Question from New Guy on Retirement Withdrawal Study Posted Here

[sfigurac]

Hi all,

I've seen a lot of good info on this forum and website so thought I'd join.

Like most here I think, I'm interested in early retirement, and am in the assessment/planning stage.

One study I saw (I'm pretty sure it was posted here) showed the results of an analysis to the question of 'How much can I safely withdraw from my portfolio?'.

As I recall, the study showed, for different lengths of retirement (30 yrs, 40 yrs, etc) the maximum percentage of a portfolio that could be withdrawn (and still have money at the end) when compared to the historical DJIA all the way back to pre-Great Depression times.

The study concluded that either 3% or 4% could be withdrawn depending on which length of retirement one used.

That looked great, and so I wanted to validate it myself, so I obtained historical DJIA data and tried to do the same analysis. It appeared to confirm the results.

However then I realized that that analysis did not appear to take into account inflation.

To take that into account, I also factored in actual historical CPI information (back to 1914) to make corrections to the DJIA.

When I did that, the result came out significantly different, with maximum safe withdrawals of as low as 1.7%, especially if one retired as recently as 1965.

Such a correction would greatly impact at what point one could retire - one might need to double the size of one's portfolio before retiring.

Could someone provide some input on this? Have I done that analysis correctly? or have I overlooked something?

Some input would be greatly appreciated.

[ERD50]

Quote:

Originally Posted by sfigurac .... One study I saw (I'm pretty sure it was posted here) showed the results of an analysis to the question of 'How much can I safely withdraw from my portfolio?'. .... The study concluded that either 3% or 4% could be withdrawn depending on which length of retirement one used. That looked great, and so I wanted to validate it myself, so I obtained historical DJIA data and tried to do the same analysis. It appeared to confirm the results. However then I realized that that analysis did not appear to take into account inflation. To take that into account, I also factored in actual historical CPI information (back to 1914) to make corrections to the DJIA. When I did that, the result came out significantly different, with maximum safe withdrawals of as low as 1.7%, especially if one retired as recently as 1965. ...

Did your DJIA data include dividends? Divs make up a substantial part of the overall growth.

A historical reporter like FIRECalc looks at all that in parallel, for every time period in its database (1871 I think) - short version, the 3%-4% numbers should be about right (historically).

-ERD50

[dallas27]

And how splits were calculates long ago also might throw the numbers off.

But good on you for even attempting to check the math.


Sent from my iPhone using Early Retirement Forum

[Midpack]

Welcome to ER.org!

In addition to dividends (which is considerable), I'm wondering exactly how you factored in inflation, and how/what DJIA periods you used. You mentioned making 'corrections to the DJIA' which could be done, though most withdrawal methodologies I've seen factor inflation into spending instead, and it's not hard to trip yourself up with either case.

Also wondering what probability of success was in your approach, the difference between a 95% and 100% success rate is considerable.

Did you include any other asset classes, or a 100% equity portfolio.

A few things that come to mind having not seen your work (not asking to see it).

I take it you don't trust FIRECALC or any of the other withdrawal calculators out there...

[FIRE'd@51]

Serious studies have used the S&P 500, which is a broad-based cap-weighted index of US stocks. The Dow is not really representative of the US stock market since it has only 30 stocks and is price-weighted, as opposed to cap-weighted. I am not at all surprised that the results would be different if the Dow were used.

[sfigurac]

Thanks for all the input.

I haven't researched the more detailed models yet but definitely will check them out.

The DJIA analysis seemed like an interesting one which is partly academic, partly reality-based.

It seemed to me that the goal of the study was to determine - if you invested all of your retirement money into a fund that matched the DJIA in performance, how much could you withdraw and be successful (i.e. not run out of money).

Some of the intriguing things I found about it were:

1. The study relies on real data (historical DJIA data) that represent actual happenings. e.g. big ups and downs etc.
2.

[sfigurac]

Sorry, I hit the wrong button and sent before finished. Let me start all over:

Thanks for all the input.

I haven't researched the more detailed models yet but definitely will check them out.

The DJIA analysis seemed like an interesting one which is partly academic, partly reality-based.

It seemed to me that the goal of the study was to determine - if you invested all of your retirement money into a fund that matched the DJIA in performance, how much could you withdraw and be successful (i.e. not run out of money).

Some of the intriguing things I found about it were:

1. The study relies on real data (historical DJIA data) that represent actual happenings. e.g. big ups and downs etc.
2. The study seems to indicate that you could invest all of your portfolio into stocks (DJIA matching fund) and have a pretty good chance of success (e.g. not running out of money).

I don't believe the study I read, or my analysis, takes into account dividends, but only assumes stock market growth. As pointed out, dividends would improve performance, but the study seems to show that even without that one could be successful.

The DJIA numbers I used I believe were just the numbers as reported in the news, without any adjustment. I'm assuming that historical stock splits were taken into account in the DJIA as they are today to represent actual capitalization.

I wanted to adjust the numbers for inflation since that could have (and actually did have) a big effect on the results, and I wanted to take into account actual inflation rates rather than the typically assumed 3% rate since inflation rates varied quite widely over the years, and the average (based on historical CPI) is actually higher than 3%.

