FIREcalc rewrite - suggestions?

[dory36]

I'm probably going to rewrite some portions of FIREcalc (http://www.early-retirement.org/fire) over the next couple of months or so.

I've had a handful of suggestions, mostly related to clarifying the instructions and explanations, and will be incorporating those.

What would you like to see added or changed?

Dory36

[Cb]

Some ideas (some easy, some not):

1) I think the entry fields for the out-year revenue streams are counter-intuitive, but otherwise the calculator is terrific!

2) Even cooler would be a calculator that asked the following:

2a) How old are you & your SO currently?

2b) Whaddya need to live on in today's dollars?

2c) Any out-year revenue streams? If so, how much & when?

2d) How much have you got now? How much can you add per year?

2e) Review & adjust the following defaults: 77/23, 100% SWR, etc, etc.

With that info the calculator could tell the user the following:

Based on actuarial tables (or user-defined drop-dead ages), a 77/23 allocation, etc, etc you:

-will require a nest-egg of $x.xM in today's $$ to allow a 100% SWR of $xx,xxx annually until you meet your maker in 20xx

-you're a bit short, Bunky. At your present rate of savings you'll be able to punch out in another xxx months, with an inflation adjusted nest-egg of $x.xM, to enable a SWR of $xx,xxx (adjusted for inflation) etc

Knowwhatimean?

Cb

[dory36]

Wow -- thanks CB!

I had been trying to think about how to make the whole thing more understandable, but hadn't been making progress, as I had been focusing more on the individual items rather than the whole approach, as your suggestions do.

Your thoughts help tremendously! You'll probably recognize your thoughts in the final result.

Dory36

[patnbj]

First of all thanks for creating the FIRECalc tool and for enhancing it!

Enhancement suggestion: In addition to the existing inflation adjustments, add the ability to specify a specific % (this could be a range from say 0 to 20%). The reason why is because I find that most of my expenses grow at a rate less than CPI/PPI. Health insurance, however is growing a lot faster than CPI/PPI. So it would be great to run through the two groups of expenses seperately in order to potentially fund them differently.

1. Please include the best and worst result here, in addition to the average:

The average (mean) portfolio balance following the withdrawal in year xx was $.

2. Allow the calculator to come up with the optimal asset allocation for the assets mentioned, instead of letting people experiment with the amounts.

3. Allow people to input asset allocations for each of the following:
Stocks
TIPS
Commercial Paper
5 Year Treasury
30 Year Treasury


Thanks!!!

[dory36]

Good suggestions! #1 and #2 can be done readily with available data, and I'll add them to the list for the next version.

Since the calculator is based on historical results rather than speculation about what might happen, #3 requires data on how any alternative investments performed since 1871 (or would have performed, if it's possible to calculate from existing data, as Intercst did for TIPS).

Dory36

[Ted]

I think that FIRECalc is the best calculator that I have seen in terms of its inherent provision for accounting for historical inflation, and its provisions for inputting changes in assets and expenditures. However, I have cautioned in other posts that past market performance does not guarantee future results. In this regard it is especially worth noting that the economic growth that the U.S. economy experienced from 1871 to recently, which was responsible for the generally high real rates of return on financial assets, was virtually unequalled anywhere else in the world at any time in history.

What I'm getting at is that attempting to "fine tune" FIRECalc too much may be counterproductive in that it will create the mistaken impression that it is better than it can possibly be at predicting future returns on financial assets.

Having said that, it would be nice if a couple of other asset categories could be included, such as high yield bonds and REIT's. The problem is, of course, that these asset classes have only been in existence for a couple of decades, and I can't think of a good way to create "synthetic" past returns for them the way that it is possible to do for TIPs.

Anyway, in modifying FIRECalc, please remember the dictum that doctors are supposed to follow: "First, do no harm."

[dory36]

Ted is certainly right that "past market performance does not guarantee future results" and about some of the remarkable results seen in the past.

That's why it's important to keep in mind that FIRECalc is a safety-oriented calculator, not one that purports to predict future returns.

It tries to predict what the market will not do, based on what it has never done. Sort of like predicting that a specific runner will not take over 15 minutes to run a mile, based on the fact that the same runner has never taken more than 12 minutes to run a mile, in 130 past races.

That's a whole lot easier to predict than how fast he will actually run the mile!

