retired at 48:

Here's the question in an example.  Assume someone aged 64 retires 1 Jan 08 with $900,000, with a 4% SWR, and after the first SWR withdrawal next Jan1 his remaining portfolio has grown to $1,000,000.  How can we not treat him the same as a new, 65 year old, retiring 1 Jan 09 (next year) who has accumulated $1,000,000?  Both should be able to use the 4% SWR, on $1MM of assets..  Money knows no past history. 

Can someone explain differently?

Continue your example, but assume that the first to retire doesn't start over.  The first person takes out $36,000 the first year (4% of $900,000).  With 3% inflation, he takes out $37,080 the next year.  That same year, the second to retire takes out $40,000.  From then on, the first will always take out less than the second.  Assuming the same investments, the first will always have more wealth (same starting point, lower withdrawals).  Do you think they both have the same chance of running out of money before death?  Obviously not.

The SWR is based on probabilities.  If the monte carlo study says that there is a 5% chance of running out of money before death, that 5% is calculated over a wide variety of starting conditions.  However, in any particular situation, you either run out of money or you don't.  You can't 5% run out of money.  You do or you don't.  The 5% is based on hundreds of starting dates and market returns, but every one of the individual histories either ends 100% out of money at death or 100% with money left over at death.

Now lets go back to your two retirees.  Clearly number 1 has a lower probability of running out of money than number 2 since number 1 always has more money than number 2.  Perhaps number 1 has a 4.5% chance of running out of money and number 2 has a 5.5% chance of running out of money.  If those were the only two starting points in history, the monte carlo would say that the (overall average) chance of running out of money is 5%, but the two individuals do not have the same chance of running out of money.

If number 1 starts over again in year two with a new 4% withdrawal to match number 2, he has just raised his chance of running out of money to match number 2's.  He may have raised it from 4% to 5% or from 4.5% to 5.5%, but he has raised his probability of running out of money.

The SWR of 4% is based on looking at all retirees, both those who retire in lucky years, just before a market rise, and those who retire in unlucky years, just before a market drop.  By re-retiring each year, you are just guaranteeing that when you finally retire for the last time, it will be in an unlucky year, just before a market drop.  With any luck (sarcasm alert), you will have raised your own risk to a 100% chance of running out of money before death. 

Al