FIRECalc updated for Q4 2008

[dory36]

FIRECalc now includes the 40+% decline in the equity and interest market from 12/07 through 12/08.

I am using the 12/08 data instead of the as-yet unavailable 1/09 data, but I didn't want to wait another month to get the impact of the 4th quarter 2008 into FIRECalc. So look for a minor correction in another month or so.

dory36

 

[dory36]

By the way... the only updated data I have is the "standard" market and interest data. So the "mixed portfolio" will not include the new data. That portfolio data was generously contributed by Bob Clyatt, the author of Work Less Live More, but that was a one time contribution, and I have no practical and accurate way to extend it to later years.

 

[samclem]

Dory,
Thanks, as always.
Maybe in your next program mod users could get an option to exclude this data set from the simulations? And any other data with a negative return> I'd like to keep my planned withdrawal rate the same without changes caused by disturbing new data . . .

 

[ESRBob]

Dory,
I need to update the figs in any case for future data runs and editions of the book. Let me see what I can do here in the next few weeks and see if I can get the updated data for you to include. Is it just 2008 or is it also 2007 we need?

 

[Rich_by_the_Bay]

First impression from re-running my numbers is that the total impact of including these data is minimal. Then I reminded myself that this represents only 3% or so for a 30-35 year retirement run.

Thanks for updating.

 

[youbet]

First impression from re-running my numbers is that the total impact of including these data is minimal. Then I reminded myself that this represents only 3% or so for a 30-35 year retirement run.

Thanks for updating.

I'm never quite sure I understand FireCalc runs completely Rich, but I think the impact is even less than you think. We now have the 1979 - 2008 30 yr run. And 2008 is the last year of that run. So there are now 109 30 yr runs available as part of the "test" and 2008 is the last year of the last run.

At least that's how I think it works. Perhaps Dory will comment.

 

[FIRE'd@51]

We now have the 1979 - 2008 30 yr run. And 2008 is the last year of that run. So there are now 109 30 yr runs available as part of the "test" and 2008 is the last year of the last run.

Additionally, 29 years of the 1979-2008 data set overlap with the 1978-2007 data set.

 

[Rich_by_the_Bay]

I'm never quite sure I understand FireCalc runs completely Rich, but I think the impact is even less than you think. We now have the 1979 - 2008 30 yr run. And 2008 is the last year of that run. So there are now 109 30 yr runs available as part of the "test" and 2008 is the last year of the last run.

Yes, on further thought I think you're right. Either way, bad as 2008 was, it will take more than a few years to topple the assumptions previously in place. Reassuring, in a backward sort of way.

 

[ERD50]

I'm never quite sure I understand FireCalc runs completely Rich, but I think the impact is even less than you think. We now have the 1979 - 2008 30 yr run. And 2008 is the last year of that run. So there are now 109 30 yr runs available as part of the "test" and 2008 is the last year of the last run.

At least that's how I think it works. Perhaps Dory will comment.

Further to that, most people are running FireCalc looking at the at 95% success number. If you were at ~95% with your numbers with 108 runs, 5 failures gives you 95.37% success.

If 1979-2008 added a failure with your profile, that would be 6/109 = 94.50% success.

If 1979-2008 added a success with your profile, that would be 5/109 = 95.41% success.

Whether 2008 really impacted that 1979-2008 run depends upon whether the bad year of 2008 was enough to flip it from success to a failure. If 1979-2008 already was going to be failure for you, even if 2008 was neutral, then it wasn't 2008 that "caused" the failure, it would just be an extension of a to-be-failed run. But a bad 2008 would make it worse. That 95% success rate does not tell you how badly you failed in the bad runs (only the worst one), just that you failed.

Which is another reason I like to use a 100% success rate, and/or analyze the end value - I think you get a bit more 'sensitivity' out of this excellent tool by doing that. A 95% success rate sort of truncates and hides some of your data.

Again, all the above is based on the assumption that *my* assumptions on FireCalc are correct, I'm pretty sure about that, better than 95% sure I'd say

-ERD50

 

[Martha]

Being a believer in black swans, I used 100% when I did FIREcalc way back when as an extra precaution.

Nevertheless, black swans are what you don't expect. Divorce. Other serious family event. Hurricane Katrina. Things that change what assets you have but are unrelated to the market. Or things that change what expenses you carefully budgeted for.

Just a reminder that a 100% success rate only relates to the market.

Thanks Dory. Stay warm.

 

[youbet]

Additionally, 29 years of the 1979-2008 data set overlap with the 1978-2007 data set.

Exactly. We've had one extra test run added. That new test run is simply the old 1978-2007 test run with 1978 deleted and 2008 added. In my case, the new 1979-2008 test run is a "pass" so no change for me.

 

[youbet]

Further to that, most people are running FireCalc looking at the at 95% success number. If you were at ~95% with your numbers with 108 runs, 5 failures gives you 95.37% success.

If 1979-2008 added a failure with your profile, that would be 6/109 = 94.50% success.

If 1979-2008 added a success with your profile, that would be 5/109 = 95.41% success.

Whether 2008 really impacted that 1979-2008 run depends upon whether the bad year of 2008 was enough to flip it from success to a failure. If 1979-2008 already was going to be failure for you, even if 2008 was neutral, then it wasn't 2008 that "caused" the failure, it would just be an extension of a to-be-failed run. But a bad 2008 would make it worse. That 95% success rate does not tell you how badly you failed in the bad runs (only the worst one), just that you failed.

Nice analysis and examples. I agree, at this point in time 2008 data only impacts one run out of 109.

