Stop worrying about the 4% rule

[johnny847]

Also keep in mind that depending on where your you might want to earn at least $1 of earned income if you want to ridiculously game take advantage of things like the EITC or savers credits.

http://www.irs.gov/Credits-&-Deductions/Individuals/Earned-Income-Tax-Credit/EITC-Income-Limits-Maximum-Credit-Amounts
http://www.irs.gov/Credits-&-Deductions/Individuals/Earned-Income-Tax-Credit/Do-I-Qualify-for-Earned-Income-Tax-Credit-EITC

Those hours working might translate to some big tax credits, depending on how much of your income is coming from Roth sources and/or how much in Roth conversions you are doing :)

For example, let's assume you spend $30k a year and have $100k in Roth IRA principle. You have 5 years you can withdraw $20k of principle and then need to make $10k/year. If you earn $10k/year, using the IRS EITC estimator with 1 child, you are eligible for an EITC of $3,305. If you bump your taxable income to $30k (maybe Roth conversions) you still have an EITC of $2,224. More kids increases this obviously and it may increase it to the point it's worthwhile to have your earned income at the ideal point you get a full EITC.

Since we're talking about gaming err using tax incentives, you might as well put $2k into a Roth IRA and get 50% of that back on AGI up to $36k from the savers credit. So in our hypothetical example, let's say you make $10k, spend $30k (remainder is from Roth IRA withdrawals). You put $2k "back" into a Roth IRA, so when you file taxes you get back $4,305 (EITC of $3305 + $1000 savers credit).

While these numbers are small it should show that a modest income ($10k) can result in a very large percentage of your yearly spending ($14.3k in this case). So your withdrawal from your portfolio is not $30k but just about $15k. This could dramatically change your withdrawal percentage in initial years.

For me personally, where this is important is considering whether to scale down your work before going completely FIRE. It may not be ideal for someone who wants to go 100% working to 0% working and never change that percentage. But for me, if I end up working say 50% or 20% for some years then options for taking advantage of the tax code become numerous. Maybe I am able to scale back to 50% and make say $40k a year gross. I could put a bunch of that into a 401k to lower AGI/MAGI and then basically take advantage of all of the above, too. Perhaps living 50% retired is not what most here want but it is an option that makes the 4% perspective different to consider other scenarios.

Regardless, our tax system is not at all designed for people like mustachians who can save/live under their means on minimal income.. it provides a lot of opportunities for someone able to skillfully control their income to really maximize tax incentives.

Also note that with Roth conversions you can basically pick your taxable income to a dollar every year - have some unused unrefundable credits (such as savers)? Just convert the right amount of pretax to Roth!

Nice write up ender.

For the bolded part, you can even do this after the fact when you file taxes if there's any uncertainty in your income.   Convert more than necessary during the year and retroactively recharacterize the correct amount during tax season. You can even rechar by October 15.

[brooklynguy]

I agree on that danger zone of the first 5-10 years for 30-year periods.  I sure hope it's true for longer periods, but I doubt there's research to confirm it.

Jeremy @ Go Curry Cracker has a blog post in the works on how to determine if/when your portfolio has gotten "too big to fail" (which, given Jeremy's thoroughness, will, I'm sure, include a survey of any existing research) and I have high hopes that it will become the early retirement community's definitive resource on the topic (and I keep mentioning it periodically in part to keep the pressure on him to finish it up so the rest of us can freeload off of his efforts).

I guess it also means that you start a new 30-year period every year until you're in your 60s or even 70s.

It's tempting to think about the chances of portfolio success for super-long periods this way, but I don't think it's really accurate, because a 30+X year retirement period is not a rolling series of 30 year retirement periods, but a single 30+X year retirement period (and perhaps this line of thinking is what caused MMM to spread his bit of misinformation that having enough to last for 30 years is essentially equivalent to having enough to last forever).

Now I side with Nord's idea that the skills that got you there will probably take you pretty far in earning money when necessary.

I got the opposite takeaway from "Voice of Reason" Nords' latest posts in this thread -- that is, reliance on the often-invoked level of safety margin that "I can always return to work" may be misplaced, because (i) running out of money in old age is a catastrophic enough scenario to necessitate insuring against in one form or another and (ii) the reality of earning income during retirement could, to some extent, end up defeating the purpose of retiring (as in the tale of Nords' brother-in-law).

[Monkey Uncle]

I used think that as well.

Now I side with Nord's idea that the skills that got you there will probably take you pretty far in earning money when necessary.

Now, $10k is nothing to sneeze at in the context of $45k/yr total spending, but it doesn't seem like enough to make up for a major portfolio failure.

You might be surprised--run some calculations in cFIREsim.

And your estimates seem really low to me in terms of time necessary to make 10-20k.

Like I said, I used to think he same way, now it seems to me that that there's so many ways to make money, it really isn't a problem.

I'm pretty confident about my estimates of the amount of money I would take in on the business ventures, as I was pretty thorough in constructing those (actually, I was probably too optimistic).  But I probably hi-balled the amount of taxes I would pay.  I forgot about all the EITC and Saver's Credit business, as it's been quite a while since I was able to qualify for those.  Ender's post is great, but I'm not sure how I would feel about claiming credits that are intended for the working poor, especially knowing that the money is coming straight out of the pockets of the still-working middle class.

Anyway, the whole point is that I don't really want to work nearly full time for half the year (or the whole year, in the case of the menial job) after I'm supposedly FIRE.

[ender]

Anyway, the whole point is that I don't really want to work nearly full time for half the year (or the whole year, in the case of the menial job) after I'm supposedly FIRE.

