Originally Posted by donheff
You may feel a little under fire here but the questions are not intended to antagonize. You introduced a novel withdrawal algorithm and naturally many of us are curious about how, and if, it works. We have the basic ongoing withdrawal setting approach including the up/down limits. But the starting amount is critical - that is why we are interested in what it is as a percent of the starting portfolio. It can't be irrelevant as you state as can be seen if you imagine starting at a high amount (e.g. 10% or even 50% like Hydro suggested).
Well the starting amount as % of portfolio is actually not relevant - it does not appear anywhere in the algorithm. But the amount itself is of course critical, as are other parameters. My input parameters are: average market gain, average market std. dev., interest rate, inflation rate, tax rate, starting total assets, stock allocation. Output parameters are average/high/low (across 100 trials) of: inflation-adjusted ending withdrawal amount, loss/gain relative to initial withdrawal, ending total assets, and a few other less important measures.
Clearly using a fixed interest rate assumption (I'm using 5%) over 40 years is oversimplifying - but trying to use a random distribution for interest as well as stock mkt would make the whole model too complex for me. The same applies to inflation rate (using 3%).
My starting withdrawal happens to be about 3% of total assets. But this is coincidence. It may well be that the model would not give good results (i.e. worst case trial ends up millions of dollars in the red) if the starting withdrawal was a much higher pct. In that case you'd have to adjust the starting withdrawal, along with stock allocation pct, to see if good results are possible. But for my purposes I kept starting withdrawal as a fixed value, and varied other parts of the model to get the best result.
If anyone's really interested in looking at it in detail, maybe I could put up a simple web page. (Of course, being ERed, this might start to seem like work...)