Most analyses of "safe" withdrawal rates are concerned with withdrawing a certain percentage of your original portfolio (we'll call that strategy [1]), that amount increasing with inflation. That'd make your portfolio income constant in today's dollars.

>By "constant in today's dollars" you mean constant in buying power.
Yes ... reduced for inflation. $10K, ten years from now, may buy what $7K would buy today: $7K would be $10K expressed in today's dollars.

Anyway, there's another strategy that's been around for a while (strategy [2]) that I always thought was pretty silly, namely withdrawing a fixed percentage of your current portfolio. The claim is that you'd never run out of money if, for example, you withdraw 5% of your current portfolio. While that's certainly true, it means that, when your portfolio is down to $1.00, you'd withdraw just five cents.

>So?
So I've done some simulations and find the latter strategy pretty good.

>You learn something every day, eh?
Yes. I always thought that, when your portfolio drops to a few dollars and you're withdrawing a few pennies, you'd be sorry you adopted that strategy.

>But it's better than trying to withdraw $20K from a $5.00 portfolio!
Exactly!

So here's the spreadsheet that I played with.

The nice thing is that ...

>So you like that strategy now?
Well, it's not my favourite strategy, but I like it better than the "usual" strategy where you withdraw some amount (increasing with inflation) and completely ignore market fluctuations.

Indeed, what will often happen with the strategy [1] is that your portfolio has grown from $500K to several million dollars but you're still withdrawing $20K (for example).

>How do I get the spreadsheet?
To download a .ZIPd file, just RIGHT-click on the picture above and Save Target ...

>That's a fictitious set of stock prices, right? Well, yes, but we could also use, say, the S&P 500 and a $100K portfolio. Perhaps the worst 30 years was from about 1965, so if we use strategies [1] or [2], with a 5% withdrawal rate, we'd run out of money in about 22 years >That's with the first strategy, eh? Yes. But, for about 20 years we'd be withdrawing less using [2] and our portfolio has a chance to grow and it winds up, at the end of the 30 years, with over $350K. >What's your inflation? Oh, sorry. It's the actual annual inflation, for the years 1965 - 1995. I should point out that the notorious 4% withdrawal rate would be okay, using strategy [1] ... but 5% ain't good. In fact, with strategy [2], you're withdrawing more when the market is high and less when it's low. >Buy low, sell high? Well, sell high ...

Figure 1

>Okay, so 1965 was a bad year to start your 30 year withdrawals, but what about ...??
Here are some other starting years, again with a S&P portfolio, starting at $100K and a 5% withdrawal:
 

>Wow! Look at the 1950s! You end up with over a million bucks! Yes, and you're withdrawing just 1%, using strategy [1]. >And with strategy 2? It's 5% of course. That's the strategy, eh? It's 5% every year. Look again at Figure 1. The income chart looks like the portfolio chart. One is 5% of the other.

>And the 1990s look pretty good, eh? Even with strategy 2 you end up with ...
Over a million. Yes, those were the days, my friend. An annualized S&P 500 return of about 14% over that 10-year time span ... and about 23% over the last half-dozen years.

>Well, suppose I definitely need a certain income, increasing with inflation and ...
Say, 3%, increasing with inflation ... just to pay the bills?

>Yes! I'll give up travel, restaurants, steak ...
Yes, I understand. You definitely need, say $15K per year from your portfolio and ...

>Yeah, but when the market is good I'd like to withdraw more for travel, steak and .. Okay, I've actually added that to the above spreadsheet. Try it! Figure 2 shows a typical example for this strategy [3]: The blue graph (as we've mentioned) is similar to the portfolio chart. Notice that it starts off very well. The green graph ignores this runup and withdraws a constant $20K (in today's dollars). >Them's strategies [1] and [2]? Actually, they're [2] and [1], respectively. Now comes [3], where we withdraw just $15K (increasing with inflation) but, each year, we calculate 5% of our current portfolio (as in [2]) and if it's larger than our $15K we withdraw that amount. That's the grey graph. There you withdraw more for your steak in those early years but then drop back to $15K (in today's dollars) and ...

Figure 2

>But that runs out of money too!
Yes. Sad, eh?

>So, what if I invest in something other than the S&P, like maybe ...?
Okay, using strategy [2] we'll withdraw 7% of our current portfolio for 30 years from that infamous 4 x 25 portfolio. If we modify the "sensible withdrawal" spreadsheet described here so that we're withdrawing 7% of the current portfolio, then we'd get a Monte Carlo probability about 90% (of ending up with the same portfolio that we started with).

>Aah, the same portfolio, eh? What about the probability of ending up with more than $0?
You mean the MC probability of our portfolio just surviving? That's 100%, of course! With [2] it's always 100%.

>But, with [2], you could be withdrawing a few dollars each year, right?
Well, that's better than running out of money entirely!
(However, in the thousands of simulations, starting with a $1M portfolio, the minimum withdrawal was about $7K).

>Okay, but which thirty years?
It's Monte Carlo, remember? We select, at random, annual returns for each asset class from the years 1928 - 2000 and random sequences of 30 actual inflation rates and do thousands of MC simulations and ...

>And what if you just withdrew 5% of your original portfolio ... increasing with inflation? I mean strategy [1].
The MC probability of surviving (meaning more than $0, after 30 years) is 87%.

>And strategy [3], for that 4x25 portfolio?
I didn't try that ...

>And using some bond component or using years other than 1928-2000 or using different allocations or ...?
zzzZZZ

See Jonathan Clements' article on Squeezing Out Cash In Retirement (Oct 12/03).