That's fine, but if the comparison is working FT 100% for 3 years or only doing 60% (maybe Mon-Wed) for 5 years it gets less straightforward.

One of the reasons I want to be able to scale back work is to spend time with kids in their earlier years once we have them. Having a reduced work schedule, while working more years but similar total hours, better allows this. It also provides more security if I want to "scale up" for whatever reason (market plummeting, unanticipated higher expenses, whatever) since I'm still doing work.

Plus, there are situations where pensions/benefits better accrue with more years service. My current pension gives me a year credit I think for something like 500 hours in a service year, so working 25% would not only provide the benefits I discussed above but it would provide me more years credit towards a pension.

For what it's worth in situations like this you may benefit from paying off a house earlier too as you reduce your risk in retirement - because you have lower spending to "cover" and volatility affects you more. I might do some musing on this to see how paying off a house early affects situations where you continue to work to meet some/all of your expenses at a higher SWR than if you still had a mortgage payment. If a decent percentage of your expenses is a mortgage then paying it off frees some cashflow.

[johnny847]

I used think that as well.

Now I side with Nord's idea that the skills that got you there will probably take you pretty far in earning money when necessary.

Now, $10k is nothing to sneeze at in the context of $45k/yr total spending, but it doesn't seem like enough to make up for a major portfolio failure.

You might be surprised--run some calculations in cFIREsim.

And your estimates seem really low to me in terms of time necessary to make 10-20k.

Like I said, I used to think he same way, now it seems to me that that there's so many ways to make money, it really isn't a problem.

I'm pretty confident about my estimates of the amount of money I would take in on the business ventures, as I was pretty thorough in constructing those (actually, I was probably too optimistic).  But I probably hi-balled the amount of taxes I would pay.  I forgot about all the EITC and Saver's Credit business, as it's been quite a while since I was able to qualify for those.  Ender's post is great, but I'm not sure how I would feel about claiming credits that are intended for the working poor, especially knowing that the money is coming straight out of the pockets of the still-working middle class.

Anyway, the whole point is that I don't really want to work nearly full time for half the year (or the whole year, in the case of the menial job) after I'm supposedly FIRE.

If the moral aspects of claiming those credits when you don't need it bother you... then just don't claim them.
This isn't a real problem.

[Monkey Uncle]

Anyway, the whole point is that I don't really want to work nearly full time for half the year (or the whole year, in the case of the menial job) after I'm supposedly FIRE.

That's fine, but if the comparison is working FT 100% for 3 years or only doing 60% (maybe Mon-Wed) for 5 years it gets less straightforward.

One of the reasons I want to be able to scale back work is to spend time with kids in their earlier years once we have them. Having a reduced work schedule, while working more years but similar total hours, better allows this. It also provides more security if I want to "scale up" for whatever reason (market plummeting, unanticipated higher expenses, whatever) since I'm still doing work.

Plus, there are situations where pensions/benefits better accrue with more years service. My current pension gives me a year credit I think for something like 500 hours in a service year, so working 25% would not only provide the benefits I discussed above but it would provide me more years credit towards a pension.

For what it's worth in situations like this you may benefit from paying off a house earlier too as you reduce your risk in retirement - because you have lower spending to "cover" and volatility affects you more. I might do some musing on this to see how paying off a house early affects situations where you continue to work to meet some/all of your expenses at a higher SWR than if you still had a mortgage payment. If a decent percentage of your expenses is a mortgage then paying it off frees some cashflow.

I get where you're coming from.  But I think it's highly situation-dependent.  I'm making close to 90k now, so in my situation it would not be as close as your 3 yrs full time vs. 5 years at 60%.  And my kid is grown, so that ship has sailed.  My job does not provide a part-time option that would still contribute to my pension (but continuing full-time increases my pension by 1% of hi-3 salary for each additional year, plus my hi-3 goes up as I continue to get COL and step increases).  The house thing could apply, though.  I probably wouldn't straight-up pay it off, as I have a 3 1/8% interest rate.  But DW and I are talking about downsizing, which has the potential to reduce our housing expenditure.  If we stick with the current house, it will be paid off in 12 years, and our annual expenses will drop by about $10k.  I project that I'm 3-4 years from FIRE, so if I were to do PT work, I'm probably looking at 8-9 years to get to where expenses drop to the point that I don't really need the extra money (unless we downsize the house).  We'll see how the numbers look when I get there, but if it's a choice between OMY (or TMY) and 8 or 9 years of PT work, I'd probably pick the former.  I'm hoping the numbers look good enough that I don't have to choose either of those options.  How good is that?  If I can manage a 90-95% success rate in cFiresim/FireCalc and an 80-85% success rate in a Monte Carlo simulator, I'd probably feel good enough to pull the plug completely.

[forummm]

Anyway, the whole point is that I don't really want to work nearly full time for half the year (or the whole year, in the case of the menial job) after I'm supposedly FIRE.

That's fine, but if the comparison is working FT 100% for 3 years or only doing 60% (maybe Mon-Wed) for 5 years it gets less straightforward.

One of the reasons I want to be able to scale back work is to spend time with kids in their earlier years once we have them. Having a reduced work schedule, while working more years but similar total hours, better allows this. It also provides more security if I want to "scale up" for whatever reason (market plummeting, unanticipated higher expenses, whatever) since I'm still doing work.

Plus, there are situations where pensions/benefits better accrue with more years service. My current pension gives me a year credit I think for something like 500 hours in a service year, so working 25% would not only provide the benefits I discussed above but it would provide me more years credit towards a pension.

