Safe Savings Rates

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
Majormajor78
Posts: 910
Joined: Mon Jan 31, 2011 8:13 pm

Safe Savings Rates

Post by Majormajor78 »

Here's an article I came across that does an interesting analysis on how much we need to save to ensure that we'll be able to maintain a safe withdrawl rate. Fluctuates quite a bit under differnt market conditions as you would expect but I think it gives an interesting double-check to see if we are hitting our goals.

http://wpfau.blogspot.com/2011/02/safe- ... ch-to.html
http://www.fpanet.org/journal/CurrentIs ... ectations/
Last edited by Majormajor78 on Tue Feb 07, 2012 1:45 pm, edited 2 times in total.
wts
Posts: 165
Joined: Tue Jul 13, 2010 12:27 am

Post by wts »

The black LMSR curve is the main contribution of this paper. In the context of Bengen’s original study, the maximum value of the LMSR curve (which is 16.62 percent in 1918) becomes the SAFEMIN savings rate from a lifetime perspective that corresponds to Bengen’s SAFEMAX withdrawal rate.


The LMSR curve suggests two things to me.
1. 16.62% is a high rate to sustain consistently over 30 years. No wonder theres a crisis looming.
2. If it can be maintained then an individual would have always saved enough for retirement regardless of sequence of returns.
User avatar
CaliJim
Posts: 3050
Joined: Sun Feb 28, 2010 7:47 pm
Location: California, near the beach

Post by CaliJim »

..
Last edited by CaliJim on Wed May 11, 2011 12:11 am, edited 1 time in total.
-calijim- | | For more info, click this Wiki
Dandy
Posts: 6701
Joined: Sun Apr 25, 2010 7:42 pm

Post by Dandy »

Getting to a high savings rate is the trick. People hear you must save 10% or more and they feel they can't do it.

What worked with me was when I started work an "old timer" said every time you get a raise increase your savings -- you'll never miss what you never had.

That was a great way to live below your means from the start.
User avatar
ofcmetz
Posts: 2465
Joined: Tue Feb 08, 2011 7:09 pm
Location: Louisiana

Post by ofcmetz »

Interesting article, thanks. Like the previous post says, if you can learn to live on 80% of 50K income then one day after raises and promotions you'll be living on 80% of 100K income and still be enjoying an increased standard of living.

I set my retirement savings rate at 15% and my total savings rate at 20% about 5 years ago after reading The Millionare Next Door.
Never underestimate the power of the force of low cost index funds.
User avatar
Topic Author
Majormajor78
Posts: 910
Joined: Mon Jan 31, 2011 8:13 pm

Post by Majormajor78 »

wts wrote:
The black LMSR curve is the main contribution of this paper. In the context of Bengen’s original study, the maximum value of the LMSR curve (which is 16.62 percent in 1918) becomes the SAFEMIN savings rate from a lifetime perspective that corresponds to Bengen’s SAFEMAX withdrawal rate.


The LMSR curve suggests two things to me.
1. 16.62% is a high rate to sustain consistently over 30 years. No wonder theres a crisis looming.
2. If it can be maintained then an individual would have always saved enough for retirement regardless of sequence of returns.


That 16.62% number might actually be a little low since the article purposes a withdrawl rate of 50% of ending salary. I suppose if the individual is getting Social Security or another source of income this would dovetail nicely with the estimated 80% final salary needed to live an adequate retirement. Want to travel and/or pick up some hobbies and you need to step that rate up to 20% at least.
sport
Posts: 12084
Joined: Tue Feb 27, 2007 2:26 pm
Location: Cleveland, OH

Post by sport »

Majormajor78 wrote:I suppose if the individual is getting Social Security or another source of income this would dovetail nicely with the estimated 80% final salary needed to live an adequate retirement.
When you retire, you don't need some percentage of your final salary. You need 100% of your post-retirement budget. This may be the same as 80% of your final salary, or it may be a lot less (or possibly more).

Jeff
User avatar
ofcmetz
Posts: 2465
Joined: Tue Feb 08, 2011 7:09 pm
Location: Louisiana

Post by ofcmetz »

jsl11 wrote:
Majormajor78 wrote:I suppose if the individual is getting Social Security or another source of income this would dovetail nicely with the estimated 80% final salary needed to live an adequate retirement.
When you retire, you don't need some percentage of your final salary. You need 100% of your post-retirement budget. This may be the same as 80% of your final salary, or it may be a lot less (or possibly more).

Jeff
Nice way to put it dude. I totally agree, but haven't heard it stated that way before.
Never underestimate the power of the force of low cost index funds.
KlangFool
Posts: 31426
Joined: Sat Oct 11, 2008 12:35 pm

Re: Safe Savings Rates

Post by KlangFool »

Majormajor78 wrote:Here's an article I came across that does an interesting analysis on how much we need to save to ensure that we'll be able to maintain a safe withdrawl rate. Fluctuates quite a bit under differnt market conditions as you would expect but I think it gives an interesting double-check to see if we are hitting our goals.

http://wpfau.blogspot.com/2011/02/safe- ... ch-to.html
Majormajor78,

I am going throw a monkey wrench to your whole calculation.

Gross saving rate = Income - living expense - Taxes..

So, as the saving rate went up, a person is living with a life style of lower expense.

Let's assume 20% gross saving rate and 25% gross taxes, this person is living on 55% of his gross income. Every year, this person will save 20% / 55% = 36% of his annual expense. Assuming 4% SWR and 0% real return, this person needs 68.75 years of working.

On the other hand, if a person saves 50% of his gross income with 25% gross taxes, this person is living on 25% of his gross income. Every year, this person saves 50%/25% = 2 years of annual expense. Assuming 4% SWR and 0% real return, this person needs 12.5 years of working.

A) Gross saving rate and expense is linked. And, it is an inversely proportional relationship.

B) How realistic is it for a person to continue working for 68 years??

What is the assumption of REAL RETURN for the 60/40 portfolio?? What am I missing here??

KlangFool
User avatar
Topic Author
Majormajor78
Posts: 910
Joined: Mon Jan 31, 2011 8:13 pm

Post by Majormajor78 »

KlangFool,

Unfortunately the blog I linked to is not the full article but only a summary. The link to the full article is labeled "(link coming as soon as possible)." Make of that what you will.

As for your math, why are you are assuming a 0% real return? The summary article claims to use historical data of real returns for stocks and treasuries over differnt time periods. "Her asset allocation during the entire 60-year period is 60/40 for stocks and bills. Data is from Robert Shiller’s webpage for the S&P 500 and Treasury bills."

