4% rule failures?

[W2R]

Quote:

Originally Posted by tiuxiu Curious if anyone has ever heard of a real life example of someone failing in FIRE when using a reasonable safe withdrawal rate, and if so how was it realized and handled? I don't mean like that garrulous legend who was in these forums before my time, I mean someone taking a 4%ish type of withdrawal from a reasonably allocated stash and realizing there was no way it could continue so they went back to work out of necessity. I don't doubt it has happened and I would think the recent economy might have produced a few. I'm trying to imagine how one realizes they are one of the lines that end up under the bar on firecalc, since it isn't always obvious and many of those lines that look sketchy end up being able to successfully support someone for over 50 years. Then taking into account future income changes like SS, the Bernicke (sp?) spending theories, etc. I'm really interested in how someone decides it is 4th and 10 and they need to punt on FIRE.

It seems to me that like many things in life, ER is a gamble. No matter what SWR you select, it is still a gamble - - because we do not know what the future will bring, or if future market performance will bear any relationship whatsoever to what we have seen in the past.

I think this is why so many here were willing to tighten our belts, or even work temporarily during the 2008-2009 market crash. Even though the portfolios of those who did so might have been OK by 2010, at the time it only seemed prudent to take a proactive stance.

[Arnie]

Quote:

Originally Posted by ERD50 It would seem that way, but it's not.

I used the "one of those lines" and "under the bar" terminology from the original post but did not mean to limit the concept to just FIRECALC. Whether it is FIRECALC or anything else, I would not rely on just one tool.

I mean that whatever logic you used to determine you could retire is just as valid (or invalid) if you rerun it years later. Think of it as though you had really not retired previously and you trying to see if you could do so now with current assumptions for portfolio size, budget, years remaining, etc.

All the various tools/techniques have assumptions and inherent weaknesses but if the results of your periodic reevaluation have gotten significantly more dismal you probably have reason to be concerned.

I suppose one could say "we've been through a real rough patch so it can only get better", but I would be uncomfortable starting a retirement underfunded on that premise - why would I be comfortable continuing one (without adjustments)?

[ERD50]

Quote:

Originally Posted by Arnie I used the "one of those lines" and "under the bar" terminology from the original post but did not mean to limit the concept to just FIRECALC. Whether it is FIRECALC or anything else, I would not rely on just one tool.

Agreed, but I think the underlying methodology makes FIRECALC one of the best tools out there. Everything else I've tried pretty much converges towards a similar ballpark.

Quote:

I mean that whatever logic you used to determine you could retire is just as valid (or invalid) if you rerun it years later. Think of it as though you had really not retired previously and you trying to see if you could do so now with current assumptions for portfolio size, budget, years remaining, etc. ... I suppose one could say "we've been through a real rough patch so it can only get better", but I would be uncomfortable starting a retirement underfunded on that premise - why would I be comfortable continuing one (without adjustments)?

Disagree (at least as far as analysis of past data, as FIRECALC does).

We didn't have two Great Depressions in a row. We didn't have two periods of high inflation like the 80's in a row. The market conditions have been cyclical.

So you are saying, "My hypothetical portfolio just barely survived the inflation of the 80's (or the GD), I need enough left over to do it all over again".

If you want enough buffer to provide some assurance that you could survive a future scenario that might be like two past bad periods strung together, then that would be a reasonable approach. But it's assuming the future will be worse than the worst of the past. And it might be reasonable to allow for that kind of buffer.

But it isn't correct to say that you can just run FIRECALC again after a bad stretch, and say that that new number is the new 'correct' number. By that logic, the you are really saying that those squiggly lines you see that dip, and then correct themselves over the long haul do not really exist - that they all ended up failures. But they didn't.

I know it's a bit tough to get your head around. I've been where you are, and it took me a while to absorb it. It seems like we are fooling ourselves, but we aren't. It's just the way the data runs. Again, future may be different, it's just a benchmark to use.


-ERD50

[Arnie]

Quote:

Originally Posted by ERD50 So you are saying, "My hypothetical portfolio just barely survived the inflation of the 80's (or the GD), I need enough left over to do it all over again".