To adjust for inflation, i started at the beginning of the period for which I had both DJIA and CPI data, and divided the DJIA number by (1+%inflation) for that year. I determined the inflation rate as the ratio of the CPI for that year to that of the previous year.

Then I determined what the next year's DJIA should be by assuming that the DJIA grew by the same percentage as the raw DJIA numbers did.

I continued that through the entire period for which I had DJIA and CPI data (1917-2013) to get adjusted DJIA numbers.

Does that seem like a reasonable adjustment?

i think that adjustment also allowed me to use a fixed dollar withdrawal amount, e.g. $20,000 each year against a starting $800,000 portfolio since all the data should now be in equivalent dollars (i.e. adjusted for inflation).

I know that this is a simple approach, but it seemed like it should provide some insight into how a retirement plan would have fared compared against actual historical events with real ups and downs in the market.

What do you think?

[racy]

Hi sfig,
The 4% rule came from work that William Bengen did in the '90s. Also, the 'rule' was based on a mix of equities and bonds. Bengen studied outcomes using stocks from 50% to 75%. He used intermediate term bonds. Google Bengen Wiki for lots of info. And, here is a link to his famous 1994 paper in the Journal of Financial Planning. http://www.retailinvestor.org/pdf/Bengen1.pdf

For a fun read on the 4% rule, check Lee Eisenberg's book, The Number. http://www.amazon.com/The-Number-Wha...rds=the+number

[Live And Learn]

Hello and welcome ! I remember the exact study you are referring to as I've gone back to it a few times. Link below:

Wade Pfau's Retirement Researcher Blog: Retirement Planning Guidelines: An Alternative to the Trinity Study

Hopefully that is the one you are looking for

[Milton]

Quote:

Originally Posted by sfigurac Sorry, I hit the wrong button and sent before finished. Let me start all over

If When this happens again - which it does to almost everybody, don't feel embarrassed! - you can use the 'edit' button to fix the error.

[jIMOh]

I'd add the typical yield of the S&P 500 is around 2.0-2.5%. That is the lowest withdraw rate I would consider reasonable, as dividends tend to keep pace with inflation based on reading I have done (based on past performance).

Quote:

When I did that, the result came out significantly different, with maximum safe withdrawals of as low as 1.7%, especially if one retired as recently as 1965.

Meaning if I hear 1.7% withdraw rate, I would immediately ask if a 2.5% withdraw rate living off dividends only would work (with a higher probability of success).

[sfigurac]

Thanks again everyone for the helpful input. That kind of input is what attracted me to this forum.

I "think" I found the original 'safe withdrawal' study that I had seen before, it's at this link:

The Retire Early study on safe withdrawal rates - Millenniam Edition.

I guess I may have not recalled all the details correctly when I tried to duplicate the results on my own. It doesn't specifically say it used DJIA data but rather "market return" data from Schiller (not sure what "market return" means, does that include the dividends?). It also seems to somehow take into account inflation, and portfolio allocation, and actually shows comparisons to the Trinity study someone mentioned.

I guess being an engineer I wanted to see how all the "nuts and bolts" of the study worked and "tinker" with it myself.

One thing I noticed this time around is their comment that for longer payout periods (50 to 60 years) the optimal stock allocation is around 82 - 85%.

But from what I read, that type of allocation isn't generally recommended, right?

[jIMOh]

Quote:

Originally Posted by sfigurac Thanks again everyone for the helpful input. That kind of input is what attracted me to this forum. I "think" I found the original 'safe withdrawal' study that I had seen before, it's at this link: The Retire Early study on safe withdrawal rates - Millenniam Edition. I guess I may have not recalled all the details correctly when I tried to duplicate the results on my own. It doesn't specifically say it used DJIA data but rather "market return" data from Schiller (not sure what "market return" means, does that include the dividends?). It also seems to somehow take into account inflation, and portfolio allocation, and actually shows comparisons to the Trinity study someone mentioned. I guess being an engineer I wanted to see how all the "nuts and bolts" of the study worked and "tinker" with it myself. One thing I noticed this time around is their comment that for longer payout periods (50 to 60 years) the optimal stock allocation is around 82 - 85%. But from what I read, that type of allocation isn't generally recommended, right?

I don't think it's about recommended or avoided, I think it is about collecting facts and making educated guesses based on past performance, variables a person can control, and accepting what a person cannot control.

For example, if I had to design a portfolio to last 60 years, I would keep it in equities, because over 60 years inflation is a risk which needs to be accounted for. Equity exposure can counter that risk.

Over shorter periods, the volatility of portfolio would pose a risk equal to inflation, and bond exposure (or diversification) can counter the risk of volatility.

[Live And Learn]

Quote:

Originally Posted by sfigurac One thing I noticed this time around is their comment that for longer payout periods (50 to 60 years) the optimal stock allocation is around 82 - 85%. But from what I read, that type of allocation isn't generally recommended, right?

FIRECalc defaults to 75% equities. If you play with FIRECalc you can actually get a chart that shows the success rate with various AA's (just select the radio button on the investigate tab).

[WestLake]

What about investing mainly in dividend stocks? I'm thinking of doing this until SS kicks in. Also, dividends are taxed lower than ordinary income.


Sent from my iPhone using Early Retirement Forum

[samclem]

Quote:

Originally Posted by WestLake Also, dividends are taxed lower than ordinary income.

So are capital gains from "regular" stocks.