I'm not really endorsing the article wholeheartedly. Obvious weakness in not being able to see sources for the historical data. Hopefully the full article will be more illuminating. Instead I found it to be a conceptually intriguing perspective on necesarry savings rates. Everybody talks about how we need to save in order to hit "our number" to allow us to retire. Differnt authors and advisors throw around lots of saving "rules of thumb" for good measure like 10% but 20+% is much better. The problem with this is we are not truly able to tell what "our number" is going to be 20-30 years in the future. Instead the article attempts to mathmatically define a minimum safe savings rate by examining historical market returns and inflation and seeing how somebody would have faired if they retired at any time and in any environment including the great depression. I think this hints at giving us a persective on hitting a solid and hopefully practicle savings rate today that will increase the odds of us being able to support ourselves in retirement. Perhaps some would argue that this is opposite sides of the same coin but I think it might be more actionable to strike for (and hopefully surpase) a solid number that we can calculate today as opposed to attempting to hit a magic number 25 years from now that may or may not be realistic. We can make any number of educated guesses about the future but all we can know for sure is the present and this is where we need to execute our plans.

You are right that there are obvious weaknesses in the summary but I'm personally looking forward to studying the more detailed full article.

Scott
User avatar
DRiP Guy
Posts: 2241
Joined: Tue Feb 20, 2007 3:54 pm

Post by DRiP Guy »

jsl11 wrote:
Majormajor78 wrote:I suppose if the individual is getting Social Security or another source of income this would dovetail nicely with the estimated 80% final salary needed to live an adequate retirement.
When you retire, you don't need some percentage of your final salary. You need 100% of your post-retirement budget. This may be the same as 80% of your final salary, or it may be a lot less (or possibly more).

Jeff
That's an excellent point, and one that needs to be addressed by every individual prior to entering retirement, or before using a tool, such as might be developed from such research. But IMHO, as long as the author properly delineates that presumption in the article or paper, as well as other major assumptions/simplifications used just to model the general approach, I don't feel it detracts from the significance of the work at all.
Last edited by DRiP Guy on Sun Feb 13, 2011 8:51 pm, edited 1 time in total.
User avatar
DRiP Guy
Posts: 2241
Joined: Tue Feb 20, 2007 3:54 pm

Post by DRiP Guy »

Majormajor78 wrote:KlangFool,

Unfortunately the blog I linked to is not the full article but only a summary. The link to the full article is labeled "(link coming as soon as possible)." Make of that what you will.

...

You are right that there are obvious weaknesses in the summary but I'm personally looking forward to studying the more detailed full article.

Scott
I could be wrong, but I think that's the very reason there is no link, or no publishing yet, to generate and respond to feedback regarding issues, errors, or omissions that could be remedied before further propagating the work.

I have personally found Wade to be very receptive and desirous of feedback, so I'd encourage anyone with what they consider significant feedback to contact him directly.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Hi, I'm the author of the article being discussed in this thread. Majormajor78, thank you for linking to it. I was on the fence about whether I should provide a link, as I feel I may have been wearing out my welcome here after my recent foray into valuations-based tactical asset allocation. As well, as I will discuss below, though my original intention in looking at this topic was to procrastinate from thinking any more about valuations-based tactical asset allocation, this article can also plausibly be interpreted as another long-term market timing scheme. That would be my strike 2, but I will talk about that below.

The full paper can be downloaded here:

http://ideas.repec.org/p/pra/mprapa/28796.html

I'd link to respond to some various comments and questions that have been raised.

CaliJim wrote:Interesting stuff and very useful.

A target nest egg dollar amount lets people go into denial about saving for retirement. Lavish spending is justified by thinking "I can make it up by saving more next year."

The approach outlined in the referenced article creates an annual savings goal, putting adherents on the hook to save for retirement each and every year.

Good stuff.
Thank you. Economists generally don't support fixed savings rates, as the goal should be to smooth consumption over one's lifetime. So my idea about fixed savings rates was that this is just a first step to explain my framework and it could be modified later. But I think you are providing an important contribution of your own here: based on behavioral economics, some people will be better off by forgetting about consumption smoothing if they are otherwise better able to commit themselves to maintaining their fixed savings rate throughout their life. That's interesting!

Just to further explain the usual economists view, Goofyhoofy at the Motley Fool board had this criticism, which I am saying I agree with:
I found it interesting but superficial. The SWR doesn't account for a lot of life changes, but then (except for the last couple years, where medical might take a big chunk) a retiree's life is generally a fairly smooth ride.

On the savings side, that happens very differently: people (generally) start off at very low salaries which increase over their 40 year career. They have kids, who take a lot of money just for growing, clothes, college, and so on. So the "savings rate" could be vastly different for a 28 year old than a 48 year old, and is much more highly dependent on "life circumstances" than a simple straight line formula or calculation might suggest. I'm not saying it doesn't pay to "start early", just that employing the marginal utility of money concept, you probably have a lot more to save later than you could ever hope to earlier.
An important point I cannot emphasize enough is that while the 4-percent rule about withdrawals is universal, there is no universal safe savings rate in my framework. Everyone has different life circumstances, and so everyone will have to determine their own safe savings rate. Should this concept turn out to be popular, I suppose some software or excel spreadsheets can be created to help do this. With this framework, a young person has to sit down and make a tentative plan for their whole life: their expected lifetime income trajectory, their expected consumption needs over their working years (such as raising kids and sending them to college, etc.), and also their expected retirement expenditures. This is not easy to do, but I've done this for myself and think it is possible. With this information, a person can formulate a "safe savings rate" based on past historical data. That savings rate doesn't have to be fixed, but so far I'm only considering the case of fixed savings rates. My framework doesn't have much to say for people who are already near retirement or who are already retired. It will be more focused at people who are early in their careers.
Majormajor78 wrote:That 16.62% number might actually be a little low since the article purposes a withdrawl rate of 50% of ending salary. I suppose if the individual is getting Social Security or another source of income this would dovetail nicely with the estimated 80% final salary needed to live an adequate retirement. Want to travel and/or pick up some hobbies and you need to step that rate up to 20% at least.
The key is that a person must first determine for themselves what their desired retirement expenditures will be. It can be more or less. But as a baseline, let me explain why I think 50% is not so low:

1. After retiring, you are not saving anymore. If you used a 16.62% savings rate, then a 50% replacement rate actually represents

50 / (100 - 16.62) = 60% of non-saved pre-retirement income

2. Also, indeed, I am assuming that people also get Social Security. So your Social Security benefit will be added on to the top of that 60%. In that regard, you are getting a pretty decent replacement rate in my baseline example.

Also, Table 1 of my paper shows some various modifications from the baseline. One is assuming a 70% replacement rate.

I fully agree with Jeff's assessment, and I think I did express that inside the actual paper.
KlangFool wrote:What is the assumption of REAL RETURN for the 60/40 portfolio?? What am I missing here??
It's based on rolling historical periods. There is no over-arching assumption.