Nope, not saying that. But let me ask you this ... Take two people, A & B. Both are the same age. Both have same budget. Both have same portfolio. Only difference is A retired five years ago and is doing a gut check to see how things are looking, and B is contemplating retirement now. Both of them see the same thing, that it does not look so hot, 2% SWR advised per whatever tool of choice. How would you advise each of them?

[73ss454]

I'm one of the ones that ran out and got a PT job in April of 09 when I thought the world was coming to an end. I'm happy to report that in Oct of this year I ended that as I'm only 5 months away from SS and our port has recovered to that of Oct of 07.

I know Dory had adjusted Firecalc for the large down draft. Does anyone know if it was again adjusted for the large come back in the last year and 1/2?

[haha]

Quote:

Originally Posted by Arnie How would you advise each of them?

"Please consult your financial advisor, legal team, accountant and/or astrologer."

Ha

[BigNick]

Quote:

Originally Posted by haha My guess is that once a portfolio that is one's only or main source of support hits $375,000, life changes, and likely forever.

I agree, in two ways:
- The one which I think you meant
- The other one, which is that if you get down to 375K and you're happy that that's enough, you've either gone lala, or you know with some certainty that you aren't going to be around long enough to run out.

In any of those cases, yep, things have changed and they ain't coming back the way they were.

[ERD50]

Quote:

Originally Posted by Arnie Nope, not saying that. But let me ask you this ... Take two people, A & B. Both are the same age. Both have same budget. Both have same portfolio. Only difference is A retired five years ago and is doing a gut check to see how things are looking, and B is contemplating retirement now. Both of them see the same thing, that it does not look so hot, 2% SWR advised per whatever tool of choice. How would you advise each of them?

I'll do one better - here's a story of THREE people

http://www.early-retirement.org/foru...ml#post1000663

You'll need to go to the beginning to catch it all.

-ERD50

[tryan]

Quote:

Raddr has been running a long-term thread on his board about the hapless Y2K ER who refuses to reduce his spending to reflect reality.

Has anyone looked at the hapless 2008 retiree?

[Arnie]

Quote:

Originally Posted by ERD50 I'll do one better - here's a story of THREE people

Thanks ERD50. I enjoyed that thread when it first appeared and just reread it. The three brothers example tells a totally different story than my A & B example though. It doesn't matter what A & B had years ago. In fact A & B lacks one of the weaknesses observed with the three brothers example - A & B are at a point now where they are both identical in their finances and are trying to decide what shape they are in going forward.

I enjoyed Ha's great answer on how to advise them. But should whoever advises them give the two of them different advice? I think not. A's likelihood of success in a continued retirement is exactly the same as B's in starting retirement, assuming they do everything the same.

It does not matter that A's original calculations supported 4% and that A went through a bad spot and (historically) should recover. Current projections inform both A & B that they are only safe with 2%. If there is any valid argument that A can continue with 4%, then the same argument says B can start with 4%. But I would not be any more comfortable with that than I would with the approach of the "hapless Y2K ER".

By the way, person A should not have waited 5 years to re-evaluate.

If part of the original question, and I may have missed it, was how does one tell when they are getting in trouble - and the answer is not, at least in part, to rerun the numbers through their original tool(s), then I think there is a problem with the tool or, perhaps more likely, the understanding and application thereof.

Again, I was not and am not trying to make this about the excellent FIRECALC. I agree with ERD50's comment that "it isn't correct to say that you can just run FIRECALC again after a bad stretch, and say that that new number is the new 'correct' number." But I cannot get my head around it being any less 'correct' for A vs. B in this example. If there is an issue it is believing that any tool can give you a 'correct' number in the sense of giving you something you can stick blindly to and be bulletproof.

Thanks folks for taking the time with me. I am, after all, confused by dryer sheets. I doubt we're disagreeing much at this point and I may just be going on because I enjoy the sound of my own voice.

-Arnie

[haha]

Quote:

Originally Posted by Arnie But should whoever advises them give the two of them different advice? I think not. A's likelihood of success in a continued retirement is exactly the same as B's in starting retirement, assuming they do everything the same.

To give you a more serious answer, I agree with your conclusion 100%. Early on I would read the explanations of how the history of one’s getting to the present effects his chances going forward- given same size port, same WR, and same required duration.

