FireCalc: Year 1, 2, etc

[kat]

Hi everybody!

I have a really basic question, but the answer seems to make a difference in my plans to sell my house.

For someone retired on January 1, 2006 and not taking any withdrawals until December 31, 2006, as recommended, which is year 1, '06 or '07?

If I plan on selling it this summer, would that mean a $xxx,xxx addition in year 1? So if I wait until '07, that would be year 2 instead?

I keep getting confused because I'm considering the December '06 withdrawal for year 1, meaning '07.

Did that make any sense?

Thanks!

kate

 

[kat]

Wow, 72 views in two hours!

 

[Nords]

Wow, 72 views in two hours!

Yeah, I get what you're implying, "and not a single @#$% answer yet!"

We're all waiting for Dory to step in with the correct answer.

 

[cute fuzzy bunny]

Fortunately we dont charge by the view.

Actually i'm not sure i understand the question.

I found some ambiguity in firecalc for which I never asked, if I used year zero or year one I always get the same numbers, so if your question is "is the first year year one or year zero", then answer from me is "I'm not sure". Dory might be the only one that can answer the question, whatever exactly it is...

 

[REWahoo]

We're all waiting for Dory to step in with the correct answer.

Or at least one that sounds more authentic than I could come up with...

 

[cute fuzzy bunny]

I believe the first year (in this case 06) counts as both year zero and year one. 07 counts as year 2.

Prrrrrrobably doesnt matter whichever way you do it. Its really a long range planning tool. Differences of six months and a year may materially change the results at the end of 30 years (albeit not much) but then again, its a projection based on historic data, not a predictor.

There...all you have to do is impugn our artistic integrity to get a lot of speculative information and opinionated statements with little real background information.

 

[MasterBlaster]

Your question is :

How do I account for investments and asset sales that don't occur on exactly the year boundaries ?

It will not be possible with the Firecalc software to run partial-year monte-Carlo simulations on all of the possible outcomes. However, the simple answer is to reduce or increase Assets that are sold/bought with an appropriate discount rate so they are start of year equivalent. This should get you pretty close to the result that you want.

So for example, if you sell your house in June for $300k (net) to your portfolio. You could increase it's value by 6 months of interest and then use Firecalc as if you sold it the next year.

 

[cute fuzzy bunny]

It will not be possible with the Firecalc software to run partial-year monte-Carlo simulations

Just to be clear, firecalc doesnt do monte carlo simulations. It uses historic data tables in the same order they naturally occurred.

Your point is correct though, it cant do part years.

 

[kat]

I almost almost almost sent Dory a pm, but there's such wisdom here too, so I thought I'd......

 

[cute fuzzy bunny]

Never confuse wisdom with bullshit

 

[dory36]

For someone retired on January 1, 2006 and not taking any withdrawals until December 31, 2006, as recommended, which is year 1, '06 or '07?

If I plan on selling it this summer, would that mean a $xxx,xxx addition in year 1?   So if I wait until '07, that would be year 2 instead?

Year 1 is 2006 in your example, and yes, the house sale proceeds if sold in 2006 would go in year 1. But as it's mid-year, you could add a bit of safety by entering it as year 2. The difference is that firecalc assumes a full year's earnings on the sale.

For the year 1 vs year 2 issues, sometimes seeing the results is more meaningful than a narrative.

Try setting up a $1/year annual withdrawal from a $100,000 portfolio, invested in 0% stocks and the rest (100%) in commercial paper. Then insert your anticipated change as a very large amount, like 250,000.

Then in firecalc run the "detailed results" option and look to see when the impact shows up.

 

[kat]

Thanks Dory! You explained a lot of things!


kate

 

[Texas Proud]

If you want to use monte carlo go to financialengine.com (or something like that...) it uses a monte carlo calc and is better than FIRECALC IMO

 

[cute fuzzy bunny]

The problem with monte carlo simulations is that they can produce more optimistic and more pessimistic results than reality might. Some are better than others at smoothing out the random stringing of years and the loss of market correlations.

