Comparing returns to those used in FireCalc calculator?

[frp user]

I've run the simulation multiple times and we traded emails on the Retire Early Forum. In your email you indicated you were conservative in the returns and standard deviations used. Do you happen to have a historical return for the length of data used in the FIRECalc calculator

I've played with various possibilities but I seem to be able to turn "failure" into "success" with what appears to be minor shifts.

Thanks.

[admin]

Hello,

I found a tool from moneychimp that seems pretty good at giving you historical return/std dev data for the S&P 500 at http://www.moneychimp.com/features/market_cagr.htm

The Index Funds Advisors site at http://www.ifa.com/portfolios/p065/index.asp has some good portfolio return/std dev data (rather than just sp500) but I don't think the numbers include expenses. Listed across the top of the page are icons for the various index portfolios and when you click on one, you get a page of data for the selected portfolio. About half way down the page is a table that has return/std dev history for periods from 5 to 79 years.

I don't have that info on FIRECalc, but I'm sure if you posted a question on the FIRECalc support part of the web site someone could get it for you.

Obviously, one of the biggest challenges in using Monte Carlo simulation (or any planning tool) is to guestimate the expected return/std dev for your portfolio.

It may seem surprising that only a .5% reduction in return can cause a successful plan to fail, but when you consider it, .5% is a pretty big chunk of the part of the return that's actually funding your spending (depending how far off retirement is).

Here's my logic (please bear with me). Consider the default 8.5% portfolio return. From that, you need to deduct 3.5% right off the bat to counteract inflation (at the default setting) leaving you with 5%. Then you have to subtract another 1.2% to pay taxes (15% of 8.5%). That leaves you with around 3.8% left to fund spending in retirement (without shinking the portfolio). So a .5% change to the rate of return is really a 15% change in the amount of spending that the portfolio can support. And all that is without considering volatility, or the standard deviation. Volatility can really drag down your chances for success because half of the time you're going to be earning less than the average return. When you get a string of bad years that are well below that average return, you're going to be digging into the nest egg and that could drag your portfolio balance down so far that you can't make it up.

If you're up for more experimentation (or to confuse you more), notice what happens when you decrease inflation and the portfolio return together by the same amount (without changing std dev). You might try a run with inflation at 0% (down from 3.5%) and the return at 5% (from 8.5%). This should help explain how they are connected. Another interesting experiment is to set the std dev to zero and play around a bit. You should notice that plans either work or fail hard, but nothing in between. It's even more clear if you set the spending policy to stable (to turn off Guyton stuff).

The trouble with all of these planning tools is that they are soooo sensitive to slight variation in inputs that it's hard to have faith in the outputs. Obviously, if you've got a spending rate that's way too high, the plan will fail hard unless you use silly estimates for return and std dev. But once you get into the grey area, it's tough to get more precision. In this area, I like to build (in my mind) a "possibilities curve" of what things might happen and what the results would be on my plan. That's the best way I know for how to think about working with these tools. Unfortunately, long range retirement planning doesn't lend itself to black and white thinking or hard answers.

Let me know how you make out...

Jim