baseline values?

[tm3]

obviously there are a number of guesses/assumptions to input, starting with retirement age (a factor under some user control) and life expectancy (a factor with less user control that we would like).

i'm curious about the choice of values for inflation, investment return, taxes, etc. clearly with a model like this one would test multiple combinations, but as a starting point do most go with long term historical values such as 3% for inflation, 12% SD for the SP500, and so on?

[jimr]

i'm curious about the choice of values for inflation, investment return, taxes, etc. clearly with a model like this one would test multiple combinations, but as a starting point do most go with long term historical values such as 3% for inflation, 12% SD for the SP500, and so on?

I like your approach of trying multiple combinations of inputs. For most plans, the inputs are so uncertain that the difference between an 80% and 100% probability of success can be meaningless. Estimates of future savings, spending, inflation, and returns are fraught with uncertainty. However, if you only change one input and see a 20% change in the probability of success, that's telling you something important about your plan's sensitivity to that input.

Also, there can be important relationships between the inputs. For example, if you reduce inflation by .5% and leave the investment return alone, the effect is very similar to increasing your estimate of the portfolio's real return by .5%. That's a pretty major change. If you want to see how changes in inflation will impact your plan without assuming increased real portfolio growth, reduce inflation and your portfolio's return by the same amount.

Personally, I usually leave most of these at their defaults to start and focus on the plan's cash flows. However, playing around with inflation, returns, and taxes can be important since a small change in any of these can have a big impact on results.

Jim