Am I understanding SWR correctly?

[audreyh1]

No re-sequencing of historical returns. Actual sequences only.

[mathjak107]

Quote:

Originally Posted by scrinch I thought that the 90% reflected your statistical chance of not running out of money looking at all historical time frames, not just the worst possible ones. Is that wrong? Does FIRECALC look at the actual sequence of historical returns, or randomly re-sequenced historical returns?

firecalc uses actual data in sequence. the 4% rule is based on the worst of times we had in actual order of how they played out. the fidelity RIP calculator uses monte carlo. i prefer monte carlo myself since over lapping time frames may never ever play out that way again as they did in historical.

case in point is where we stand today . never before have low interest rates and high stock valuations existed in our historical data as it played out

did you know if we removed the two worst time frames the swr would have been a 6.50% average ?


https://www.kitces.com/blog/what-ret...ly-based-upon/

[scrinch]

I use Monte Carlo myself as well. Who's to say that what has happened is the only thing that can happen? On the other hand, random combinations and sequences do fail to catch the economic controls or tendencies behind what actually has happened. I wish I were smart enough to build those tendencies into my Monte Carlo combinations.

What you mean, I think, is that the 4% SWR result stems from what did happen during the worst time periods, but that the overall analysis includes all time frames.

[mathjak107]

the over all analysis used all time frames just so they could identify the worst time frames. then withdrawal rates were tested until a safe level was found with a high level of successfully not running out of money over 30 years.

it is those worst of times the 4% rule is based on.

[scrinch]

Right. But without all those good periods in addition to the worst ones, FireCalc wouldn't be able to estimate the "chance of success" of any withdrawal rate/investment scenario.

Edit: okay, I see you weren't talking specificaly about FireCalc in your first post. Still, to define a SWR as having a 90% POS, the full population of results (or at least the bottom 11%) has to be known.

[mathjak107]

yes and no.

each time frame is looked in seperate context of a 30 year period in the trinity study , bill bengan's work and firecalc.

the results of each 30 year period are looked at . each period contains the best and worst of only its time frame.

then the worst period is pulled out and analyzed to see what could be drawn off it.

i believe 4% did fail in 1965-1966 depending who's data you use and there may have been one or 2 more so when all time frames are considered a percentage of pass fail can be arrived at.

hypothetically if you looked at 10 periods of time and one failed then you had a 90% success rate. the best of each period is not mixed into other periods of time when using historical data like firecalc. if we then take that failed period and see what draw rate let it pass you have the basis for determing a safe withdrawal rate.

we had 17 great years from 1987 to 2003 averaging almost 14% cagr for 17 years . but unless your time frame was part of that those number don't help your period .

in monre carlo simulatons you could pair the best of the gains with the worst of the gains or the worst gains with worst sequencingA and get combo's historical never saw,


i think the best example is the fact that while 1987 to 2003 were fabulous times for gains the time frame leading in were the worst of times.

20 years of poor markets followed by double digit inflation crushed retiiree's as well as few in the accumulation stage were able to save any money for when the best of times came.

so in that period the fact we had the greatest markets wasn't so great since few had much money invested because the time frame before was so poor.

the flaw in historical data is that may never play out in that order again.

[pb4uski]

The dilemma with "safe" withdrawal rates of 3-4% depending on your age when you retire is that since it is effectively based on a worst case scenario it is more likely than not that one would either work longer than you really needed to or die with a huge stash unless you are unlucky enough to retire at a bad time (like 1965-1966) and actually have an adverse sequence of returns.

That links into another thread about updating your withdrawals to allow more spending if your sequence of returns is favorable rather than unfavorable, which I think makes a lot of sense (unless one wants to die with a huge stash).

[kannon]

Very interesting reading. As a new retiree I can see the importance of sequencing. of course we hope we start out retirement with good investments and not hit a few bad first years.

My question is since we are considering Roth IRA conversions, our WR skyrockets over our normal planned WR to cover expenses. To me pulling out a few years of sizable chunks of our money to pay for Roth conversions is about the same as starting out retirement with a few bad years of investments. Which we are being cautioned can lead to a bad retirement financially speaking.

Am I comparing apples and oranges? Are the Roth IRA conversion WRs the same as having a bad market?

Thanks

[pb4uski]

A Roth conversion is not a withdrawal and doesn't affect your WR. It is just moving money from one pocket (a taxable IRA) to another pocket (the Roth IRA). Also, if you sell and buy the same ticker (Stock A to Stock A) when you do the Roth conversion it also has no effect on your asset allocation either.

[REWahoo]

Quote:

Originally Posted by kannon Am I comparing apples and oranges?

Yes. You are comparing paying (what you hope to be lower) income taxes now instead of later on a portion of your retirement savings to a decline (loss) in the value of your overall portfolio due to a market decline. The Roth conversion should increase the net value of your portfolio in the long run. The market decline certainly won't.

And like Pb4 says, a Roth conversion isn't a withdrawal. You aren't spending that money.

[mathjak107]

Quote:

Originally Posted by pb4uski The dilemma with "safe" withdrawal rates of 3-4% depending on your age when you retire is that since it is effectively based on a worst case scenario it is more likely than not that one would either work longer than you really needed to or die with a huge stash unless you are unlucky enough to retire at a bad time (like 1965-1966) and actually have an adverse sequence of returns. That links into another thread about updating your withdrawals to allow more spending if your sequence of returns is favorable rather than unfavorable, which I think makes a lot of sense (unless one wants to die with a huge stash).

