I posted a similar piece the other day but Jack pointed out that it accounted for inflation twice. So, I removed that post and went back and looked at the numbers.

The attached PDF is a summary of what I found. It was an interesting study. I’m going to go back and include a fixed income class and see where that leads.

To run these numbers, I used the following assumptions:

• I assumed $1,000,000 on the first day of retirement.

• I deducted an income amount based on the initial withdrawal rate (3% – 10%).

• I invested the remainder in the S&P 500 Total Return Index (no fees are transaction costs were deducted as this was based on the actual index and not a product).

• At the end of each year, I calculated the inflation amount based on the CPI for that year and adjusted the next year’s income based on that number. I deducted that income from the current balance and invested the rest in the S&P 500.

• I performed this calculation for each thirty year period.

• For those periods when the funds were exhausted before the thirty years was up, I took the balance available as income for the final year. Sometimes that number was really low.

Here is the summary of all the data:

**UPDATE:** For those who are interested, here is a **15-page PDF that looks at the numbers in depth at a 4% inital withdrawal rate**. I was going to make them all available until I found out how big the files are.

Very interesting. So an initial 4% withdrawal rate (then adjusted up/down for inflation every year afterwards), has a 93.9% chance of lasting you 30 years if invested solely in the S&P 500.

Well, just one more thing you missed. Don’t withdraw all at the beginning of the year. (Who does that?) My calculations were done using monthly data, taking money out each month.

So in your calculations, you’re missing some returns.

I suspect you will find that that accounts for the failures.

Well…

That’s what the stats say but there’s always a chance the future will be different. For instance, if you retired at the beginning of 2008 with $1 million and took out 4%, you’d ended 2011 with $705,000. Not a great way to start out retirement.

The withdrawals are at the end of the year, Jack.

The way I did the spreadsheet makes them look like they are at the beginning but they’re at the end.

Jack,

I think the failures come from the fact that some 30-year periods had higher inflation rates than others. Most of these “studies” are done with hypothetical inflation rates.

And I’m sorry for the confusion on the withdrawals.

I was using the CPI for inflation. Of course, the monthly data on the S&P500 doesn’t go back quite that far, so one would have to interpolate.

I have monthly total returns for the S&P going back to January 1926. It was the S&P 90 until February 1957.

Jlp) Good point: never begin retirement during a recession…

> The withdrawals are at the end of the year, Jack.

OK — that’s not what you said in the post. In the post you say it’s taken out at the start.

Yes. The first year’s income is taken out BEFORE the rest is invested for that year. Then, at the end of the first year, another withdrawal is made for the next year.

JLP#11 seems to contradict JLP#4.

Jack, you could add six months to the results to approximate what you are saying..

I’m not sure what the issue is here. If you retire at the end/beginning of the year, where are you going to get the income for the first year? Aren’t you going to take it off the top and invest the rest? Then, at the end of the first year, you pull out the income for the next year? How else would you do it?

Interesting..

JLP) Jack is trying to get a few more months of retirement out of the simulation. So if you have $1M in the bank, you are going to take out 4%; which is $40k the first year.

You simulated that you take out all $40k in January of year 1, whereas Jack wants you to take out $3,333 in January/Y1, then the next $3,333 in February/Y1, etc. Though you would need to adjust how much you take out monthly for inflation (instead of yearly), and you are also leaving money in the market (slightly) longer to grab more growth.

No, you aren’t. You are going to take it out maybe a month at a time. Do you save all of the money you’re going to invest, and put it all in at the end of the year? So why would you take it all out at the beginning of the year?

The difference is small — about 0.15% with a nominal 8% return, and about 0.20% with a nominal 12% return — but over 30 years, that can add up. A total of about 4.9% to 5.5% in total assets after 30 years.

I see what you are saying. Actually, I would probably put a couple of years’ worth of income needs into a cash/interest bearing account.

Remember, these are just hypotheticals.