For what it's worth in situations like this you may benefit from paying off a house earlier too as you reduce your risk in retirement - because you have lower spending to "cover" and volatility affects you more. I might do some musing on this to see how paying off a house early affects situations where you continue to work to meet some/all of your expenses at a higher SWR than if you still had a mortgage payment. If a decent percentage of your expenses is a mortgage then paying it off frees some cashflow.

I get where you're coming from.  But I think it's highly situation-dependent.  I'm making close to 90k now, so in my situation it would not be as close as your 3 yrs full time vs. 5 years at 60%.  And my kid is grown, so that ship has sailed.  My job does not provide a part-time option that would still contribute to my pension (but continuing full-time increases my pension by 1% of hi-3 salary for each additional year, plus my hi-3 goes up as I continue to get COL and step increases).  The house thing could apply, though.  I probably wouldn't straight-up pay it off, as I have a 3 1/8% interest rate.  But DW and I are talking about downsizing, which has the potential to reduce our housing expenditure.  If we stick with the current house, it will be paid off in 12 years, and our annual expenses will drop by about $10k.  I project that I'm 3-4 years from FIRE, so if I were to do PT work, I'm probably looking at 8-9 years to get to where expenses drop to the point that I don't really need the extra money (unless we downsize the house).  We'll see how the numbers look when I get there, but if it's a choice between OMY (or TMY) and 8 or 9 years of PT work, I'd probably pick the former.  I'm hoping the numbers look good enough that I don't have to choose either of those options.  How good is that?  If I can manage a 90-95% success rate in cFiresim/FireCalc and an 80-85% success rate in a Monte Carlo simulator, I'd probably feel good enough to pull the plug completely.

Similarly, we have a good situation setup where we're saving a lot of money mostly due to VLCOL. But I would prefer to take a different job that pays less, maybe requires DW to get paid less/be unemployed for awhile to find something, and has a VHCOL. So we would probably not save anything, or very little. Figuring out the timing of how long to keep the status quo (which is fine, but not interesting to me beyond money) vs FIRE vs job more in line with my passion (but still a job I might not like) vs a job which would pay me a lot more but is still something I'm not interested in (status quo but with more money and more stress). The last one is the only one I've ruled out. I'm stuck with the status quo for at least 8 months (half of it on leave) due to impending parenthood.

[brooklynguy]

For what it's worth in situations like this you may benefit from paying off a house earlier too as you reduce your risk in retirement - because you have lower spending to "cover" and volatility affects you more. I might do some musing on this to see how paying off a house early affects situations where you continue to work to meet some/all of your expenses at a higher SWR than if you still had a mortgage payment. If a decent percentage of your expenses is a mortgage then paying it off frees some cashflow.

This is starting to get a bit off topic, but this isn't really accurate.  If you choose to retain a mortgage in retirement instead of paying it off, then the pile of investments that you have in lieu of paying off the mortgage (which can be mentally accounted for as the portion of your portfolio that is "earmarked" for servicing the mortgage) will "cover" the monthly mortgage expense and the decision to retain the mortgage therefore has no effect on your overall cashflow (putting aside tax considerations, transaction costs, and impact on eligibility for means-based financial incentives, if any (but, generally speaking, if the math favors keeping a mortgage before consideration of these factors, it will continue to do so even after consideration of these factors)).  It is true that retaining a mortgage can increase your "risk," but if the mortgage rate is low enough and the remaining life to maturity is high enough to justify the strategy of retaining it instead of paying it off, then that risk is probably lower than the equivalent risk of early retiring on a SWR-based retirement strategy in the first place (plus, retaining the mortgage can actually decrease certain risks in retirement, such as the risk of inflation eating away at your portfolio).  There's lots more discussion and detail on all of this in the various "to prepay or not to prepay" mortgage threads.

[sol]

I did a little FIREcalc meta-analysis to get a better feel for Safe Withdrawal Rates.

For each specified time period, I just recorded the success probabilities for each SWR.  I used the default 75/25 stock-bond split and a million dollar portfolio with 30k/year for 3%, 40k/yr for 4%, etc. 

The data are a little lumpy because I only did integer SWRs and the results are based on historic records, where random chance plays some role in distorting these lines from nice smooth curves.

[sol]

Some things to note on this graph:

1.  3% is 100% safe for all periods.  There are no periods in the historical record where a 3% SWR has ever failed, regardless of timing or duration. 

2.  Periods longer than 40 years are basically statistically identical.  Lasting 40 and 60 are a basically equally likely.  This may just be a result of there being so few 50 to 60 year periods in the historical record that they are largely identical.  Most of the 60 year periods are going to contain most of the same years as all of the 50 year periods, so these lines are mostly the same.

3.  The biggest increases in SWR come from shortening your time period.  This suggests to me that annuities like social security are hugely valuable, probably worth significantly depleting your nest egg.  For example, if you can buy a social security supplemental annuity at age 52 to cover all of your expenses after age 62, and by so doing lower your expected period of retirement drawdown from 30 years to 10 years, you can effectively increase your withdrawal rate on the remaining nest egg from 4% to 8.5% (from green's 95% to blue's 95% line) without negatively impacting your success probability.  That means it would be worth spending over half of your nest egg to get that annuity.  Short planned drawdown periods seem to make retirement super easy.

4.  The 50% success rate line on the vertical line gives you the average expected value for any time period.  For example, for a 30 year period the average safe withdrawal rate that has historically been about right is about 6%.  For periods longer than 40 years it's been closer to 5%.

5.  For periods 30 years or longer, this line gets steeper as you lower your SWR up to 4%.  This means you continue to see increasing large benefits from your continued savings up until you reach about 4%, but working to lower your SWR below 4% suddenly gets much harder.   The odds just don't move very much for your efforts beyond that.

