Firecalc not the whole picture?

[palomalou]

Okay, I run Firecalc several times a week, changing various parameters. Also i-ORP. SInce Firecalc should not be a deciding factor, what else should I be doing? I've run the Merrill edge and the TRowePrice, but they each have problems of omission, such as not allowing for part-time work after retirement or for the sale of our house, or trying to figure out what we need by what we earn now (irrelevant, as we now save roughly half of what we earn). Spouse has a spreadsheet as well but is constantly tweaking it to try to be sure it is reliable.

[REWahoo]

While no one should bet the farm on the results provided by any retirement calculator, I haven't found anything any better than FIRECalc when it comes to the logic behind it.

[Midpack]

You can run them all and create as many spreadsheets as you want, but there is no 'deciding factor.' There are so many variables and uncertainties, the best you can do is test your plan against past history (FIRECALC does that well IMO), decide how you think the future will compare with the past and build in any additional safety factor that allows you to sleep at night (if needed), have a plan B, monitor your progress and remain flexible. If you pull the plug with rigid expectations on your spending, taxes, investment returns, inflation, Soc Sec and other income sources, life span(s) and all the other variables - I guarantee many will be wrong (could be worse, or better!).

The FIRECALC graphs illustrate the uncertainty well. A given plan could fail before 30 years, or end with a bequest many millions more than the retiree started with - those and anything in between were real possibilities that actually would have happened from 1873 thru ending in 2011. And those are results showing the variability of real returns only! There are many other variables.

If you're waiting for a calculator/spreadsheet with no 'ommisions' - it will never come, guaranteed.

That said, (like me and others) you may need to try several calculators/spreadsheets to fully appreciate there is no silver bullet. You're (understandably) going through convincing yourself now, hopefully you'll get there before you drive yourself batty...

[braumeister]

You can get some interesting insights from running Firecalc with various time frames.

Don't use only your most likely projected future, but try various shorter or longer spans of years to vary your perspective.

[Animorph]

Sounds pretty similar to what I did. The calculators should give you a level of confidence. The spreadsheet should give you some guidance for your particular situation and as an aid to track your retirement financial health. But what really helped me was seeing that we had enough flexibility in our spending to cope with unexpected events. It's the future, there is no certainty, even if you worked an extra 10 years.

[kyounge1956]

Have you also used a Monte Carlo simulator? I'm not familiar with i-ORP and the others you mentioned. My favorite Monte Carlo is Flexible Retirement Planner, and I don't think it has any of the flaws you mentioned--it's possible to input part-time income, a lump sum such as sale of a house, and it doesn't force you to use a fraction of your current income for your retirement income needs.

[REattempt]

I think the logic and models behind Jim Otar's excel retirement optimizer are as good or better. Same basic principles, some different assumptions but Is my choice. I'd recommend his ebook (PDF) for 5-6$. Explains everything and is a great book on distribution portfolios.

http://www.retirementoptimizer.com/

YMMV

[REattempt]

Btw, Jim has a great section that explains the problems with Monte Carlo simulations. Best I've read on the subject. As a result, I won't use them.

[67walkon]

FireCalc works pretty well. You have to play with it a little if you have unusual circumstances, but it at least gives you some comfort.

[Rustward]

If you are just looking for another calculator, you can try Fidelity's Retirement Income Planner.

I really don't see any of these tools warn users about retiring after a market runup into a steeply declining market. I don't expect them to predict the future, just mention the possibilities. For example, if you concluded from any of these calculators in 2007 that you were good to go -- but just barely -- and you pulled the plug in 2008, where would you have been in early 2009?

[Free To Canoe]

The "whole picture"?
Money is just a part of the retirement picture.
Possible questions:
Will my money last longer than me?
Am I spending / saving too much?
Am I spending / saving too little?

Or is it...is Firecalc the best retirement income planner?

These questions bring more questions (of course).

[rodi]

I use them all. Fidelity's RIP, i-Orp, Firecalc, the one provided by my work (financial engines), another one provided through my former megacorp's pension website...
And of course a home-brewed spreadsheet.

As mentioned - firecalc is great for backtesting against previous market cycles and historical data.
I-orp is great for helping you what-if the tax strategy of what money to use when...
So they offer very different information - all of it useful.

If you don't have a magic date of full cola'd pension/benefits - you just have to sock the $$ away and keep adjusting estimates and running the numbers till you reach that "sleep at night" point.

[photoguy]

I think FIRECalc is a great tool for retirement planning. It's one step more detailed than a back of the envelope calculation and realistically I think that's the best one can do. The logic behind it is sound and although it might not account for every little tweak in your withdrawal plan it's only a matter of time before someone includes them as options.

The problem I have with FIRECalc is that I think people tend to place too much reliance on it's predictions. For example, if FIRECalc says a given WR rate has a 5% chance of failure that means only that in historical data their were 5 failures out of 100 or so cycles. That doesn't mean the probability of failure going forward is 5%.

It's well known that with data driven (statistical) models that performance on live data is ALWAYS worse than when the model is run on historical data. Sometimes the performance drop is small and sometimes it can be very large. Thus I think the way to interpret FIRECalc results is that it is a sanity check to make sure you are at least in the right ballpark. If the withdrawal method failed miserably historically it's probably not a safe bet going forward.

Monte Carlo methods can offer more flexibility in what you simulate and a greater number of "cycles" compared to FIRECalc but ultimately they are only as good as the assumptions that you put into it. For retirement planning and MC simulation you would need to

(1) Make an assumption about the distribution of returns in a given year, and
(2) Make an assumption about how a sequence of consecutive years might be distributed.