I cannot see how it possibly could matter

Because the markets tend to be mean reverting over long periods of time, (as to value ratios, not level) I believe that the market has some sort of memory. Its history does matter, as opposed to the model of it being a Monte Carlo generator. But the market has no idea when you or I dropped in, and would be indifferent to this information if it had it.


Ha

[cho oyu]

Quote:

Originally Posted by Arnie It does not matter that A's original calculations supported 4% and that A went through a bad spot and (historically) should recover. Current projections inform both A & B that they are only safe with 2%. If there is any valid argument that A can continue with 4%, then the same argument says B can start with 4%. But I would not be any more comfortable with that than I would with the approach of the "hapless Y2K ER". -Arnie

There's a difference in assumptions in the two scenarios:

Scenario 1: planning for retirement only looking with a forward model(i.e. your person B)

Scenario 2: planning for retirement by including both recent past and modeled future results assuming temporal autocorrelation (your person A); the answers to this simulation are merely a subset of answers to scenario 1

Therefore, scenario 1 is a more robust answer to a monetary survival question.

As to whether it is better? That depends on how well one thinks that the future can be predicted based on the past. Ultimately, which imperfect prediction method allows you to peacefully sleep each night?

[FIRE'd@51]

Quote:

Originally Posted by Arnie It does not matter that A's original calculations supported 4% and that A went through a bad spot and (historically) should recover. Current projections inform both A & B that they are only safe with 2%. If there is any valid argument that A can continue with 4%, then the same argument says B can start with 4%. But I would not be any more comfortable with that than I would with the approach of the "hapless Y2K ER".

While I see your point, it is inconsistent with a FIRECALC probability prediction because of the 5-year time difference.

At this point A has 25 years to go and B has 30 in terms of measuring the success of a FIRECALC prediction. Certainly FIRECALC would say that A has a higher probably of success for 25 years than B for 30 at the same withdrawal rate. By saying they both now have 30-year horizons, you are implicitly saying that A should have based his initial withdrawal rate on a 35-year time span.

[ERD50]

Quote:

Originally Posted by Arnie Thanks folks for taking the time with me. I am, after all, confused by dryer sheets. I doubt we're disagreeing much at this point and I may just be going on because I enjoy the sound of my own voice. -Arnie

Well for me, explaining something helps me to make sure I understand it, so I get something out of it too.

So I see where you are coming from, and I agree that if at a specific point in time, there is no difference between them in their SWR if they have the same portfolio $ and the same # of years going forward (edit - I see FIRE'd@51 touched on this). Consistent with the Three Brother's story, it is how they got there - but we can ignore that for now.

But consider this - FIRECALC doesn't make any assumption of what the past was when you jump in. The failures reported have you starting off at the worst possible time (that's why they fail!). But, if you started with $1M, and 5 years later you have half that, you are already five years into a period of very bad times. FIRECALC is now going to put you back at the beginning of a bad period, rather than having that $1M 5 years in. So it extends your bad streak by 5 years.

So....... in that case, FIRECALC is being over-conservative, and actually BOTH of the people could get by with a higher SWR at that point. They still have $500K after 5 years of bad times. IOW, bad periods are pretty much defined by coming off a high period. But these two people came off a lower period. Yes, statistically they can take a higher SWR than normal with the same success rate.

Now whether you are comfortable with taking that higher rate or not is a personal gamble. But that is (to the best of knowledge) how to interpret what FIRECALC does with that scenario.

-ERD50

[Moemg]

Quote:

Originally Posted by Arnie . By the way, person A should not have waited 5 years to re-evaluate. . -Arnie

Absolutely ! I retired in 2007 just when the s--t was hitting the fan and had I continued along the 4% plus inflation I would have done serious damage to my portfolio .

[Arnie]

Quote:

Originally Posted by ERD50 Now whether you are comfortable with taking that higher rate or not is a personal gamble. But that is (to the best of knowledge) how to interpret what FIRECALC does with that scenario.

I understand and agree with your FIRECALC explanation. I guess I can blame my lack of comfort with using higher WR's in the A & B example on recency bias. If I was up against that scenario in reality I might want to "average" the FIRECALC results with those from other tools.