Oh, it is nice to see that DRiP Guy has joined the discussion. He's my #1 critic :D I mean sincerely that it is nice he's here. So I have one more important issue to discuss which I promised at the beginning of this post. Why don't I go ahead and hit "Submit" now and that come back later with the issue of whether this is just a long-term market timing scheme.
Last edited by wade on Thu Feb 17, 2011 5:22 am, edited 1 time in total.
User avatar
DRiP Guy
Posts: 2241
Joined: Tue Feb 20, 2007 3:54 pm

Post by DRiP Guy »

wade wrote:I was on the fence about whether I should provide a link, as I feel I may have been wearing out my welcome here after my recent foray into valuations-based tactical asset allocation.

[snip valuable info]

.... this article can also plausibly be interpreted as another long-term market timing scheme.

...Why don't I go ahead and hit "Submit" now and that come back later with the issue of whether this is just a long-term market timing scheme.
Wade:

Just to address the 'wearing out the welcome' part -- I think we have had some pretty contrary-thinking characters around who added a lot of life and interesting dialog to the board over the time since it's inception, and they hopefully illuminated and even got some folks to open their personal thinking up, although it is true it was also sometimes along with their blood pressure!

:lol:

That said, however, this is very much a 'board with a mission,' and so while some of the staunchest supporters of other methods have done valuable service in contributing here, it is true (at least to my own observations) that even the best of them in reasoning skill, and the most congenial, eventually tired of the struggle of swimming upstream against a powerful current -- the majority of the board are indeed adherents to the philosophy of Bogle, at least as they interpret it.

I suspect this might be your fate as well, at some point, but I'd personally like to invite you to stay and liven the dialog, since you seem to me to be a genuine investigator of various approaches to addressing the financial dilemmas we all must deal with, and are here to discuss.

Even if what you propose ultimately really does fall under the guise of 'market timing', it would appear a solid minority of almost 10% here feel that most are already timing anyhow (see link)... so for those at least it might prove invaluable to consider.

As for me, I have already made my own proclivities abundantly plain on numerous occasions, so I need not establish that I am a bit of a stick in the mud as to my own willingness to engage in doing any more guessing about Mr. Market than absolutely required to meet my personal minimum goals, so we needn't rehash all that -- I'd just say that as for me, I feel if we can enjoy the fruits of "Market Timer" and Munchkin Man, and many other free-style types, then we likely have room for a Professor of Economics dedicated to exploring how to fund retirements.

My own recent revelation is documented at the link here -- I guess if you are willing to constrain the definition sufficiently (and that's a whole lot of constraint, IMHO!) then anyone who doesn't literally buy just once and the hold forever without re-balancing or other change, is committing the sin of market timing anyway -- it's just a matter of degree...
http://www.bogleheads.org/forum/viewtop ... highlight=

:wink:
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Is this paper just another long-term market timing scheme?

I would like to say no, but I think a plausible case can be made for both sides of the issue. So a Point / Counterpoint:

This paper is not a market-timing scheme

This paper just applies the methodology designed by William Bengen (who explored safe withdrawal rates) to the entire lifetime of working and saving. Over rolling periods that include both a working phase and a retirement phase, I calculate the savings rate needed in the working phase to finance one's desired retirement expenditures. Then I find, what was the highest savings rate required in history to finance one's retirement expenditures (for the baseline individual I use), and I call that the "safe savings rate," which basically corresponds to the "safe withdrawal rate" in terms of caveats, as described in previous research. Valuations may play a role here, but I wish I could just say that I don't care about the underlying mechanism that explains the results. The safe savings rate is what it is.

This paper is a market-timing scheme

But I can understand the other side of the issue as well, as I heard explained to me by both someone who strongly opposes valuation-based investing, and someone who strongly supports valuation-based investing. The idea can be summarized from an anonymous comment on my blog:
Your point appears to be that year-to-year investment returns are not independent.
Well, yes. I'm saying that the problem with traditional retirement planning advice is that it treats the 30-years before retirement as being independent from the 30-years after retirement, when in fact, I do believe there is a link. That is the basis for why I wrote:

"But the savings plan should be adhered to regardless of whether it seems one is accumulating either more or less wealth than is needed based on traditional criteria."


Image

Image

I don't want to re-explain everything here, so please see the blog post or paper for background details. These figures show how the MWR responds to past market returns. Figure 2 directly shows that from year to year, high asset returns will push down the MWR for a new retiree in the subsequent year, while low asset returns push up the MWR for a new retiree in the subsequent year. That right there I think provides a resolution to the safe withdrawal rate paradox described in the Boglehead's wiki on safe withdrawal rates.

Figure 4 then shows a tendency that people who experienced bad luck during their working years tend to get redemption after retirement. The x-axis shows the fixed savings rate over 30-years that someone needs to accumulate 12.5 x final salary at retirement (This is how I portray traditional retirement planning advice: a 50 percent replacement rate / a 4 percent safe withdrawal rate = 12.5 time final salary is needed). The y-axis shows the subsequent 30-year MWR during retirement. People who were unlucky when they worked and had to use a higher savings rate to meet their wealth accumulation goal, then tended to enjoy the opportunity to have used a higher MWR as well. Likewise, people who were lucky when they worked, and could use a lower savings rate also tended to find out that their MWRs were on the low side as well.

As such, I am saying that those who "saved enough" which I define not by their wealth accumulation, but rather by their savings rate, should turn out to be okay, barring the caveat that their lifespan turns out to provide a new low that has not yet been experienced in the historical data. I'm also saying that their actual withdrawal rate doesn't matter. It could be low or high, depending on when key parts of their career took place during a bull or bear market.

I am basically exploiting this information in my Figure 5. Does it make me a market-timer? Maybe - ? See above for why I would prefer not to think so.

Image

But this also leads to a new take on Figure 5. If you don't believe that the relationship in Figure 4 will continue to hold in the future, and if you want to treat retirement as independent from work, and therefore you want to follow "traditional retirement planning advice" then you need to make your savings decisions based on the savings rates shown in the blue curve.

But if you trust the relationship in Figure 4 as being something meaningful, then you can go ahead and use the black curve of Figure 5 to make your savings rates decisions. I view that as my main point.


I forgot to mention in my last post, but there was also another important comment on my blog:
Savings rate can be considered an antithetic variable with smaller variance than withdrawal rate or wealth accumulation. However, I think you need to consider how a responsible human would behave. Would it be responsible for a 1921 to start withdrawing >9% and hope for good investment returns so they don't run out of money half way through retirement? Even if a study says they could have, would they really have the guts to do it? Maybe if they have a pension and/or government retirement benefits that would provide for their minimal retirement needs, this strategy would be appropriate for a portion of their investment that provide some bonus income for luxuries. But I don't see it being a viable strategy for someone to use for their needs.
This is fair enough. It would have been rather scary for that 1921 retiree to enter retirement with wealth of only 5.52 times final salary. There is no question. But the point I am trying to make is that things would have worked out okay for this guy after all. He's not even the one who experienced the worst outcome in history.