FIRECalc makes just one assumption: The next 30 years (or so) will not be any worse than the worst performance we've ever seen in this country for a term of the same duration, no matter when it started. *

And FIRECalc asks just one question: Would my withdrawal strategy and asset allocation have survived the worst we've ever seen?

Good years do nothing to improve success in FIRECalc, since a positive portfolio of $1 is treated exactly the same as a portfolio in the billions.

This brings into question the asset allocation exploration idea. *Now we're into looking for the best strategy rather than one that would have survived all the financial disasters of the past 130 years.

I did some preliminary playing around, but the only algorithm I can think of off the top of my head is to optimize for ending portfolio balance. But this would heavily weight those years when inflation was high, and discount those years when inflation was low. That doesn't produce useful information for someone trying to allocate their investments for tomorrow.

Logical: Something that would have worked in each of the past 130 xx year terms has a high probability of working in the 131st xx year term.

Illogical: The strategy that would have produced the best average results in 130 *xx year terms will produce the best results in the 131st xx year term.

So, I'm now thinking it's a bad idea to add some search tool for best asset allocation...

Thoughts?

Dory36

[intercst]

dory36 asks,

This brings into question the asset allocation exploration idea. Now we're into looking for the best strategy rather than one that would have survived all the financial disasters of the past 130 years.

I did some preliminary playing around, but the only algorithm I can think of off the top of my head is to optimize for ending portfolio balance. But this would heavily weight those years when inflation was high, and discount those years when inflation was low. That doesn't produce useful information for someone trying to allocate their investments for tomorrow.

Logical: Something that would have worked in each of the past 130 xx year terms has a high probability of working in the 131st xx year term.

Illogical: The strategy that would have produced the best average results in 130 xx year terms will produce the best results in the 131st xx year term.

So, I'm now thinking it's a bad idea to add some search tool for best asset allocation...

Thoughts?

------------------------------

There's a whole group of disaffected former TMF posters engaged in just such an exercise over on the www.nofeeboards.com

I'm sure we'll hear about it if anything of value emerges.

intercst

[Ted]

In terms of maximizing the probable return, the "best" asset allocation is 100% high risk stocks!

But the "catch," of course, is that the higher the probable return, the higher the "risk" of loss -- which for retirees could be devastating. Thus, the challenge for most retirees is to select a portfolio mix in which the "reasonable worst case" scenario would allow their last dollar to be spent on their funeral.

The past results of different hypothetical asset mixes that FIRECalc analyzes presumably provide a reasonable means of estimating the probability that a particular asset mix will finance a particular level of real (inflation adjusted) expenditure for a particular number of years in the future. But the returns provided by FIRECalc, in my opinion, are not inherently "conservative," in that they are based on a period of extraordinary economic growth. So if FIRECalc says that a particular strategy would have succeeded in 99 percent of past periods, I would guess that its probability of future success would be more like, say, 85%. (The exception is that a portfolio of 100% TIPS would have nearly a 100% chance of providing a real rate of return equal to their current pre-inflation yield.)

Fortunately, retirees today have the option of diversifying their assets more than in the past, in a way that it is reasonable to believe will reduce downside risk.
TIPS are a particularly safe bet, and REITs and even high yield bonds have the likelihood of delivering higher total returns than conventional bonds, with less risk than stocks.

I think that the best planning strategy is to test various allocations between stocks and bonds with FIRECalc, and then to partially substitute some of these other assets for stocks and bonds in a person's real portfolio, realizing that there is no "ideal" asset allocation that can be predicted in advance.

[JWR1945]

Please include more options on the re-balancing period. Currently, the calculator re-balances annually. I have seen two sources that suggest that less frequent re-balancing is a better idea.

If you wanted to get fancy, you could introduce all sorts of thresholds (e.g., the portfolio got out of balance by xx% or the total portfolio increased or decreased by yy% from the previous year). But the more basic...re-balance every zz years...kind of re-balancing is the most important.

Thanks. Have fun.

John R.

[MRGALT2U]

I see danger here. Have you heard of paralysis by analysis? I never use anything more advanced than a
pencil and the back of an envelope. Keep it simple

[JWR1945]

The worst feature of safe withdrawal rate calculators in general is how they treat withdrawals. The same withdrawal amount (in real dollars) is taken every year. This is the equivalent of dollar cost averaging...but during distribution instead of accumulation. In a sideways market this would normally reduce the average cost of shares sold by 5% to 10%.