[FUEGO]

I have thought a lot about this issue over the last couple years. Soon I plan to post a new thread with some recent research I have done. Until then, I'll add to the discussion by linking a previous bit of research I did:

http://www.early-retirement.org/foru...ess-35074.html

In that thread, I presented some research I did. If you want to skip to my conclusions, here they are: Go back to work if at 10 years you have less than 60% of your initial portfolio value or if at 15 years you have less than 50% of your portfolio value. This would greatly reduce your odds of running out of money, but not send you back to work prematurely.

[Katsmeow]

Quote:

Originally Posted by ERD50 Disagree (at least as far as analysis of past data, as FIRECALC does). We didn't have two Great Depressions in a row. We didn't have two periods of high inflation like the 80's in a row. The market conditions have been cyclical. So you are saying, "My hypothetical portfolio just barely survived the inflation of the 80's (or the GD), I need enough left over to do it all over again". If you want enough buffer to provide some assurance that you could survive a future scenario that might be like two past bad periods strung together, then that would be a reasonable approach. But it's assuming the future will be worse than the worst of the past. And it might be reasonable to allow for that kind of buffer. But it isn't correct to say that you can just run FIRECALC again after a bad stretch, and say that that new number is the new 'correct' number. By that logic, the you are really saying that those squiggly lines you see that dip, and then correct themselves over the long haul do not really exist - that they all ended up failures. But they didn't. -ERD50

I don't think so. It is sort of like the guy who rolled heads and the question is how likely is it that his next roll would be heads. Some would say not because it is less likely someone will roll two heads than one head and one tail. But they are comparing apples and oranges. When no rolls have been made it is less likely that someone will roll two heads. But once one roll has been made (and been heads) that person still has a 50-50 chance of rolling heads.

So, yes, it is very unlikely someone will have two sets of extremely bad market returns close in proximity. However, if A retires, has the bad market returns, then that has already happened to him. He does not have less likelihood of bad future market returns than B who hasn't retired. A's overall probably of success may be higher because he might have less years of retirement to plan for. But future economic conditions will be the same for both him and B.

[ERD50]

Quote:

Originally Posted by Katsmeow I don't think so. It is sort of like the guy who rolled heads and the question is how likely is it that his next roll would be heads. Some would say not because it is less likely someone will roll two heads than one head and one tail. But they are comparing apples and oranges. When no rolls have been made it is less likely that someone will roll two heads. But once one roll has been made (and been heads) that person still has a 50-50 chance of rolling heads. So, yes, it is very unlikely someone will have two sets of extremely bad market returns close in proximity. However, if A retires, has the bad market returns, then that has already happened to him. He does not have less likelihood of bad future market returns than B who hasn't retired. A's overall probably of success may be higher because he might have less years of retirement to plan for. But future economic conditions will be the same for both him and B.

But FIRECALC isn't MonteCarlo. There is no rolling of dice, it is just reporting on the past scenarios. It has nothing to do with the likelihood of it happening. It either happened or it didn't. If it happened it's considered.

And the fail scenarios you get don't start 5 years into the bad period, that would shorten the bad period by 5 years and that wouldn't be the worst of the worst. And they don't start at the beginning of the second stage of a double-whammy bad market, the worst would be both stacked together (if/when that happens). So for that case, we are now talking about someone surviving a bad market and then throwing two more sequential bad markets on top of that with the FIRECALC failure. That no longer reflects history, and is outside the realm of what FIRECALC does.

And this assumes the future is no worse than the past, but that is all the tool can do. You can adjust for that if you wish (not a bad idea at all), but don't double-adjust w/o considering what it is really telling you.

-ERD50

[Nords]

Quote:

Originally Posted by 73ss454 I know Dory had adjusted Firecalc for the large down draft. Does anyone know if it was again adjusted for the large come back in the last year and 1/2?

Dory not only took care of all the "routine" maintenance and bug-fixing and upgrades, as well as educating those who thought a feature was actually a bug, but every year he also downloaded Schiller's databases of asset-class performance. I remember there used to be a lag of months until the previous year's data was available for Dory to download.

Either FIRECalc or E-R.org was a significant commitment, especially for a guy who spent most of his time on the Intracoastal Waterway. Together they amount to nearly a full-time job.