Here is a bonus figure that I didn't put in the paper which shows the evolution of accumulated wealth before and after retirement for the 1921 and 1966 retirees (1966 had the lowest MWR in history) when using a 16.62% savings rate:

Image

After saving at 16.62%, the 1921 retiree had to use a 9.06 withdrawal rate to obtain her desired retirement spending level. But even using a such a high withdrawal rate, she still had over 2 multiples of final salary left as wealth at the end of 30 years.

I tend to be a rather risk-averse person myself, but I guess I am trying to say that it is okay to sometimes go ahead and take the plunge into retirement if that is what you want to do. If you saved with a responsible savings rate, you should be okay regardless of your wealth accumulation. The future might throw a wrench into the works, such as if the U.S. begins a process of terminal decline, but otherwise you should be okay.
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Post by magellan »

IMO, this is a very important contribution and a great insight.

I've long wanted to figure out a way to "exploit" the fact that the worst-case max withdrawal rates in history all came after very steep market run ups. IIRC, the historical data supports a SWR over 5%, if you subtract out portfolio gains from the last 2 years before retirement. Wade's bonus figure above, showing annual portfolio value for the 1966 retiree really drives this home. They got the worst historical withdrawal rate based on their retirement portfolio value, but look at how much that value exploded in the last years before retirement. Still, I was never really comfortable with leveraging this because it reeks of data mining and is based on only a handful of data points that aren't even independent.

Wade's insight to trace this back to the savings rate seems like a good one. Let's face it, it just seems wrong that the traditional SWR methodology can result in dramatic differences in withdrawal amounts for two retirees that had identical savings habits, but retired just a year or two apart. That just doesn't seem like it could be right.

Anyhow, as far as Wade wearing out his welcome - I say hell no. This is exactly the kind of out-of-the-box thinking that makes this forum so interesting and thought provoking. Please keep it up.

Jim
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Post by magellan »

btw - Reuters blogger Felix Salmon has some additional comments on wade's paper.
Felix Salmon wrote:But the bigger message certainly resonates with me: spend less effort on trying to boost your annual returns, when you have very little reason to believe in your alpha-generation abilities, and spend more effort on maximizing your savings every year.

Investing can be exciting, especially when it’s done wrong. You follow the markets rising and falling, you obsess about your retirement-fund balance, you rotate out of this and into that, you read books and magazines and blogs to try to learn more about what to do. You might even, in a moment of weakness, find yourself watching CNBC. Budgeting, by contrast, is like going on a diet: it’s a drag, and it’s hard to get any pleasure or excitement out of it. But the latter is much more likely to get you well-set in retirement than the former.
Sounds rather bogleheadish to me...

Jim
User avatar
DRiP Guy
Posts: 2241
Joined: Tue Feb 20, 2007 3:54 pm

Post by DRiP Guy »

magellan wrote:Let's face it, it just seems wrong that the traditional SWR methodology can result in dramatic differences in withdrawal amounts for two retirees that had identical savings habits, but retired just a year or two apart. That just doesn't seem like it could be right.

Jim
(chuckle) Respectfully, it is wrong -- it's wrong because the 'traditional' methodologies don't result in dramatic differences for retirees that retire just a couple of years apart. The ones I am familiar with would tag 'em both at about.... oh, let's say.... 4%.

:wink:
grberry
Posts: 234
Joined: Fri Jul 20, 2007 1:16 pm
Location: Boston, MA

Post by grberry »

A few thoughts.

1) I note that one of the assumptions in the paper is that "Portfolio administrative and planning fees are not charged" (page 4). A unrealistic assumption, as there will be mutual fund fees, but one that Bogleheads in general will come closest to. Anyone building a spreadsheet tool should include an input for the portfolio fee level.

2) The paper mostly uses a 60/40 asset allocation. I see that table 1 (page 15) also shows 40/60 and 80/20 fixed allocations. In the long run, alternative allocations should also be considered, including those using international equity and allocation rules with varying allocations over time (e.g. age in bonds). Again, an optimized spreadsheet tool would be open to such allocations, although I don't know if alternative slices have as much history available.

3) The example in the paper is that of an individual whose real income is constant for the last 30 years of working. This should eventually be extended to include people whose real income grows over time. A good spreadsheet would also have an input parameter here.

Ultimately, the optimized version of this would be based on a massive Monte Carlo simulation of both 1) future investment returns and 2) future income levels, as both are likely to vary in unpredictable ways. I'm not sure that we have as a society any data on which to simulate future income paths, but at the least the chance of a period of unemployment or disability ought to be included.

Finally, I do think that this is a great conceptual expansion of planning to limit risk. I hope that you will get the paper, or an expanded version of it, published in a journal.
User avatar
jh
Posts: 1998
Joined: Mon May 14, 2007 11:36 am
Location: USA

Post by jh »

...
Last edited by jh on Sat May 07, 2011 5:40 pm, edited 1 time in total.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

DRiP Guy wrote:
magellan wrote:Let's face it, it just seems wrong that the traditional SWR methodology can result in dramatic differences in withdrawal amounts for two retirees that had identical savings habits, but retired just a year or two apart. That just doesn't seem like it could be right.

Jim
(chuckle) Respectfully, it is wrong -- it's wrong because the 'traditional' methodologies don't result in dramatic differences for retirees that retire just a couple of years apart. The ones I am familiar with would tag 'em both at about.... oh, let's say.... 4%.

:wink:
Jim: Thank you so much for your comments!

DRiP Guy, I'm not understanding your response to Jim. You did catch that he said "withdrawal amount" rather than "withdrawal rate," right? My understanding is that Jim is referring to the safe withdrawal rate paradox.
grberry wrote:1) I note that one of the assumptions in the paper is that "Portfolio administrative and planning fees are not charged" (page 4). A unrealistic assumption, as there will be mutual fund fees, but one that Bogleheads in general will come closest to. Anyone building a spreadsheet tool should include an input for the portfolio fee level.
I agree with this. My program has administrative fees built in so that I can add them very easily, but I left them out to be consistent and comparable with most of the existing research. I do find the general lack of fees in existing research to be rather vexing though.
grberry wrote:2) The paper mostly uses a 60/40 asset allocation. I see that table 1 (page 15) also shows 40/60 and 80/20 fixed allocations. In the long run, alternative allocations should also be considered, including those using international equity and allocation rules with varying allocations over time (e.g. age in bonds). Again, an optimized spreadsheet tool would be open to such allocations, although I don't know if alternative slices have as much history available.
I will definitely incorporate age-based allocation rules, and I would like to consider international equities. I have data going back to 1900 rather than 1871 for them.
grberry wrote:3) The example in the paper is that of an individual whose real income is constant for the last 30 years of working. This should eventually be extended to include people whose real income grows over time. A good spreadsheet would also have an input parameter here.
Yes, definitely. My assumption of constant real wages will generally lower the necessary savings rate for two reasons:

1. most people have lower incomes when they are young, which means that their savings when they are young will have less time to compound with growth.

