Monday, February 19

Monte Carlo Retirement Simulation

Something new that I learned about after starting to read personal finance blogs was Monte Carlo Simluation. Today I decided it was time to play around with my numbers on a calculator so I searched the web and found this one which is accompanied by a nice article explaining what it does and why there is a need for it.

The most interesting thing that I found in experimenting with the numbers is that there is less risk (in Monte Carlo terms) in leaving your investments in 100% stocks (even with the increased volatility) if you have a long retirement planned. I found this quite fascinating actually, as I want to have a very long retirement so all the models I ran had 30, 40 or even 50 years of retirement. What this means is that all the long run assumptions came true for me, although volatility was certainly factored in.

I decided to put very little focus on the accumulation phase, since I want it to be relatively short and instead look at the retirement phase, so I set the pre-retirement years to 0 and played around with the retirement numbers.

One interesting question that arose from this exercise is, what should my target success rate be? The Monte Carlo calculator's main result is a percentage chance that your investment will last a certain number of years. A lot of the examples give you the 50% number, but a 50% chance of failure seems extremely high to me. Kind of scary high really. Ideally, I'd rather have a 100% chance of my money lasting through retirement.

But on the flip side, holding out for a very high success rate means that in most cases you will have saved "too much". This means delaying your retirement to save more and working longer than you really had to. That isn't desirable either. Much as it scares me, I think that it makes sense to choose a lower number. One site suggested that above 80% you aren't gaining much (the 80/20 rule again?) so I have chosen 80% as my target number for now.

Also, the author talks about an "adaptive model" that can choose to delay retirement, cut back on retirement expenses or save more in a bad year. I believe that my retirement will certainly be adaptive because I will be actively keeping track of my money and responding to all changes. In fact, most of the stories of early retirees that I have read tell of actually increasing the value of their investments even though they did not plan to do so, so that their retirement grows more secure as time passes, not less. (if they are aware and intelligent about managing their money)

I am hoping to user Monte Carlo simluation to help me to narrow in on my retirement Number and Date. Today was a lot of playing, and I tried a few calculators (which unfortunately gave varying results) so I'm not ready to hold on to anything particular yet, but I do feel that this is a useful tool that I can work with further.

Today was mostly a holiday for me, but I worked a little:

Today's billable hours: 1.5
Today's contracted hours: 0


Preet said...

I use the Monte Carlo Simulation for all my clients - I tell you it's the greatest thing since sliced bread. While the concept is fairly simple, getting a good grip on the exact values of the variables is the tricky part. It's hard using standard deviations for portfolios when there is only a short history for the investments, so it is better IMHO to use a portfolio constructed with modern portfolio theory and extrapolate standard deviations from the theory.

As for what percent success rate you should use? I think it is more of a personal choice but, 75% on the low end and 85% on the high end is usually what I like to see.

I have a financial plan sample on my blog that incorporates the monte carlo analysis if you want to check it out - just look for the "Actual financial plan" topic in the search function and then you can download it if you like.

Would be interested in hearing your thoughts on it...