Which is another reason I like to use a 100% success rate, and/or analyze the end value - I think you get a bit more 'sensitivity' out of this excellent tool by doing that. A 95% success rate sort of truncates and hides some of your data.

Agreed. I also find it very useful to take the ending values and do a histogram, observe that for normalcy and calculate x-bar and sigma. Sometimes I also do that with some of the intermediate points (say, every five years) just to get a feeling for what the variation in intermediate and end values could be even though ultimately the portfolio survives.

We've discussed this before and I think we agree that even though a portfolio is destined to "survive," it may experience some significant volatility along the way that will test the mettle of even the most experienced, stoic investor! Just witness all of our wailing and hand wringing regarding the current economic cycle, the condition of our portfolios, etc., being expressed in many current threads.

 

[youbet]

Being a believer in black swans, I used 100% when I did FIREcalc way back when as an extra precaution.

Nevertheless, black swans are what you don't expect. Divorce. Other serious family event. Hurricane Katrina. Things that change what assets you have but are unrelated to the market. Or things that change what expenses you carefully budgeted for.

Just a reminder that a 100% success rate only relates to the market.

Thanks Dory. Stay warm.

There are lots of ways to make the test "conservative." Using a 100% success rate is one. Padding expenses, understating other sources of income, increasing the length of the withdrawal period and not including "stashes of 'reserve' cash" in your submitted portfolio value are others.

I think the most important thing for folks to do with this tool is to stop and think about what FireCalc is actually doing and what the outcomes actually mean. Taking advantage of the Excel data output and just looking it over is very useful. It's even better to do some quick and simple Excel statistical analysis.

100% success rates do not guarantee lack of anguish during the withdrawal period unless you personally can handle potential large drops in portfolio value without concern........

 

[W2R]

100% success rates do not guarantee lack of anguish during the withdrawal period unless you personally can handle potential large drops in portfolio value without concern........

Well, but if you know your "anguish threshold" you can specify that the portfolio has to contain xxx dollars at all times. I usually double what I think xxx should be, just on principle.

But events like Hurricane Katrina can make you [-]feel [/-]so powerless and really things are taken out of your hands. Worse than any disaster movie I have ever seen.

 

[youbet]

Well, but if you know your "anguish threshold" you can specify that the portfolio has to contain xxx dollars at all times. I usually double what I think xxx should be, just on principle.

Good point. And that's another way to make your Firecalc testing more "conservative." Just use the minimum value feature. Bear in mind, however, this method may not completely eliminate "anguish."

Example. You enter a $1,000,000 portfolio and do a FireCalc run which survives at 95%. You then modify this so that anything below $500,000 is a "failure." To continue to have a 95% success rate, you'll now need a much larger beginning portfolio given a constant withdrawal rate ....... let's guess..... say $1,500,000. Now, the same dip that would have brought you to zero in the first run will bring you to the vicinity of $500,000 now. Still about a $1,000,000 loss, although you have the security of still having $500,000 left. But you had to come up with an extra $500,000 up front! Thee's no free lunch, sadly......

Try looking at the FireCalc output graph for a test run with a 100% success rate. Lines moving from left to right near the center of the distribution show relatively stable portfolio performance. Perhaps your portfolio only varies in real value +/- 20% or so over the 30 yrs despite your ongoing withdrawals. That would be nice and is completely possible. Even better, observe that some lines trend up continously. You're spending money yet your portfolio keeps growing. Sweet! But, alas, some lines trend down and lead to low portfolio values midstream and near zero by the end. Those outcomes are all 100% successful, but if that's what I wind up with (and I'm sure off to that kind of start! ) I'll be gritting my teeth the whole time.

Ya just gotta learn to live with it.

 

[samclem]

We've discussed this before and I think we agree that even though a portfolio is destined to "survive," it may experience some significant volatility along the way that will test the mettle of even the most experienced, stoic investor! Just witness all of our wailing and hand wringing regarding the current economic cycle, the condition of our portfolios, etc., being expressed in many current threads.

Exactly. An allocation and withdrawal rate with a 100% historical track record of success might be little comfort as the investments all lok like they are going into the toilet at the same time. I try to imagine how I'd feel and react as I ride one of those lines down into a deep trough without certainty of what lies ahead. The "X% of year-end value" withdrawal method has a lot going for it when this volatility is considered.

 

[ERD50]

Well, but if you know your "anguish threshold" you can specify that the portfolio has to contain xxx dollars at all times. I usually double what I think xxx should be, just on principle.

Good reminder. I had not played with that in a while.

A quick couple runs shows that with the others at defaults, if you say that you don't want to see your portfolio drop by more than half, your SWR would need to be down around 2.1% to get a 94.5% success rate.

Reversing the Eq/Fixed ration from 25/25 to 25/75 improved it just a bit, a 2.375% SWR gets you 94.5% success.

SS, pension, and longer/shorter period will change the numbers, so everyone needs to run those for their own situation and draw-down tolerance. But that outlines the general trend.

-ERD50

 

[youbet]

I try to imagine how I'd feel and react as I ride one of those lines down into a deep trough without certainty of what lies ahead. The "X% of year-end value" withdrawal method has a lot going for it when this volatility is considered.

In addition to the "X% of year-end value" withdrawal method, I'm looking at test runs with more conservative (lower beta) allocations. I understand my withdrawal rate will need to be lower (I'm already retired so can't add to the portfolio amount) but the reduced volatility may add more to my quality of life than a few grand extra in the budget!