2. Most people will have higher incomes near retirement, so if their desired replacement rate is a percentage of final salary, this is a separate reason that will require more savings.

It turns out that I wrote a paper a few years ago that is relevant to this, as I compare the work histories of actual workers to hypothetical workers such as the one I use in this paper. In my current paper, it does help to make more realistic that I assume the person only saves for 30 years, rather than the more commonly used 40 years case. It is in the somewhat obscure Journal of Income Distribution, but I have a working paper:

"Assessing the Applicability of Hypothetical Workers for Defined-Contribution Pensions"
http://ideas.repec.org/p/ngi/dpaper/07-11.html
grberry wrote:Ultimately, the optimized version of this would be based on a massive Monte Carlo simulation of both 1) future investment returns and 2) future income levels, as both are likely to vary in unpredictable ways. I'm not sure that we have as a society any data on which to simulate future income paths, but at the least the chance of a period of unemployment or disability ought to be included.
About Monte Carlo for future investment returns: I wouldn't trust it much for this case. I use Monte Carlo a lot, but I always assume that asset returns are not related over time, as I had never really thought about this much until recently. Thus, there would be no link between the pre-retirement and post-retirement period. New Monte Carlo methods could be created to allow valuations to be important (at least, new for me, I think other people have attempted this already), but at the end of the day their importance will be dictated by the importance you give them in your underlying assumptions. I would prefer to stick with historical data. But I will at least try this out on the historical data for other countries, as that might be useful.

About Monte Carlo for income paths, you are giving me flashbacks to the early stages of when I wrote my dissertation, as I envisioned such a thing with transition probabilities in and out of disability and unemployment. In the end, I abandoned all that. But I do think that trying these methods on the income paths of past actual workers could be useful. In that regard, the Panel Study of Income Dynamics has now been around long enough to give us full career paths that are not topcoded by the maximum taxable Social Security income level, unlike the dataset I used in the paper I mentioned above.
User avatar
DRiP Guy
Posts: 2241
Joined: Tue Feb 20, 2007 3:54 pm

Post by DRiP Guy »

First, from Wade:
wade wrote:DRiP Guy, I'm not understanding your response to Jim. You did catch that he said "withdrawal amount" rather than "withdrawal rate," right? My understanding is that Jim is referring to the safe withdrawal rate paradox.

Then, this is what he refers to, from Jim:
Let's face it, it just seems wrong that the traditional SWR methodology can result in dramatic differences in withdrawal amounts for two retirees that had identical savings habits, but retired just a year or two apart. That just doesn't seem like it could be right.
1. I did not catch the word 'amount' verses 'rate,' so yes I erred -- please excuse my sloppy reading.

2. But even with that happy correction, I am loath to agree that there is any real 'paradox' at work here at all, nor has there ever been:

* If I saved carefully and diligently for all my life, from a boy, and only manage to amass $400K at the end due to lower lifetime salary, and my friend saved not a whit all his life, until one year prior to retirement, at which time he somehow salted away $1.4MM all in one fell swoop due to inheritance, then I think no one would have an issue with saying he can still clearly withdraw a higher *amount* than I, but the historical record shows we can both withdraw more or less 4% over 30 years, if there is yet to be a fancy new proven backtested scheme out, and we want to do it at a flat continuous relatively safe rate, based on historical information, increased only for inflation. I'm hoping these are not largely in dispute.

* The fact is that if a different retiree pair (perhaps twins), with their only difference in saving rate or anything from each other is that they entered the workforce a year apart, and in the year that brother Abe retired and his wife was so conservative that she made him go to 100% fixed ($1MM), while Brother Barry stayed at a 80/20 allocation and lost 50% ($500K) to a sudden unexpected but temporary crash - say, another 9/11 -- before he could move to their planned 50/50 retirement target, then I think (but I believe Wade is working to dispute with data?) that it is clear as can be that poor Brother Barry is gonna get only half the payout that Abe does, but they can both still likely withdraw about 4% from the remaining nestegg -- using the standard fixed tables from Trinity, and assuming Abe's prescient allocation shift does go back to ~60/40 or 50/50.

I understand that you want to explore modifying this, increasing it, and making it flexible, and so have many others who came before. So, while I have amended my initial error in reading, I have to still ask: is there still some point I am missing: Where is the paradox?

PS -- Yes, I have read Kitces on this:
www.kitces.com/assets/pdfs/Kitces_Report_May_2008.pdf

Perhaps the 'real' paradox, if there is one, is the one about scarcity-thinking, here:
http://www.fa-mag.com/component/content ... &Itemid=27

I suppose at the end, I come back full circle to the 'no paradox camp, as in this article:
This is the first misuse of the guideline. It is not intended as a means by which to manage or adjust spending. It is intended as a reasonableness benchmark for making financial decisions. In the first year, the first retiree made the decision to retire and established a budget based on the belief that a 4 percent withdrawal rate was safe. After withdrawing 4 percent and experiencing a 21 percent investment loss, he now has 25 percent less wealth. He was unlucky and his retirement plan got off to a bad start. He cannot continue to spend $80,000 plus inflation if he still wants an 80 percent chance of maintaining a constant inflation-adjusted withdrawal rate.

The 4 percent guideline is reasonably safe but not 100 percent safe. Actual events must be considered and future spending must adapt. The paradox of the two retirees is not really a paradox, it is a misunderstanding of the meaning of the 4 percent guideline.
http://www.orlandomedicalnews.com/prope ... ne-cms-494



First there is a mountain then there is no mountain, then there is;

First there is a mountain then there is no mountain, then there is.


--Donovan
Last edited by DRiP Guy on Tue Feb 22, 2011 9:26 pm, edited 1 time in total.
KyleAAA
Posts: 9497
Joined: Wed Jul 01, 2009 5:35 pm
Contact:

Post by KyleAAA »

I think a 30 year career is a bit low. Most people will work more like 40 years. A lot can happen in that last 10 years.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

KyleAAA wrote:I think a 30 year career is a bit low. Most people will work more like 40 years. A lot can happen in that last 10 years.
I think it will be hard enough to convince people to save consistently for 30 years. If you can convince them to save for 40 years, then more power to you. For the baseline case I consider, if the only thing you change is that someone saves for 40 years instead of 30 years, then the "safe savings rate" is 8.77% instead of 16.62%. On the other hand, if the person waits and doesn't start saving until the final 20 years, the safe savings rate jumps up to 35.91%. Please use this information to convince people to start saving as early as possible!