6.  One way to look at this graph is to randomly sample it by printing it out and and throwing darts at it.  The first dart that lands anywhere between the red and orange lines is your retirement lifespan lasting between 20 and 60 years.  In order for a 4% SWR to fail you, your dart would have to land between the red and orange lines in the top left corner of the graph, to the left of the 4% line, AND you'd still have to live at least 20 more years to deplete it.

[MDM]

I did a little FIREcalc meta-analysis to get a better feel for Safe Withdrawal Rates....

Nicely done!

[sol]

Another way to look at this same information is to ask how long, on average, a given SWR should last.  I did this by iteratively seeking the number of years that FIREcalc needed to give me a 50% success rate for each specified SWR.

So by comparison with the above graph, the statistically "correct" average SWR for a 30 year period is about 6%, which is where the green line (for 30 years) is at 50% on the above graph.  This new graph below flips that relationship around to find the average time for each specified SWR, rather than the average SWR for each specified time, but it gives the same value: a 6% SWR should last about 30 years before being depleted to zero.

This plot reports an infinity value for 4% SWR or less, because these low SWRs never fall to 50% success rates over any time period.  On average, SWRs below 5% will last forever.  Most of you already know this, because you already know that the average ending portfolio value for a 4% SWR plan is roughly twice the starting value.

I find this graph more interesting than the first one, in some respects.  Once you've managed to save up 10x your expenses, your nest egg should last you an average of 15 years.  Saving anothe ~10% to get down to 9% SWR only gets you to 20 years.  But once you cross below that 6% SWR level (roughly 17x expenses) the time periods really start to take off.  Each tiny additional bit of savings is then buying you a huge number of additional years of retirement.

[PowerMustache]

Sol, I really like that last graph. In addition to illustrating the safety of 4% from yet another angle, it also provides an encouraging metric to show the big impact of those dollars saved in the final years leading up to FI (for those of us who think 4% is a good SWR). Before seeing this chart, I was thinking those years building up the net worth to achieve 4% rather than 6% WR might become a bit boring from a tracking perspective -- it's not as exciting to watch the portfolio grow that last 30% as it is to watch it grow the first 30%. Now, I plan to visualize this chart for inspiration during that final 30% of net worth growth before FI.

[brooklynguy]

I did a little FIREcalc meta-analysis to get a better feel for Safe Withdrawal Rates....

Nicely done!

I agree, great work!  But I'm curious why you used FIREcalc instead of cFIREsim, given the known flaw in [EDIT] cFIREsim's FIREcalc's bond data?  However, I just ran a few comparison simulations to spot check the differences in the outputs between the two alternatives, and, with the default allocation of only 25% bonds, the discrepancy in the bond return calculation method does not appear to have a material impact on the outputs for purposes of your analysis.

In addition, I have a quibble with this portion of your analysis:

3.  The biggest increases in SWR come from shortening your time period.  This suggests to me that annuities like social security are hugely valuable, probably worth significantly depleting your nest egg.  For example, if you can buy a social security supplemental annuity at age 52 to cover all of your expenses after age 62, and by so doing lower your expected period of retirement drawdown from 30 years to 10 years, you can effectively increase your withdrawal rate on the remaining nest egg from 4% to 8.5% (from green's 95% to blue's 95% line) without negatively impacting your success probability.  That means it would be worth spending over half of your nest egg to get that annuity.  Short planned drawdown periods seem to make retirement super easy.

This analysis assigns no value to having a positive terminal portfolio value at the end of the retirement period (which may be a valid assumption for some people, but, all else being equal, I think most people would prefer to have something instead of nothing at retirement end (i.e., death) in order to pass on to their heirs, give to charity, throw themselves a Great Gatsby level party of a funeral, or whatever).

In your example, spending half of your portfolio to purchase the supplemental annuity (which would have no impact on your success probability and almost no impact on your withdrawals in absolute dollars) would cost you a historically median amount of $258K in terminal portfolio value for every $100K of your portfolio that was spent on the annuity.  That is, if your portfolio was $1 million at age 52, and you spent half of it ($500k) on an annuity, then, at the projected end of the retirement period (age 82, in your example), you would be left with a portfolio of $0 (but, of course, you would still have the entitlement to continue collecting the annuity payments if you continue to live beyond that point), whereas, had you not purchased the annuity, you would at that point have had a terminal portfolio having a historical median amount of $1,290,000.  Given that all else is practically equal between the two scenarios, I'd rather have a $1,290,000 portfolio when I turn 82 than have zero dollars and the right to collect $40K per year for the remainder of my life.

[Cheddar Stacker]

Sol, You are awesome. Thank you.

[SuperSecretName]

Regardless, our tax system is not at all designed for people like mustachians who can save/live under their means on minimal income.. it provides a lot of opportunities for someone able to skillfully control their income to really maximize tax incentives.

Great write-up, and I agree on the tax code.  It is not meant for people like us.  Given how badly big business cheats optimizes their taxes, I don't feel bad about doing the same with mine.

[Clean Shaven]

Sol, You are awesome. Thank you.

X2. Thanks!

[forummm]

[brooklynguy]

In recognition of sol's extraordinary efforts towards examining SWRs from new angles, I propose that we rename this thread "Dr. Sol or:  How I Learned to Stop Worrying and Love the 4% Rule."

[Cheddar Stacker]

This article is nothing new for anyone around here:
http://www.marketwatch.com/story/how-traditional-retirement-formulas-fall-short-2015-07-15?dist=afterbell

I linked to it here because I think it highlights just one more reason to stop worrying about the 4% rule even though that's not really his conclusion.

