Tuesday, April 24, 2007

Reader's Question #5

Jordan left a question on the How Much Do I Need to Retire - Part V post that I thought needed it's own post to address.

Do you have any knowledge or forecasting the success of your retirement strategy with the mathimatical principal called "Monte Carlo".

It's the idea that there is no average return, so it tries to run your investment scenarios against historical variability thousands of times to tell you how likely your investment strategy will be to succeed for X years. (that's my understanding at least)

I don't think most people use it in their calculations or are even aware of what it is.

I've seen some retirement and planning software that offers this feature and thought it seemed very relevant for long term planning and "what if" situations.

I'm sure it could the topic could cover several posts, but I thought I'd ask what you know of it, if you can recommend any specific software and if there is a reason you would or wouldn't use it.

A free online tool is available along with some interesting articles from a retired software developer at:
http://www.flexibleretirementplanner.com/

Thanks very much for your input.
A: To start off with I'm familiar with the Monte Carlo simulation, we actually use them at work to help solve multiple variable problems into a single solution (for those of you who want to know the detailed mathematical uses check out this article).

In retirement calculations they are often used to help simulate the market by producing 'random' investment returns. The calculator will often run an enormous number of calculations over the years of your retirement savings and actual retirement to give you a probability of success that your withdrawals will work (for example 90% or 50%) for your entire planned retirement (yes you have to pick when you die). Since the returns will alter slightly on each run you can input the exact same data more than once to and get slightly different results each time. If your still a bit confused about how this works check out this article (it is a very basic description).

So overall the Monte Carlo can provide some insight on how a portfolio will do over a long period of time. The key word in that last sentence was 'insight' it will not produce a prediction of the future or address all your concerns. It will give you data, but it's up to you to decide if you can live with a 95% probability of success or 85%.

There are numerous calculators out there, some free and some that ask you to buy a copy, that all do the same general thing. The problem with these are they are almost always geared to the US market and often you can't just find the one you want with the correct features. For example I would like one that I could run different phases of retirement through with different pools of cash to generate an overall probability of success for my plan, but I just can't see to find what I'm looking for so far.

If you are from the US, I would suggest using FIREcalc which is likely the best free calculator out there (which is why I have a link to it on the left hand side under Tools called FIRE Calculator).

A few words of caution here. Monte Carlo simulations can be very useful if you know what your doing. If your new to retirement planning or you don't have a great understanding all the variables you can input into these calculators you will most likely just scare yourself to death with some weird numbers and then get depressed about this entire retirement planning thing. Don't worry your retirement plan can still work without a simulation.

My personal view of these calculators is they tend to be used by those who want assurance that they will be fine. The want to know they can survive the great depression again and the interest rates of the 1980's again, which is a bit useless in my opinion. Those events were highly unusual in terms of the history of the market. It's rather like building a house based on the 1 in a 100 year storm weather conditions. Yes your house will stand up to anything, but your going to be dumping in a lot of money to insure yourself for a rare event.

3 comments:

Anonymous said...

Hello Again,

RE: Canadian Tour of Personal Finance blogs #2
I've scheduled the next tour on May 7, 2007.
The Money Diva has accepted to be the host for this event.

Check out http://www.canadian-money-advisor.ca for more details.

I've set up a new email for the tour
tour (dot) canadian (at) gmail (dot) com

BTW... did you see that Google's "Brand" is worth $66.4 Billion?
Google's "Brand" is worth $66.4 Billion?

Monty Loree
http://www.canadian-money-advisor.ca

Jim said...

One comment on FIRECalc is that although it's a great retirement calculator, it calculates its results by back-testing against actual historical market returns rather than running a Monte Carlo simulation.

The difference is that you get far less "test runs" since there aren't that many possible "retirement lifetimes" in the historical record. The upside is that even though there are less scenarios, you can be sure that you're testing against actual scenarios rather than some theoretic model (as in Monte Carlo).

I'm the author of the planner that the original questioner referred to and I think you did a good job of summing up how to interpret the results of these tools.

The real value in the Monte Carlo approach is in the flexibility it can give you in trying to model your plan. For example, in an MCS based calculator, you can try a run where you punch in a market crash in year 5 of your retirement and see what that does. Also, the flexibleRetirementPlanner supports an approach where the withdrawals can be varied depending on how things are going. This is probably closer to how actual retirees will behave rather than keeping spending fixed right up to the end.

Anyhow, I hope you'll give my planner a look. I happen to think IT'S the best one out there (although I may be a little biased).

Jim

Tim Stobbs said...

Jim,

Thanks for that comment and feedback. I forgot to mention FireCalc is historical based rather than Monte Carlo (sorry for the confusion to everyone).

CD