DRiP Guy wrote: * If I saved carefully and diligently for all my life, from a boy, and only manage to amass $400K at the end due to lower lifetime salary, and my friend saved not a whit all his life, until one year prior to retirement, at which time he somehow salted away $1.4MM all in one fell swoop due to inheritance, then I think no one would have an issue with saying he can still clearly withdraw a higher *amount* than I, but the historical record shows we can both withdraw more or less 4% over 30 years, if there is yet to be a fancy new proven backtested scheme out, and we want to do it at a flat continuous relatively safe rate, based on historical information, increased only for inflation. I'm hoping these are not largely in dispute.
I agree.
DRiP Guy wrote: * The fact is that if a different retiree pair (perhaps twins), with their only difference in saving rate or anything from each other is that they entered the workforce a year apart, and in the year that brother Abe retired and his wife was so conservative that she made him go to 100% fixed ($1MM), while Brother Barry stayed at a 80/20 allocation and lost 50% ($500K) to a sudden unexpected but temporary crash - say, another 9/11 -- before he could move to their planned 50/50 retirement target, then I think (but I believe Wade is working to dispute with data?) that it is clear as can be that poor Brother Barry is gonna get only half the payout that Abe does, but they can both still likely withdraw about 4% from the remaining nestegg -- using the standard fixed tables from Trinity, and assuming Abe's prescient allocation shift does go back to ~60/40 or 50/50.

I understand that you want to explore modifying this, increasing it, and making it flexible, and so have many others who came before. So, while I have amended my initial error in reading, I have to still ask: is there still some point I am missing: Where is the paradox?
I hope I am not changing some "rule" about it, but I don't think the paradox applies for people who have different asset allocations.

I agree with what you wrote here, and I don't find any controversy about that.

The paradox is simply if these two brothers have everything the same including the same asset allocation, but retire one year apart from each other, and that year happens to bring a significant drop to both of their portfolios, then the brother retiring later will have a lower withdrawal amount despite both of them using the 4% rule.
DRiP Guy wrote:I suppose at the end, I come back full circle to the 'no paradox camp, as in this article:
This is the first misuse of the guideline. It is not intended as a means by which to manage or adjust spending. It is intended as a reasonableness benchmark for making financial decisions. In the first year, the first retiree made the decision to retire and established a budget based on the belief that a 4 percent withdrawal rate was safe. After withdrawing 4 percent and experiencing a 21 percent investment loss, he now has 25 percent less wealth. He was unlucky and his retirement plan got off to a bad start. He cannot continue to spend $80,000 plus inflation if he still wants an 80 percent chance of maintaining a constant inflation-adjusted withdrawal rate.

The 4 percent guideline is reasonably safe but not 100 percent safe. Actual events must be considered and future spending must adapt. The paradox of the two retirees is not really a paradox, it is a misunderstanding of the meaning of the 4 percent guideline.
I understand that the 4% rule is just a guideline and is not set in stone. I think here is where we are going in a different direction though: you are suggesting that after a major market drop, the already retired person should lower his withdrawals (despite it not being required by the 4% rule) to keep things in harmony.

On the other hand, I am saying that those already retired don't need to make any changes to their initial withdrawal rate, but that new retirees can go ahead and enjoy a higher withdrawal rate than the previous retiree so that both the old and new retirees can enjoy roughly the same withdrawal amount.

Did I express your position correctly? I'm sorry if I didn't, but based on my understanding, it seems like we've reversed our roles! I always argued that 4% is too high for recent retirees, while you stood by 4%. Now look at us. Well, I guess this is actually consistent with what you've believed all along: go ahead and start your retirement with a 4% withdrawal rate, but always stand ready to lower your withdrawals should things take a turn for the worse. That sounds fine to me. It's the Guyton/Klinger view, and I have no objection to that. The reason why I was more worried than you about whether 4% is too high for the initial withdrawal rate, is because I was implicitly assuming that it would be very bad if the retiree ever had to lower his withdrawals after retiring. But both of these models are competitive and will serve the different needs of different people. I feel like we might be on the same wavelength about this now...
SpecialK22
Posts: 844
Joined: Tue Sep 01, 2009 3:16 pm

Post by SpecialK22 »

wade wrote: The paradox is simply if these two brothers have everything the same including the same asset allocation, but retire one year apart from each other, and that year happens to bring a significant drop to both of their portfolios, then the brother retiring later will have a lower withdrawal amount despite both of them using the 4% rule.
That indeed shows the rigidity of the 4% rule and does appear to create a paradox. What's even more interesting is the brother who retired first will have a slightly smaller portfolio (at the time the second brother retires) due to one year of withdrawals and one year less of contributions. I wonder how (say) basing the 4% withdrawal rate on the average portfolio balance of the three years preceding retirement would fare. Granted, the paradox still exists--particularly for any drastic changes in portfolio value during those three years--but differences in withdrawal amounts have the potential to be less steep.
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Post by magellan »

SpecialK22 wrote:I wonder how (say) basing the 4% withdrawal rate on the average portfolio balance of the three years preceding retirement would fare.
I wondered this myself and worked up a spreadsheet based on Bob's cool MWR spreadsheet. I used a conservative approach that used the lower of the portfolio balance at retirement, or 1, 2, or 3 years back. The results were nearly a 1% increase in the SWR. As I said above, the problem with this is that there are really only a few data points that drive the entire thing and it seems like data mining to find just the right algorithm that fits the data.

Here's a link to the spreadsheet. Also,here's another spreadsheet that has summary data for the lookback algorithm and also considers what the MWR would be if we only look at retirements that started in years with PE10 less than 15 or 20.

These were more "playing around" spreadsheets that I cooked up pretty quickly. They could easily have errors in them, so take the results with the appropriate tons of salt.

Jim
grberry
Posts: 234
Joined: Fri Jul 20, 2007 1:16 pm
Location: Boston, MA

Post by grberry »

wade wrote:
I think it will be hard enough to convince people to save consistently for 30 years. If you can convince them to save for 40 years, then more power to you. For the baseline case I consider, if the only thing you change is that someone saves for 40 years instead of 30 years, then the "safe savings rate" is 8.77% instead of 16.62%. On the other hand, if the person waits and doesn't start saving until the final 20 years, the safe savings rate jumps up to 35.91%. Please use this information to convince people to start saving as early as possible!
Put everything I mentioned above on the back burner. This analysis is the most important one to get published! If it can also be done with a couple other asset allocation rules, great. But this can make a real difference in people's lives if we have it as an example to point to.
thedude
Posts: 422
Joined: Wed Mar 07, 2007 6:44 pm

Post by thedude »

Dandy wrote:Getting to a high savings rate is the trick. People hear you must save 10% or more and they feel they can't do it.

What worked with me was when I started work an "old timer" said every time you get a raise increase your savings -- you'll never miss what you never had.

That was a great way to live below your means from the start.
Or go to grad school and live off of $21K a year in an expensive city. It's enough to live off of but not enough to buy useless crap. If people would just remember how little they needed when they were students, perhaps they'd be able to save a little bit more. It's worked great for me.
User avatar
LH
Posts: 5490
Joined: Wed Mar 14, 2007 2:54 am

Post by LH »

Interesting thread : )
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

wade wrote: These figures show how the MWR responds to past market returns. Figure 2 directly shows that from year to year, high asset returns will push down the MWR for a new retiree in the subsequent year, while low asset returns push up the MWR for a new retiree in the subsequent year.

