 years than it did when were sitting in 2008? i.e. is it being too optimistic?

I'm sure I'm just missing a basic tenet of the calculator here, sorry for the disorganised thought!

Bolded - Firecalc doesn't really assume a four fold increase in the next 10 years, but then again it didn't in 2008 either.
Since it uses historical information, if you are using a 30 year retirement for example, the information from 2009-2017 is built into a retirement starting in 1987.
The Bengen 4% withdrawal concept is probably best used to initially calculate if one is in a good place in order to retire.
There are very few folks who actually follow this withdrawal concept year after year including the inflation increase, as this type of spending is not typical of a real retirement spending pattern.
Thus as Travelover mentioned, more folks are gravitating to a more variable withdrawal pattern, which rewards up markets, but reins in the spending to an extent in down markets.

No. And for simplicity, let's just say that historically, FIRECalc says he's 100% fine retiring in 1928 with $X, but not in 1929 after the drop.

So remember that 1928 included surviving the 1929 drop. It's already baked in. If you recalculate in 1929, you are mostly double-counting.

It's a little hard to get your head around (it was for me). But remember that FIRECalc bases the numbers on the worst scenarios. Follow the squiggly lines, and you see very many of the rise and rise and rise. FC is being conservative. So the 1928 retire is told he can only take X% to survive. Now in 1929, his adjusted withdrawal % is much higher due to the drop, but history says he survives. The % withdraw will go up as the market dips, but that was already accounted for.

So the 1929 retiree running these numbers gets an overly conservative number, because now whatever $ he has are 'worth more', but FIRECalc doesn't try to account for this (there's no 100% sure method), it's all based on worst case.

-ERD50

+1 So effectively this is the other side of the coin concept of what you brought up in your text which I bolded above.

 

[Bir48die]

Run some different scenarios using an Excel spreadsheet with both your budget and your overall holdings. Should either give you piece of mind or a plan B

 

[Ready]

I rerun the calculations every year just to see how they would change. I think most of us take a look at FIRECalc from time to time just to see how the numbers look based on current net worth. But in theory we shouldn’t have to, since as others have said the sequence of returns risk is already baked into the results.

Realistically if the market tanks most of us will pull back spending a bit, and if the market soars, we may loosen the purse strings a bit.

Try running the scenario where you ask for a spending level that will give you a 100% success rate. That is the withdrawal rate tha should withstand even the worst cycles in stock market history.

 

[TwoByFour]

No. And for simplicity, let's just say that historically, FIRECalc says he's 100% fine retiring in 1928 with $X, but not in 1929 after the drop.

So remember that 1928 included surviving the 1929 drop. It's already baked in. If you recalculate in 1929, you are mostly double-counting.

-ERD50

The bolded part is not really right, although it might not matter.

1929 is historically the worst 30-year sequence in the last 100 years. If the OP did a FireCalc run of $1mm starting value and it passed 100%, then we know the 1929 sequence succeeded with a starting value of $1mm. If the OP now wants to test his remaining 29 of retirement after 1 year of retirement, his starting value for FireCalc is whatever his actual assets are at the end of his first year of retirement (let's call that X), so he puts that in to FireCalc and it runs all 29-year sequences with that starting value. But the 30-year sequence stating in 1928 with a starting value of $1mm (the original run) is not equivalent to a 29-year sequence starting in 1929 with a starting value of X, unless the performance for 1928 is exactly equal to the performance of his actual first year of retirement.

Now if his actual performance in the first year of retirement is better than the performance of 1928, he has nothing to worry about, he will pass at 100% still. But if the performance is worse, you don't know what will happen unless you run FireCalc with the actual asset value at the end of his 1st year of retirement. As it is, the market fell about -12% in 1928 so if the OP's actual first year is better than that, he has nothing to worry about.

But this was an easy case: 100% success on the initial, pre-retirement run of FireCalc, and retesting after just one year of retirement. If someone had less than 100% success on the initial run, and/or retested FireCalc after say, 5 years of retirement, it would be more complicated - the success rate could indeed change. In particular, if in the pre-retirement run the only sequence of years that failed was the 1929 sequence, and the actual performance of year-1 of retirement was positive, when retesting FireCalc, the number going into the terrible 1929 sequence will be much higher than it was on the initial run and it could succeed on the rerun, which would put the remainder of his retirement at 100%.