It sort of reminded me of this GCC post: http://www.gocurrycracker.com/the-worst-retirement-ever/ because he cherry picked an awful year to retire (1970) and still showed ways to succeed.

The article in summary: if you retire in a very bad year for sequence of return risk, and you draw 5%/year adjusted for inflation, and you invest conservatively in a 50/50 stock bond portfolio, you will run out of money in year 38. By adjusting to 60/40 and still drawing 5% you would still have 180% of your original starting value after 43 years.

My conclusion: So draw 4%, or invest more aggressively than 50/50. Or don't be so rigid with spending exactly 4/5% plus inflation. Pretty simple.

[brooklynguy]

The article in summary: if you retire in a very bad year for sequence of return risk, and you draw 5%/year adjusted for inflation, and you invest conservatively in a 50/50 stock bond portfolio, you will run out of money in year 38. By adjusting to 60/40 and still drawing 5% you would still have 180% of your original starting value after 43 years.

Something seems grossly amiss with this data (but first of all, note that the tables appear to be using nominal figures, so, on an inflation-adjusted basis, you wouldn't have "180% of your original starting value after 43 years" even assuming that the tables' data is accurate -- but that's just a footnote, because the data in the tables seems to be wildly inaccurate).

As you summarized, the table shows that a 50/50 portfolio with a 5% WR would not run out of money until year 38.  But cFIREsim says a 50/50 portfolio with a 5% WR starting in 1970 would have run out of money in less than 20 years.  Similarly, as you said, with a 60/40 portfolio, the table shows a positive balance of more than 180% of the original balance (in nominal terms) after 43 years, but cFIREsim shows total portfolio depletion in less than 20 years for this scenario as well.

What gives?  Obviously there are some discrepancies between the underlying data used by cFIREsim and this article's tables (for example, cFIREsim uses a total equity index while the article uses the S&P 500) and probably also in other assumptions (investment fees, timing of withdrawals, etc.), but none of this should be sufficient to explain such outrageously different results.

[forummm]

The article in summary: if you retire in a very bad year for sequence of return risk, and you draw 5%/year adjusted for inflation, and you invest conservatively in a 50/50 stock bond portfolio, you will run out of money in year 38. By adjusting to 60/40 and still drawing 5% you would still have 180% of your original starting value after 43 years.

Something seems grossly amiss with this data (but first of all, note that the tables appear to be using nominal figures, so, on an inflation-adjusted basis, you wouldn't have "180% of your original starting value after 43 years" even assuming that the tables' data is accurate -- but that's just a footnote, because the data in the tables seems to be wildly inaccurate).

As you summarized, the table shows that a 50/50 portfolio with a 5% WR would not run out of money until year 38.  But cFIREsim says a 50/50 portfolio with a 5% WR starting in 1970 would have run out of money in less than 20 years.  Similarly, as you said, with a 60/40 portfolio, the table shows a positive balance of more than 180% of the original balance (in nominal terms) after 43 years, but cFIREsim shows total portfolio depletion in less than 20 years for this scenario as well.

What gives?  Obviously there are some discrepancies between the underlying data used by cFIREsim and this article's tables (for example, cFIREsim uses a total equity index while the article uses the S&P 500) and probably also in other assumptions (investment fees, timing of withdrawals, etc.), but none of this should be sufficient to explain such outrageously different results.

This isn't a very useful study. It's a backtested portfolio that conveniently uses the asset classes that went crazy during that time period. "100% Stocks" means 20% LC, 20% LCV, 20% SC, 20% SCV, and 20% REIT.

[brooklynguy]

This isn't a very useful study. It's a backtested portfolio that conveniently uses the asset classes that went crazy during that time period. "100% Stocks" means 20% LC, 20% LCV, 20% SC, 20% SCV, and 20% REIT.

Where does it say that?  Doesn't the article say the "100% stock" portfolio is the S&P 500?

EDIT:  Oh I see now - I misread the reference to the "column at the right" to mean the "100% stock" column and didn't notice the separate S&P 500 column.

[Cheddar Stacker]

Ok then. I didn't verify anything. Disregard.

[forummm]

I'll be interested to see that analysis, Ender. I assume you'll also include the differences that result from longer or shorter remaining mortgage terms, and higher and lower interest rates. With only 5 years left on the term, perhaps the WR can be really high since you'll have the huge drop in spending in 5 years. With the typical 6% interest rates historically, paying off your mortgage seems like a no-brainer. It's with the historically low 3% rates now that make us wonder.

[brooklynguy]

Ender, your analysis in the above post is flawed for the same reason I described in post # 107 above.  I'm reluctant to go into too much detail because it seems off topic for this thread and has already been extensively discussed in several of the "to pay off or not to pay off" mortgage threads (like this one), but in my view the proper way to think about it once you have a sufficient portfolio balance to pay off the mortgage in full is to mentally divide your portfolio into two buckets:  one devoted to servicing the mortgage principal + interest, and another devoted to servicing all of your non-mortgage related expenses.  So, if your mortgage rate is low enough and remaining life to maturity high enough to justify retaining the mortgage, then paying it off does not result in increased chances of history-based portfolio success (instead, it can result in decreased chances of history-based portfolio success, in addition to leaving you with a lower historical terminal portfolio value).  When you run your cFIREsim/FIREcalc analyses, don't forget to set the principal + interest payments to "not inflation adjusted" (while all the non-mortgage related expenses, including property taxes and insurance, should be adjusted for inflation).