Don't you mean to say "will push down the CHANGE IN the MWR..." ???

You're way of looking at this is excellent. Vertical & horizontal thinking...

So your regression equation shows that if your investment gain is zero, you need to reduce your previous year's SWR by 31%.

But a 5% gain essentially negates that reduction since .06 x 5 = .30

My brain wants to answer the question "what is the Multiplication Factor" (ie last year's SWR x the Mult Factor = This years SWR). But that would make it difficult to discover the correlation of investment return.
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

The regression line for Fig 4 says that a 11.5% savings rate is required for the "mental benchmark" 4% SWR:

1.59 + .22(11.5) = 4.12%

Close to the "standard benchmark" to save 10%.

However the equation's Confidence Level is not excellent.

So what is the average savings rate of black line in Fig 5 ? It seems to be a bit higher than 11.5%. But I'm not sure if they should give the same results...

Also your 50% replacement rate (even though you insightfully showed it to be actually 60%, by correctly not counting what you were saving) seems to be a lower standard than the "common rule of thumb" of 80%. But as you pointed out, additional money from SS and pensions are usually unspoken asuumptions.

Now, if you could expand the equation to where I could plug in the % of my retirement spending which will be met from [SS + Pensions] then my "true" required savings rate will emerge...
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

Can we extract something from the blue line in Fig 5 regarding the magnitude of the timing of "bad returns" during retirement?

I notice that the highest required savings rate occurred for someone who retired about 8 years before the Crash of 1929.

So is ~8 years into retirement the worst case for bad sequences?

Can we look at the blue line and get some insight about the most dangerous time for bad sequences to occur?
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Bongleur wrote:
wade wrote: These figures show how the MWR responds to past market returns. Figure 2 directly shows that from year to year, high asset returns will push down the MWR for a new retiree in the subsequent year, while low asset returns push up the MWR for a new retiree in the subsequent year.

Don't you mean to say "will push down the CHANGE IN the MWR..." ???
Thank you everyone for the comments.

Bongleur, your question is very valuable as I can try to clarify some things that I may not have explained clearly.

Image

So, I do not mean to say "will push down the CHANGE IN the MWR..." Well, I guess this can be applicable too, but it isn't what I was trying to say.

A high positive return leads the change in MWR to be a negative number. That means the MWR for the next year will be less than this year. In other words, it will get pushed down.
Bongleur wrote:So your regression equation shows that if your investment gain is zero, you need to reduce your previous year's SWR by 31%.
These are percentage points, not percentages. And I think you are saying this a bit backwards.

The regression equation says that if investment returns = 0, then the change in MWR is +0.31

This means, for example, if the MWR was 6 percent for a retiree last year, it will be 6 + 0.31 = 6.31 percent for a retiree this year after the market had stayed flat.

Also we must be clear to distinguish between MWRs and SWRs. The MWR is a number you can't know until 30 years after you retired. It is something calculated after the fact. The SWR is the withdrawal rate that can be thought of as safe. So, traditionally, 4% is thought of as the SWR, though people understand that most retirees will be lucky enough to enjoy a higher MWR. They just don't have anyway to know this in advance, which is why it could be a good idea to stick with 4%. [Or can MWRs be predicted in advance? That is a background theme of this paper, and also a more specific theme of the paper I have on predicting MWRs].
Bongleur wrote:But a 5% gain essentially negates that reduction since .06 x 5 = .30
Yes, a 5% real gain on the portfolio (approximately) will mean that the MWR stays constant for two years in a row.

To put more context to this, consider these changes in MWRs with respect to the up and down movements you see in this figure:

Image

The 1920s was a great decade for investors, and you can see the MWR drop dramatically throughout the decade. On the other hand, the 1970s was a pretty miserable decade for investors, and you can see the MWR rise dramatically throughout that decade.
Bongleur wrote:My brain wants to answer the question "what is the Multiplication Factor" (ie last year's SWR x the Mult Factor = This years SWR). But that would make it difficult to discover the correlation of investment return.
Again, I want you to replace SWR with MWR. But that being said, I don't have anything in this paper to specifically answer that. But this is the kind of issue I considered in another paper:

"Predicting Sustainable Retirement Withdrawal Rates Using Valuation and Yield Measures."
http://ideas.repec.org/p/pra/mprapa/27487.html
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

Going back to the eqn for Fig 2:

A portfolio loss of 21.833% means that you need to reduce your withdrawal rate to ZERO...

Is this the point of portfolio failure, or could the withdrawal rate recover (to original???) if you actually did cease withdrawals for that year...

The eqn fails because even an infinitely large gain in the subsequent year is being multiplied by the paused year's withdrawal value of zero percent.

A little confused by very tired too...

Wondering if taking out a loan for 100% of your expenses, and hoping for future gains to pay it off, could be a way to have a chance of preventing failure...

Hypothesis: If you must exceed your SWR, take out a loan for the excess amount and hope that future gains will be high enough to pay it off without again exceeding your SWR. This will prevent portfolio failure at this time due to exceeding your SWR. Keep rolling the loans...

Corollary: Bankruptcy is a Free Lunch because you always get to keep _something_.
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Bongleur wrote:The regression line for Fig 4 says that a 11.5% savings rate is required for the "mental benchmark" 4% SWR:

1.59 + .22(11.5) = 4.12%

Close to the "standard benchmark" to save 10%.

However the equation's Confidence Level is not excellent.
I'm afraid that is not what this figure is about. Rather, it is that if you were lucky enough to have worked at a time that only required you use use an 11.5% savings rate to accumulate 12.5x your final salary by retirement, then your luck runs out after retirement because your maximum sustainable withdrawal rate will only be about 4.12 percent.

Yes, there is some variability around that. It is not exact. But an R2 of 0.79 is not too shabby.

Bongleur wrote:So what is the average savings rate of black line in Fig 5 ? It seems to be a bit higher than 11.5%. But I'm not sure if they should give the same results...
Please try reading the paper again. And if you could use more background, the Bogleheads Wiki is a great place to start:

http://www.bogleheads.org/wiki/Safe_Withdrawal_Rates
Bongleur wrote:Also your 50% replacement rate (even though you insightfully showed it to be actually 60%, by correctly not counting what you were saving) seems to be a lower standard than the "common rule of thumb" of 80%. But as you pointed out, additional money from SS and pensions are usually unspoken asuumptions.
Right, but it isn't an unspoken assumption. I said it :D
Bongleur wrote:Now, if you could expand the equation to where I could plug in the % of my retirement spending which will be met from [SS + Pensions] then my "true" required savings rate will emerge...
I've taken care of this. First, figure out your social security benefits, defined-pension benefits, part-time work, etc. Then, deduct that from the amount you want to spend in retirement. What is left over is the replacement rate that you need from your retirement savings. The 50% I use refers to this final amount.
Bongleur wrote:Can we extract something from the blue line in Fig 5 regarding the magnitude of the timing of "bad returns" during retirement?