For example, for #2 one might develop a model that has some sort of mean reversion built into it (or not). The problem is that the study of investing returns is not a science like chemistry or physics where one has a realistic/mechanistic mathematical model. What's going to happen in practice is that the statistician or modeler is going to look at historical data (the same data FIRECalc uses) to set probability distributions governing assumption #1 and #2. So even though you could run millions of "cycles" in an MC simulator I don't believe the results are going to be any more useful than FIRECalc (and could be substantially worse if your assumptions are bad).

[Midpack]

Quote:

Originally Posted by photoguy It's well known that with data driven (statistical) models that performance on live data is ALWAYS worse than when the model is run on historical data.

Enjoyed your post overall, but care to explain "ALWAYS worse" a little further?

[rescueme]

Quote:

Originally Posted by Rustward If you are just looking for another calculator, you can try Fidelity's Retirement Income Planner. I really don't see any of these tools warn users about retiring after a market runup into a steeply declining market. I don't expect them to predict the future, just mention the possibilities. For example, if you concluded from any of these calculators in 2007 that you were good to go -- but just barely -- and you pulled the plug in 2008, where would you have been in early 2009?

Just an FYI on RIP.

It will forecast your possible results a/o Jan 1 of the following year forward, and it will also "discount" your current portfolio value to just under 10% in order to forecast that "aw sh**" situation you mentioned.

I retired in early 2007, and had been using RIP for about five years before that. While I use other forecast software, I still perfer it due to the year by year breakdown it does, along with budget changes as you age.

2008 was the actual "blip" for me, however RIP showed it would not severly impact my total plan, even though it was yet unknown on the day of my retirement. Today? Some five years after retirement my "actual" is better than planned and 2008 was just a bump in the road.

[photoguy]

Quote:

Originally Posted by Midpack Enjoyed your post overall, but care to explain "ALWAYS worse" a little further?

Not sure if I explained this well, but here goes:

When you develop a statistical model to predict some quantity you will typically test it on historical data to get a measure of how accurate it is. For example, a weather prediction system might be 80% accurate in tests on historical data (i.e., backtesting).

However when it comes to applying such a system on live data (i.e. to the future), the model will always be less accurate than what we saw. For example, it might only be correct 70% of the time.

This discrepancy comes about because the historical data which we used to build the model is finite. There's a much wider range of weather patterns that could occur beyond what we've seen in the 100 or so years that we've collected data from weather stations / satellites / etc.

Another factor that might throw the model off in the future is the fact that the system is changing (non-stationary). I.e all of the green house gases we are pumping into the atmosphere is changing the weather system but much of the data that went into the model is from the first part of the 20th century where there were relatively lower emissions.

Now FIRECalc is very careful to say that it is not a prediction system and only tabulates the historical success rate but many people use that probability as an estimate of future success rates. There is nothing wrong with this, but both of these issues (finite data, changing system) should be considered when one evaluates how accurate the number coming out of FIREcalc may be towards our future retirement outcomes.

[Nords]

Quote:

Originally Posted by palomalou Okay, I run Firecalc several times a week, changing various parameters. Also i-ORP. SInce Firecalc should not be a deciding factor, what else should I be doing?

So, FIRECalc is a gateway drug?

In addition to Otar's calculator you could try Kotlikoff's ESPlanner (which Wade Pfau just reviewed) and a subscription to Financial Engines.

If none of them object to your plans, then you can ER and spend all day running them.

[Rustward]

Quote:

Originally Posted by rescueme It will forecast your possible results a/o Jan 1 of the following year forward, and it will also "discount" your current portfolio value to just under 10% in order to forecast that "aw sh**" situation you mentioned.

I didn't realize that. Thanks.
Fidelity does provide a rather lengthy "Methodology" document about RIP that I have not fully read.

[traineeinvestor]

I used FIRECalc, a few others and my own spreadsheet. I also keep a print out of Sam Savage's article on "The Flaw of Averages" in my retirement planning folder (I also have his book) as a rather blunt reminder that the starting point/first few years are critical in determining whether or not the plan will go the distance.

The other things that give me comfort:

1. I have over engineered our budget - I added an arbitrary 20% to expected retirement expenses and I made no allowance either for the fact that our children will one day become financially independent or that DW wants to continue her part time job (at least in the short term )

2. the net rents and dividends will meet out expenses - meaning we should not have to draw on principal to pay the bills.

Given that we have no pensions, SS etc to fall back on, we've taken a very conservative approach (which is the only reason I am still working).

[pb4uski]

I pulled the trigger on RE based on an exhausting (to me, not necessarily exhaustive) multitude of analyses:

1. Quicken Lifetime Planner (and Budget) - which I consider to be great for a basic plan and what things one needs to think about but has a significant flaw of only being deterministic (you have to provide it a portfolio return rate). To accommodate that flaw I sent the portfolio return 200 bps lower than the historical average return associated with my AA to be conservative.

2. Vanguard financial planning - I had gone through the Vanguard financial planning process a few times over the last 7 or so years and went through it one more time.

3. Financial Engines - since this comes free to Vanguard customers I have run it a few times a year just to see that we were on track.

4. FIRECalc - one of my favorites as it shows a wide range of results.

5. SmartMoney Retirement Planner - I didn't spend a lot of time with this but it seemed to be a reasonably credible free deterministic planner if you don't have access to Quicken's Lifetime Planner

6. Guided Choice Quick Advice tool http://guidedchoicenow.com/quickadvice/ a last minute addition that sealed the deal

7. ORP

Each of these were within my range of comfort so I decided that rather continuing to work (which I still enjoyed to some extent) to make our heirs richer that it was time for us to retire, kick back, and enjoy life more.