EDIT:  For the sanity of anyone reading this in posterity, this post was made in response to a post by ender that was apparently subsequently deleted (as was the post that post # 107 was responsive to, but in that one I had quoted the portion I was responding to so that one should be easier for future readers to make sense out of).

[a1smith]

^^  I agree.  My mortgage is 2.97% and there is no way I'm paying it off early.  Easy math problem.

[makincaid]

Awesome graph Sol!

I would love to see it for 80% survival as well.

[a1smith]

. . . . .

I find this graph more interesting than the first one, in some respects.  Once you've managed to save up 10x your expenses, your nest egg should last you an average of 15 years.  Saving anothe ~10% to get down to 9% SWR only gets you to 20 years.  But once you cross below that 6% SWR level (roughly 17x expenses) the time periods really start to take off.  Each tiny additional bit of savings is then buying you a huge number of additional years of retirement.

Your graph is also a great way to illustrate that reducing your expenses even a little (for SWR < 6%) has a huge impact on how long your money will last due to the high sensitivity (slope) in that region of the graph.

[MDM]

Your graph is also a great way to illustrate that reducing your expenses even a little (for SWR < 6%) has a huge impact on how long your money will last due to the high sensitivity (slope) in that region of the graph.

Good point.  See linearized sensitivities below for various variables in the simplified "time to FI" calculation, given the first five numbers as inputs.

Quick calculation of "Time to FI"
Planned Withdrawal Rate	WR	4%
Annual Savings Invested	S	50000	$/yr
Annual Retirement Expenses	E	40000	$/yr
Current Assets Invested	A	100000	$
Investment return	r_	6%
Time to FI	t	11.6	yr
Sensitivity of Time to FI for each input
E/WR, stash needed	G	1000000	$
	dt/dr_	-0.6	yr/%
	dt/dS	-0.2	yr/$K
	dt/dG	0.9	yr/$100K
	dt/dA	-1.8	yr/$100K
	dt/dE	2.3	yr/$10K
	dt/dWR	-2.3	yr/%
Calculate Savings needed based on time
Desired time to FI	t	10.00	yr
Annual Savings Needed	S	62281	$/yr

[EscapeVelocity2020]

Sensitivity of Time to FI for each input
E/WR, stash needed	G	1000000	$
	dt/dr_	-0.6	yr/%
	dt/dS	-0.2	yr/$K
	dt/dG	0.9	yr/$100K
	dt/dA	-1.8	yr/$100K
	dt/dE	2.3	yr/$10K
	dt/dWR	-2.3	yr/%

Interesting summary, but why compare dt/dS in yr/$k vs. dt/dE in yr/$10k?  For most Mustachians, I would think it is just as likely to be able to save an extra 10k/yr for the earnings years as it would be to permanently lower annual expense from 40k to 30k.  This would then make both sensitivities about equal, I think...

[MDM]

Interesting summary, but why compare dt/dS in yr/$k vs. dt/dE in yr/$10k?

Don't remember why.  Do remember not thinking about it all that much.  Probably started down the list with unit ratios, increased the ones in the middle, then picked an in-between number for dE.

For most Mustachians, I would think it is just as likely to be able to save an extra 10k/yr for the earnings years as it would be to permanently lower annual expense from 40k to 30k.  This would then make both sensitivities about equal, I think...

The sensitivities do have the same order of magnitude.  In this case, looking at two significant digits, dt/dS=-1.5$/10K.  The calc'ns are on the 'Misc. calcs' tab of the case study spreadsheet if anyone would like to explore more.

[ysette9]

... dt/dS in yr/$k vs. dt/dE in yr/$10k? ...

Ah, be still, my heart. I love it when you talk nerd to me! :)

[MoonShadow]

... dt/dS in yr/$k vs. dt/dE in yr/$10k? ...

Ah, be still, my heart. I love it when you talk nerd to me! :)

https://www.youtube.com/watch?v=BdB0P9fx7Es

See what you made me do!

[a1smith]

Your graph is also a great way to illustrate that reducing your expenses even a little (for SWR < 6%) has a huge impact on how long your money will last due to the high sensitivity (slope) in that region of the graph.

Good point.  See linearized sensitivities below for various variables in the simplified "time to FI" calculation, given the first five numbers as inputs.
 . . . . . .

About what operating point are the equations linearized at?  Is it the values listed under Quick calculation of "Time to FI" heading?

[MDM]

Your graph is also a great way to illustrate that reducing your expenses even a little (for SWR < 6%) has a huge impact on how long your money will last due to the high sensitivity (slope) in that region of the graph.

Good point.  See linearized sensitivities below for various variables in the simplified "time to FI" calculation, given the first five numbers as inputs.
 . . . . . .

About what operating point are the equations linearized at?  Is it the values listed under Quick calculation of "Time to FI" heading?

Yes, those are the "first five numbers" in question.  The equations are in the linked spreadsheet mentioned a few posts back.

[forummm]

https://www.kitces.com/blog/how-has-the-4-rule-held-up-since-the-tech-bubble-and-the-2008-financial-crisis/

The 4% rule has been much maligned lately, as recent market woes of the past 15 years – from the tech crash of 2000 to the global financial crisis of 2008 – have pressured both market returns and the portfolios of retirees.

Yet a deeper look reveals that if a 2008 or even a 2000 retiree had been following the 4% rule since retirement, their portfolios would be no worse off than any of the other "terrible" historical market scenarios that created the 4% rule from retirement years like 1929, 1937, and 1966. To some extent, the portfolio of the modern retiree is buoyed by the (only) modest inflation that has been occurring in recent years, yet even after adjusting for inflation, today’s retirees are not doing any materially worse than other historical bad-market scenarios where the 4% rule worked.