In real life, you have four or five or more years of cash equivalents available. To a limited extent, it makes sense to reduce withdrawals when the market has fallen.

A constant annual withdrawal rate is great for academic purposes. It makes a study transparent. You can understand why things happen. With a real life portfolio it is costly.

There is an opportunity along these lines. Quite a few retirees have already adopted a similar approach. They are holding on to their stocks and drawing down their cash equivalents as long as possible in the hope that stocks will do better...eventually.

Have fun.

John R.

[JWR1945]

johngalt
I see danger here. Have you heard of paralysis by analysis? I never use anything more advanced than a
pencil and the back of an envelope. Keep it simple

Our thoughts are not very different along these lines.

The very nature of any safe withdrawal rate calculator means that we are engaged in a form of datamining. It is OK to draw attention to various relationships in the data. But it is not OK...dangerous even...to rely on them unless you can identify credible and reasonable underlying causes and effects. Then you rely on that underlying causes and their effects, not on the data.

There is a tremendous amount of uncertainty behind any projection. To the extent that we can identify causes and effects, we reduce uncertainty. To the extent that we just look at numbers, we are fooling ourselves.

Have fun.

John R.

[Ted]

It appears that johngalt and JWR1945 appreciate the limitations that are inherent in using any analysis of past investment returns to predict the future, although, as I have said, I think that FIRECalc can provide a better estimate than others because of its flexibility in allowing changes in assets invested, and inflation-adjusted withdrawals, to be made at selected future times.

A basic characteristic of human nature is that many people are forever searching for a financial "guru" who will tell them how to invest their money for maximum gain. Most of the financial services industry earns unwarranted returns at the expense of investors by implying an ability to "beat the market." (They often push the limits of legality in doing this. A major exception is Vanguard, although even they seem tempted to depart from strict adherence to the efficient markets concept since John Bogle is no longer in charge.) Some people might therefore be tempted to regard FIRECalc as their "guru," when, unfortunately, it's not.

JWR1945 mentioned the desirability of retirees reducing their expenditures when the market is down. While I agree that this is a desirable strategy for individuals, the paradox is that when everyone does the same thing, it causes economic activity to slow (i.e., a recession to occur) to everyone's detriment. This is one reason why I oppose the idea of "privatizing" social security, thereby shifting more of people's assets into stocks and bonds. It would probably accentuate business cycles and market fluctuations, and in the process cause a lot more people to be wiped out financially. Back in the "good old days" (when elderly people were relatively few) their families or the county poor farm took care of them, but that's not a very practical approach any longer.

[MRGALT2U]

Okay. I laid aside my stubby pencil and tried FIREcalc.
I'll admit it is pretty cool. I expected it would tell me
that I had to go back to work immediately. Happily,
it looks like smooth sailing. Think I'll buy a boat .

[JWR1945]

Oops! Ted, I misspoke.

Instead of saying In real life, you have four or five or more years of cash equivalents available. To a limited extent, it makes sense to reduce withdrawals when the market has fallen, I should have said...to reduce the portion of each withdrawal that comes from stocks.

In general, this would be difficult to do. But it may be possible to do something.

OTOH, I have seen a study with withdrawal amounts based on a constant percentage of the portfolio balance. It looked promising.

Have fun.

John R.

[wzd]

Quote:

This brings into question the asset allocation exploration idea. Now we're into looking for the best strategy rather than one that would have survived all the financial disasters of the past 130 years.

Why not define best as the best survival rate? I think that is more in line with what Firecalc is - and I thought that is where some of the allocation % numbers came from.

Wayne

[dory36]

The best overall long term asset allocation is a pretty well-studied issue. They use the same data as FIRECalc.

The published reports indicate optimal stock allocations of 48%, 66%, 74%, 77%, 82%, and 85% for pay out periods of 10 Yrs, 20 Yrs, 30 Yrs, 40 Yrs, 50 Yrs, and 60 Yrs respectively.



(From http://www.retireearlyhomepage.com/restud1.html.)

FWIW, I went through several days of exploration with the data I acquired for use with FIRECalc, and found that, surprise surprise, the calculated results matched the publlished numbers.

Dory36

[wzd]

oops....think my tongue got stuck in my cheek there.. I was thinking that Firecalc was using the allocation numbers automatically, but just went and looked and realized I had been putting them in myself, by copying from that very page.

REWRITE suggestion (since that is the topic), and it's an easy one - put the numbers you just quoted on the form as help!


Wayne