The problem with firecalc is that for long runs (say 25, 35, 40 years), all of the runs since 1970-1980 are truncated, so they're very likely to be "successful". Leaving results correlated to the modern economy and financial worlds on the table doesnt thrill me. I almost wish there was an option to "fill" the missing years with something, but "something" would probably make for a 35 page post "discussion".

However, its likely that neither approach may be needed. Most portfolio's fail in simulation on the Great Depression or the sideways/stagflation period of the 60's/70's. So if you stopped at being 90% or 95% successful, chances arent that you've got great odds of making it...chances are you will if we dont see a scenario like those two ever again. If you think situations like those two wont happen during your lifetime, then as long as you have 25x or more of your annual withdrawal, invested 40/60 to 70/40, you probably wont see a failure. If you try firecalc and succeed in going through those two periods (which will give you 100%), chances are we wont see anything much worse than those two.

If you're a real glutton for punishment, take the worst terminal value firecalc produces from the detail report and rerun it with that as your portfolio amount. Basically this would test to see if you made it through two sequential worst case trips through the depression and stagflation.

Bet you can sleep well if you can make it through that test.

 

[dory36]

From the FAQ in Firecalc:

How does FIRECalc compare to "Monte Carlo" simulations I have read about?

Both FIRECalc and Monte Carlo simulations are trying to get away from using a single historical or assumed average rate of return and inflation rate. Neither can predict the future, so both have to come up with some means of putting together numerous potential future scenarios. In effect, both programs are rolling dice to come up with scenarios that presumably could happen in the future. Such a simulation needs many sample situations -- the more the better, as long as they are realistic -- to give meaningful answers.

One approach to this, which is what a Monte Carlo simulation does, is to define a lot of rules about how much inflation might happen, how the markets might perform, and so forth. Expert analysis is used to insure that the range of such values is realistic. Then they roll the dice many times to create many possible future sets of market conditions, inflation, and so forth, all applied to your portfolio and withdrawal plans. They are using their theory about what is possible, and then rolling the dice to see how different combinations of years might impact you. Presumably they are limiting the possible market returns in a single year to some range that they believe is within reason.They would also limit inflation to some range, and probably tie interest rates to inflation somehow. Using these hypothetical situations allows them to "roll the dice" many hundreds of times for each portfolio. To the extent that their rules about possible future market conditions are reasonable, their approach is certainly sound.

In FIRECalc, and in the REHP spreadsheets on which FIRECalc was based, we use no theory, and certainly don't claim to know enough to say what might happen. Since we're looking at safety and not optimizing portfolio balance or income, the task seems simple. Would a given strategy survive the worst that has ever happened in the US, or not?

With FIRECalc, you're accepting that the next 30 years or so won't be any worse than the worst we've seen in our actual history. With a Monte Carlo simulation, you're accepting their definitions about what is possible in the future.

 

[mb]

The problem with firecalc is that for long runs (say 25, 35, 40 years), all of the runs since 1970-1980 are truncated, so they're very likely to be "successful".

One possibility is to loop around and continue with the data from year 1, 1850 or what ever the starting year is.

You do something similar in some simulations in the physical sciences. For example in chemical physics if an atom wanders out of one side of the control volume you just bring it in on the other side.

Of course for FIRECALC you would lose the "previous history effects" at the discontinuity but that may be less important than including recent market history.

MB

 

[cute fuzzy bunny]

The problem with that is that there is no correlative 'hook' between ending in 2005 and picking back up in 1871. The starting point for the data set can be a point of contention...the 1871 point was when the US economy just started recovering from the civil war. The next decade or two were a time of strong economic recovery and good times for investors. Some say that starting with such an optimistic period that is so decoupled from modern economic times doesnt really help the modern investor determine survivability. I think theres some truth to that.

And you're right...given the huge bull market runup in the 90's, its likely that any period including that 6 year span would be successful.

It all boils back down to "did I survive the depression and 60's/70's stagflation/sideways" periods. If you did, you're probably in great shape. If you didnt, but survived the other periods, then you're fine as long as no sustained badness happens...if you're getting firecalc failures all over the place, then you need more money or to spend less. You're not gonna make it.