[brooklynguy]

I just came here to post the Kitces article and I see forummm beat me to it.  I thought the article was terrific, but note that, as usual for Kitces, his numbers (based on a 60/40 stock/bond portfolio) are a bit more optimistic than what cFIREsim reports for the same allocation (but otherwise using its default settings) (as discussed in this thread).

Also, I wish he would have included an equivalent chart to the first "historical worst year comparison" chart using real instead of nominal data (and it seems odd that he didn't).

[Nords]

I just came here to post the Kitces article and I see forummm beat me to it.  I thought the article was terrific, but note that, as usual for Kitces, his numbers (based on a 60/40 stock/bond portfolio) are a bit more optimistic than what cFIREsim reports for the same allocation (but otherwise using its default settings) (as discussed in this thread).

You guys both beat me.  Now I'm going to have to tweet about it instead.

Raddr's been tracking this on his forum for over a decade.  Here's the initial setup:  http://raddr-pages.com/forums/viewtopic.php?f=2&t=1208
"I've finally gotten around to calculating the 2004 year end balance of the portfolio of a hypothetical investor who retired at the end of 1999 with a 75:25 mix of S&P500 stocks and 6 mo. commercial paper."
Here's the 2014 update, at a withdrawal rate of over 10%:  http://raddr-pages.com/forums/viewtopic.php?f=2&t=1208&start=390#p52998

Here's a more in-depth discussion of various asset allocations:
http://www.early-retirement.org/forums/f28/firecalc-and-the-hapless-y2k-retiree-69942.html

Yet the Hapless Y2K Retiree still might make it.  And if his portfolio can survive this mindless spending, then I think every human starting with a 4% SWR can make it to at least Social Security.

[k9]

Right, so now we're assuming that the future will be worse than anything we've ever had and as bad or worse than any other country has ever had in the last 200 years as well?

Seems pretty paranoid.

Not sure about what you mean here. According to the data from your other post (https://1.bp.blogspot.com/_7pcnNDkYOFE/TJLpRdCn9uI/AAAAAAAAAOI/e4WRVdW-D4U/s1600/Capture2.JPG), the worldwide SWR is about 0.25%. 400 years of income. That's pretty scary.

As a French investor, I should target a 0.95% SWR. A 4% WR for 30 years would have failed most of the time. It kinda sucks.

[arebelspy]

You cut off the rest of my post, taking that out of context.  Read the rest of it.

If you want to argue for a sub-1% SWR (or .25%, or whatever number you pick), like I said, you're essentially saying that the future will be worse than anything we've ever had and as bad or worse than any other country has ever had in the last 200 years as well.

That seems pretty pessimistic.

In other words, yes, your post of needing a sub-0.25% SWR is saying the future will be much worse than ever, anywhere, in the last several hundred years.

I just don't see it, and if that is the case, I don't think any SWR will help, even if it's 0.01% as we'll be in total economic collapse.

[k9]

Oh, anyway, I'm not really scared about that SWR stuff. As mentioned a lot of times even in this topic, there are many options beside SWR. As a French citizen, I have a substantial pension waiting for me starting from my 60s, and I can work more if needed until that. I will probably inherit from my parents before I am not able to work anymore. Health insurance is quite cheap here. And I could well be dead before I retire, anyway, so...

I'm just questioning the geographical bias of US investors. Your country has had an amazing century and you seem to take that for granted. No war. No hyperinflation. No sovereign default. No communist or fascist dictatorship. No multidecades bear market. This is exceptionally good.

4% is the worst case scenario for the best case scenario. I think it is a reasonable overall tradeoff (and I'm aiming for that), but it cannot just be considered an outlier.

[MoonShadow]

I'm just questioning the geographical bias of US investors. Your country has had an amazing century and you seem to take that for granted. No war. No hyperinflation. No sovereign default. No communist or fascist dictatorship. No multidecades bear market. This is exceptionally good.

There are a great many of us here that would argue that was not simply luck, and many others that would argue that we've had quite a few close calls on all of that.

[Nords]

I'm just questioning the geographical bias of US investors. Your country has had an amazing century and you seem to take that for granted. No war. No hyperinflation. No sovereign default. No communist or fascist dictatorship. No multidecades bear market. This is exceptionally good.

So you're essentially saying that past history is no indication of future performance?

The point of this thread is that the 4% SWR is a reasonable estimate of success.  If you're trying to insure against failure then you'd use annuities and enough savings that you'd never have to touch the principal.  But for people who are wondering "How much is enough?" the 4% SWR is a starting point.

No war.

Speaking as a military veteran, I find this statement incomprehensible as written.  Feel free to elaborate.

[MoonShadow]

No war.

Speaking as a military veteran, I find this statement incomprehensible as written.  Feel free to elaborate.

No warfare within our national borders.  On our soil.

You have to remember, he is speaking from the perspective of a French citizen.  When you are born French, war comes to you.

[k9]

I'm just questioning the geographical bias of US investors. Your country has had an amazing century and you seem to take that for granted. No war. No hyperinflation. No sovereign default. No communist or fascist dictatorship. No multidecades bear market. This is exceptionally good.

So you're essentially saying that past history is no indication of future performance?

The point of this thread is that the 4% SWR is a reasonable estimate of success.  If you're trying to insure against failure then you'd use annuities and enough savings that you'd never have to touch the principal.  But for people who are wondering "How much is enough?" the 4% SWR is a starting point.

Yeah, I agree with that. But the point was raised at some point that the 4% SWR was a worse case scenario, and that at no point in history the situation could have been worse. This is somewhat biased.

No war.