The good thing about it being so conservative is healthcare costs have been running at 2x inflation.

that has been killing tye bernicke's study which said retirees should need a lot less inflation adjusting than these calculators figured.


just having better than worst case scenario's has been leaving slack in the plan for soaring healthcare costs

[rmark]

The historical 4% SWR was reduced to about 3% net by taxes.


The historical 4% SWR is only proper for someone retiring expecting 25-30 years of spending, retiring much earlier or later changes the withdrawal rate.

[Meadbh]

Quote:

Originally Posted by pb4uski Good point ... I saw the Canadian pension thing but thought perhaps it was from a prior job and that the OP was in the US. Same principle applies though... do a hypothetical return and include that cost in your expenses. Given OP is Canadian, 10% may not be enough.

I just met with my accountant. Turns out my 2014 tax efficiency was better than anticipated. I am getting a big refund and my overall tax rate was just over 5%.

[Pilot2013]

I have a simple question... Was the OP stating his assumptions (and the associated required starting balance of $1.48 million), based on having that same $1.48 million when he is dead and buried?

SORRY- just saw the answer to my question in his 2nd paragraph.

[Meadbh]

I saw this article posted on another forum and thought it might be of interest here.

Safe Withdrawal Rates In Retirement: Is the 4% Rule Safe?

[CaliforniaMan]

I like the old SWR rule automatically increased with inflation, because it is absolutely the worst thing you could do. No one would actually blindly follow it if they saw their portfolio collapse in the first part of retirement. So it gives a pretty good basic idea of what maximum withdrawal rate would have worked in the past.

For the future, well nobody knows. I am suspicious of people taking in other variables and generating regression models, because it will always be able to beat a simple system. But one can never tell when you are just fitting to the data, and if you fail, you can fail miserably. The old SWR model is simple and stupid and easy to understand, which is precisely why I like it.

Will I follow it? Of course not, I just view it as my worst case past experience model. If we experience a long down trend and high inflation, hey my spending will take a big hit, but hey, we live in the real world.

[hnzw_rui]

Quote:

Originally Posted by CaliforniaMan I like the old SWR rule automatically increased with inflation, because it is absolutely the worst thing you could do. No one would actually blindly follow it if they saw their portfolio collapse in the first part of retirement. So it gives a pretty good basic idea of what maximum withdrawal rate would have worked in the past. Will I follow it? Of course not, I just view it as my worst case past experience model. If we experience a long down trend and high inflation, hey my spending will take a big hit, but hey, we live in the real world.

+1. I like using 3-4% SWR for figuring out the minimum I need to save but plan on using variable withdrawals and a different spending model when I actually retire.

[Midpack]

Quote:

Originally Posted by scrinch I thought that the 90% reflected your statistical chance of not running out of money looking at all historical time frames, not just the worst possible ones. Is that wrong? Does FIRECALC look at the actual sequence of historical returns, or randomly re-sequenced historical returns?

95% (not 90%) is the probability most often associated with the 4% SWR studies. Both SWR studies and FIRECALC use all actual sequence of returns (and inflation) history, not randomly selected or Monte Carlo returns (though an option in FIRECALC).

And while they use all time frames, the worst possible 5% of all time frames define what is SAFE by definition. SWR does not concern the possible upside, only where the lower "safe" thresholds have been. So in a sense SWR and FIRECALC are only using the worst possible time frames...

[Rich_by_the_Bay]

Quote:

Originally Posted by accountingsucks I am planning to have a 3% withdrawal rate and after tax expenses of $40K. This would imply i need $1.32 Million. Now, with 2% inflation, my expenses rise $800 per year but at say 10% tax bracket I need another $80 for taxes for $880 total. $880X33 means i need to build my portfolio from year 1 by $29040 just to keep up with inflation. So to account for both the inflation growth plus my $40K annual spend, I would need $40000+29040 = 69040/1320000 = 5.2% total return on my portfolio assuming I never intend to draw down principal which is what I am aiming for. Am I missing something? Seems abit of a lofty expectation given a 60/40 portfolio. I guess withdrawal rates contemplate you drawing down the balance over time but my preference is not to do that until I get Canadian pension in my mid 60's.

Scott Burns discusses how to spend more for discretionary expenses in the early retirement years, then use inflation-adjusted earnings to ratchet down over time.

A bit off topic but interesting nonetheless...

[mathjak107]

Quote:

Originally Posted by Midpack 95% (not 90%) is the probability most often associated with the 4% SWR studies. Both SWR studies and FIRECALC use all actual sequence of returns (and inflation) history, not randomly selected or Monte Carlo returns (though an option in FIRECALC). And while they use all time frames, the worst possible 5% of all time frames define what is SAFE by definition. SWR does not concern the possible upside, only where the lower "safe" thresholds have been. So in a sense SWR and FIRECALC are only using the worst possible time frames...

which when quantified works out to anytime we see less than a 2% real return average the first 15 years of a 30 year time frame a red flag should go up . that is the danger point of failing and on track to be one of the worst of times ..