 

[REWahoo]

1929 is historically the worst 30-year sequence in the last 100 years.

You sure about that?

 

[TwoByFour]

You sure about that?

Yes and no. If you look at the S&P 500 with dividends reinvested and ignore inflationary effects, the answer is yes. If you look at those other things the answer might be different. And this is based only on lowest annualized return over the 30-year sequence. When you are withdrawing money, dividends are not being reinvested which will mean that an investor in decummulation might experience a different "worst" sequence. But for the sake of the discussion I am not sure it matters which sequence is the worst one.

 

[Dtail]

Yes and no. If you look at the S&P 500 with dividends reinvested and ignore inflationary effects, the answer is yes. If you look at those other things the answer might be different. And this is based only on lowest annualized return over the 30-year sequence. When you are withdrawing money, dividends are not being reinvested which will mean that an investor in decummulation might experience a different "worst" sequence. But for the sake of the discussion I am not sure it matters which sequence is the worst one.

I always heard that 1966 was the worst 30 year sequence, yes it is nitpicking and doesn't really matter for your concept.

 

[Dtail]

The bolded part is not really right, although it might not matter.

1929 is historically the worst 30-year sequence in the last 100 years. If the OP did a FireCalc run of $1mm starting value and it passed 100%, then we know the 1929 sequence succeeded with a starting value of $1mm. If the OP now wants to test his remaining 29 of retirement after 1 year of retirement, his starting value for FireCalc is whatever his actual assets are at the end of his first year of retirement (let's call that X), so he puts that in to FireCalc and it runs all 29-year sequences with that starting value. But the 30-year sequence stating in 1928 with a starting value of $1mm (the original run) is not equivalent to a 29-year sequence starting in 1929 with a starting value of X, unless the performance for 1928 is exactly equal to the performance of his actual first year of retirement.

Now if his actual performance in the first year of retirement is better than the performance of 1928, he has nothing to worry about, he will pass at 100% still. But if the performance is worse, you don't know what will happen unless you run FireCalc with the actual asset value at the end of his 1st year of retirement. As it is, the market fell about -12% in 1928 so if the OP's actual first year is better than that, he has nothing to worry about.

But this was an easy case: 100% success on the initial, pre-retirement run of FireCalc, and retesting after just one year of retirement. If someone had less than 100% success on the initial run, and/or retested FireCalc after say, 5 years of retirement, it would be more complicated - the success rate could indeed change. In particular, if in the pre-retirement run the only sequence of years that failed was the 1929 sequence, and the actual performance of year-1 of retirement was positive, when retesting FireCalc, the number going into the terrible 1929 sequence will be much higher than it was on the initial run and it could succeed on the rerun, which would put the remainder of his retirement at 100%.[/QUOTE]

Bolded - isn't one then effectively using Firecalc in the classic Bengen 4% scenario, but actually is really bringing into concept a "% of remaining portfolio" scenario?
Not sure of my comment, but asking.

 

[TwoByFour]

Bolded - isn't one then effectively using Firecalc in the classic Bengen 4% scenario, but actually is really bringing into concept a "% of remaining portfolio" scenario?
Not sure of my comment, but asking.

No, at least not if you are asking about the withdrawal method. The withdrawal method does not matter for what I wrote. What I wrote is just basic math really and has nothing to do with withdrawal method or amount. Or whether the worst year is 1929 or 1966.

 

[Dtail]

No, at least not if you are asking about the withdrawal method. The withdrawal method does not matter for what I wrote. What I wrote is just basic math really and has nothing to do with withdrawal method or amount. Or whether the worst year is 1929 or 1966.

Ok.

 

[braumeister]

What I wrote is just basic math really

Shouldn't us liberal arts majors have some kind of retirement planning system too?

 

[REWahoo]

Shouldn't us liberal arts majors have some kind of retirement planning system too?

Some of us do - even though for us, numbers is hard.

 

[audreyh1]

1929 is historically the worst 30-year sequence in the last 100 years.

No, it's not. 1966 is worse. As are several pre-1920 starting sequences.

Inflation matters.