I've done quite a bit of analysis recently using FIRECALC on the % of remaining portfolio method where you take a fixed percentage of your retirement portfolio at the end of each year (no adjusting for inflation - just whatever the portfolio gains or loses). This is a method where income can vary quite a bit, so it's only appropriate if you have a lot of discretionary expenses, and can cut back in years after your portfolio does poorly. The upside is that you don't run out of money, your income grows more quickly if the portfolio does well during the initial years, and as a consequence, the ending portfolio tends to be smaller compared to the constant spending method. If your portfolio already provides more income than you need, and you have a lot of discretionary expenses, this might be a good choice, especially if you would like to take advantage of initial good market years. This also has pretty good characteristics for ending not to far from where you started, even after inflation.
Assuming 50% total stock market, 50% 5year US Treasuries, and a 30 year period, 106 runs starting from 1871:
If you withdraw 4.25% of the value of the portfolio each year, after 30 years the average remaining portfolio will be 103% of your starting portfolio in real terms (that means after inflation), and the lowest ending portfolio was 52% of your starting, and the highest ending portfolio was 225% of your starting portfolio.
I consider this to be the more or less "break even" case. But you also might have to tolerate a cut in income by slightly more than half, although that is unlikely. Still, you could go through a period of declining real income for several years, so you have to decide how to deal with that, or whether you have enough discretionary spending that you can cut back without it hurting too much.
If you lower your withdrawal rate to something like say 3.33%, the average ending portfolio was 130% of the starting portfolio, in real terms. The lowest 69% of the starting portfolio, and the highest 300% of the starting portfolio. In the 1966 case, the portfolio and thus income would have dropped to 54% of the starting amount, in real terms, and you would have lived with many years of declining real income before recovering. But you would not have run out of money as you would have it you had used the constant inflation adjusted spending withdrawal. And the 1966 run ended the 30 years a little above break even in real terms with the 3.33% withdrawal rate. It was around 80% for the higher 4.25% withdrawal rate.
Here are the rates I have studied. It gives you an idea of how this withdrawal method behaves with a narrow range of withdrawal rates, and gives you an idea of how historically it has behaved, and what kind of worst income reduction and ending portfolios resulted. Again, these are in real terms, so if inflation is high, you might not see an income reduction in nominal terms - it looks like everything is growing, but your spending power would be hit because your portfolio wasn't keeping up with inflation for a while.
% withdrawal remaining portfolio | average ending portfolio real | lowest ending portfolio real | highest ending portfolio real | Worst case real income reduction (1966 run) | Ending real portfolio(1966 run) |
4.25% | 103% | 52% | 225% | 46% | 78% |
4.00% | 112% | 56% | 244% | 48% | 85% |
3.50% | 130% | 66% | 287% | 52% | 99% |
3.33% | 137% | 69% | 300% | 54% | 105% |
3.25% | 141% | 71% | 308% | 54% | 107% |
3.00% | 152% | 77% | 333% | 57% | 116% |