Speaking as a military veteran, I find this statement incomprehensible as written.  Feel free to elaborate.

Yeah, sorry. MoonShadow explained. No war *on your soil*. Your country has never been invaded, your cities being destroyed, your country being ruled by a foreign army, etc. Except for military men like you, this makes a huge difference. Even if you end up winning the war, you have a whole country to build back. This has a huge impact on the economy, obviously. Your REITS portfolio kinda suffers, for instance ;)

[brooklynguy]

Not sure about what you mean here. According to the data from your other post (https://1.bp.blogspot.com/_7pcnNDkYOFE/TJLpRdCn9uI/AAAAAAAAAOI/e4WRVdW-D4U/s1600/Capture2.JPG), the worldwide SWR is about 0.25%. 400 years of income. That's pretty scary.

Just to clarify (and to elaborate on what Rebs said), the linked table shows that the "worldwide" (where the 17 countries listed in the table constitute the entire universe of countries in the "world") SWR was 0.24% (for Japan) using a 50/50 stock/bond allocation if you define SWR to mean the WR that historically resulted in success 100% of the time (commonly referred to as the "safemax" in the SWR literature).  Even in the US, the historical safemax is below 4% under the parameters most often used (in this table, it is 3.66%).  Most early-retirement-planners in this forum use a more reasonable (but still extremely conservative) threshold for what constitutes a "safe" WR, like a 95% success rate.  The table does also show the SWRs for the various countries using a 90%-success-rate threshold, and Japan is the single extreme outlier with a SWR (using that definition) below 1% -- but more importantly, even in Japan, a 4% WR succeeded most of the time (61.1% of the time).

As a French investor, I should target a 0.95% SWR.

Again, this is only true if you want to be extremely conservative and use the absolute historical worst case as your hurdle to clear (a hurdle that was set, keep in mind, based on a pretty conservative asset allocation -- as the other tables in the Pfau article make clear, stock-heavier portfolios in France had considerably higher SWRs)).

A 4% WR for 30 years would have failed most of the time. It kinda sucks.

This is not true.  The table says that a 4% WR in France worked most of the time (55.6% of the time).

[forummm]

Not sure about what you mean here. According to the data from your other post (https://1.bp.blogspot.com/_7pcnNDkYOFE/TJLpRdCn9uI/AAAAAAAAAOI/e4WRVdW-D4U/s1600/Capture2.JPG), the worldwide SWR is about 0.25%. 400 years of income. That's pretty scary.

Just to clarify (and to elaborate on what Rebs said), the linked table shows that the "worldwide" (where the 17 countries listed in the table constitute the entire universe of countries in the "world") SWR was 0.24% (for Japan) using a 50/50 stock/bond allocation if you define SWR to mean the WR that historically resulted in success 100% of the time (commonly referred to as the "safemax" in the SWR literature).  Even in the US, the historical safemax is below 4% under the parameters most often used (in this table, it is 3.66%).  Most early-retirement-planners in this forum use a more reasonable (but still extremely conservative) threshold for what constitutes a "safe" WR, like a 95% success rate.  The table does also show the SWRs for the various countries using a 90%-success-rate threshold, and Japan is the single extreme outlier with a SWR (using that definition) below 1% -- but more importantly, even in Japan, a 4% WR succeeded most of the time (61.1% of the time).

As a French investor, I should target a 0.95% SWR.

Again, this is only true if you want to be extremely conservative and use the absolute historical worst case as your hurdle to clear (a hurdle that was set, keep in mind, based on a pretty conservative asset allocation -- as the other tables in the Pfau article make clear, stock-heavier portfolios in France had considerably higher SWRs)).

A 4% WR for 30 years would have failed most of the time. It kinda sucks.

This is not true.  The table says that a 4% WR in France worked most of the time (55.6% of the time).

Just buy the whole world. Why increase your risk by concentrating in one country? Even US investors shouldn't do that.

[Monkey Uncle]

Just buy the whole world. Why increase your risk by concentrating in one country? Even US investors shouldn't do that.

Yep: http://www.schwab.com/public/schwab/nn/articles/Global-Perspective-the-Right-International-Stock-Allocation-May-Be-More-Than-You-Think?cmp=em-QYB

See the graph in item number 3.  All the upside of the U.S. market with less downside.

[k9]

As a French investor, I should target a 0.95% SWR.

Again, this is only true if you want to be extremely conservative and use the absolute historical worst case as your hurdle to clear (a hurdle that was set, keep in mind, based on a pretty conservative asset allocation -- as the other tables in the Pfau article make clear, stock-heavier portfolios in France had considerably higher SWRs)).

I know, I know. I was being rhetorical, just to underline the fact that 4% is not the worst that ever happened. I think 4% is good no matter what. If one really doesn't feel it, he should aim for 3%, no less. I mean, even if you did just as well as inflation (a 0% real return), 3% lasts 33 years and 4 months. And you don't need a very agressive allocation to do better than inflation on the long run.

And what about black swan events ? Yes, they happen (and that is what Pfau's chart shows). But you can hardly do anything about that. And that's, btw, why I agree it's silly to aim for 100% success in backtestings & simulations. 100% doesn't exist.

Oh, for the record, just to show why I think it's kinda wrong anyway to worry about that 0.95% SWR. That outlier happened because of WWII. But a French citizen retiring just after WWII was actually in a very good situation. They were the very first ones to access social security pensions (they are more substantial than in the US), and didn't have to pay for that. So, whatever got crushed during the war was somewhat given back afterwards. Pretty cool for an all-time worst.

TLDR : don't suppose 4% is ultra-safe. Don't worry about it either. Nothing is ultra-safe.