I notice that the highest required savings rate occurred for someone who retired about 8 years before the Crash of 1929.

So is ~8 years into retirement the worst case for bad sequences?

Can we look at the blue line and get some insight about the most dangerous time for bad sequences to occur?
After you retire, the very worst time for a bad sequence of returns is immediately after retiring. The effects of bad sequences decrease as you move beyond the retirmeent date.

I explain about this point a lot in this paper:

Will 2000-Era Retirees Experience the Worst Retirement Outcomes in U.S. History? A Progress Report after 10 Years."
http://ideas.repec.org/p/pra/mprapa/27107.html

The blue line in Figure 5 is about the savings rate you need in the 30 years before retirement, so it has no connection to events from after retirement.

Bongleur wrote:Going back to the eqn for Fig 2:

A portfolio loss of 21.833% means that you need to reduce your withdrawal rate to ZERO...

Is this the point of portfolio failure, or could the withdrawal rate recover (to original???) if you actually did cease withdrawals for that year...

The eqn fails because even an infinitely large gain in the subsequent year is being multiplied by the paused year's withdrawal value of zero percent.

A little confused by very tired too...

Wondering if taking out a loan for 100% of your expenses, and hoping for future gains to pay it off, could be a way to have a chance of preventing failure...

Hypothesis: If you must exceed your SWR, take out a loan for the excess amount and hope that future gains will be high enough to pay it off without again exceeding your SWR. This will prevent portfolio failure at this time due to exceeding your SWR. Keep rolling the loans...

Corollary: Bankruptcy is a Free Lunch because you always get to keep _something_.
About Figure 2, please see my explanation up above in the post before this post. When you see the change in MWR= -1 that just means the MWR falls by 1 percentage point for that new retiree. It is not some kind of portfolio failure.

I don't really think your loan idea sounds too responsible, but you can always exceed your desired withdrawal rate for an emergency. It just increases the chances that you will run out of wealth sooner. You don't need a loan to do it though.

Best wishes
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

re Fig 6 in the blog post:

What savings rates correspond to the withdrawal rate (red line) being less than or equal to the max sustainable rate (black line) 85%, 90%, and 95% of the time?

Wondering if this "backtest" can help a person answer the question of "given my starting $$, can I sustain some rate above 4%?"
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Bongleur wrote:re Fig 6 in the blog post:

What savings rates correspond to the withdrawal rate (red line) being less than or equal to the max sustainable rate (black line) 85%, 90%, and 95% of the time?

Wondering if this "backtest" can help a person answer the question of "given my starting $$, can I sustain some rate above 4%?"
That is a good question. You are basically asking for the "Trinity Study" treatment provided for William Bengen's original SWR study.

My current study corresponds with William Bengen's original SWR study. I haven't given it the "Trinity study" treatment yet, so I can't answer your question right now.

But I will do this sometime. It will not be too difficult for me to check, but I just haven't done it yet.

But about the last part of your question, how much money you have at retirement is not at all related to the 4% rule in traditional retirement studies. I am only trying to make it relevant now in this study by switching the focus to the "safe savings rate" rather than the "safe withdrawal rate".

More generally, if you want to know what your best guess for your MWR is, I suggest you rely on the method described in my other paper about predicting MWRs.

I have a simple Excel spreadsheet on my blog to help do this, but I STRONGLY recommend that you read the paper first because the Excel spreadsheet is very basic and not user friendly. You need to understand the meaning of the various inputs.
http://wpfau.blogspot.com/2010/12/this- ... about.html
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

>
More generally, if you want to know what your best guess for your MWR is, I suggest you rely on the method described in my other paper about predicting MWRs.
>

Which other paper? URL ?

I like that you are providing a completely different way of analyzing the SWR problem. If your way and the "traditional" way give the same answers to some of the same questions, then we have some confidence that the answers to questions that are only answerable by one method or the other are robust.

We know with a lot of confidence that we can spend 4% of whatever we have. In fact, I think it is relied upon in real life.

But sometimes we also know that it is not enough, or that we don't want to leave lots of money on the table when we die. So putting confidence levels on other rates is important.

***

You can backtest more data sequences by doing only 15 or 20 or 25 years. This is useful, because many people don't actually live longer than that.

If you also define success as having enough money in the bank at the end of the shorter time period to carry forward to 30 years using an annuity or some such conservative minimum survival strategy, then you can increase the number of data points that have some statistically significant correlation to the 30 year period we are most interested in.

***

This makes me think about something else. If you have "plenty" of money, you don't need an annuity. But buying one will get you a much larger "extra" amount compared to someone who buys an annuity because he must do so to meet his minimum needs.

So that _guaranteed_ "extra" amount, being pretty large, might be large enough, and therefore be preferable to risky investments with your margin money even though the risky investments probably create more "extra" money. Because your marginal utility of wealth is rather low to start with.
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Bongleur wrote:>
More generally, if you want to know what your best guess for your MWR is, I suggest you rely on the method described in my other paper about predicting MWRs.
>

Which other paper? URL ?
Thanks for all your comments. I will think about them some more.

The paper is this one:

"Predicting Sustainable Retirement Withdrawal Rates Using Valuation and Yield Measures."
http://ideas.repec.org/p/pra/mprapa/27487.html
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox »

I think you are on the right track looking at savings rates. This confirms what I have been thinking, that it really all comes down to how much you consume vs. how much you produce. I think these principles are universal and apply to everyone and everything, beyond stocks, bonds, money. It is applicable to workers today, ancient civilizations, reservoirs and the water supply, bees making honey, animals storing food for the winter, bears storing fat to consume while they hibernate.

I've mentioned before that I am not a huge believer in proving things by backtesting. You can really get caught up in the minutia of interest rates, historical PE ratios, dividend payout ratios, etc. and lose track of the underlying principles of production, consumption and saving. My philosophy is to derive principles from simple logic and basic arithmetic that a 2nd grade can understand. Later, it makes sense to examine past data to see if the ideas held up in the past.

So I concluded from simple arithmetic that a baseline savings rate is 50%. Produce 100 bushels of grain per year, consume 50 bushels, and stockpile 50 bushels each year in the silo. If you do this for 7 fat years, then you can live off them for 7 lean years. Or work for 30 years saving 50% and retire for 30 years on the accumulated savings.

This assumes zero real return, meaning that the amount of grain in the silo or gold in the vault doesn't magically increase, but you also don't lose any grain to rot.

Now if you can magically get some real return, your savings rate decreases. I created a spreadsheet to model production, consumption and savings and came up with this simple table.

Code: Select all

Savings Real Return
Rate    Needed
50      0%
40      1.02%
30      2.14%
25      2.78%
20      3.53%
15      4.43%
10      5.65%
5       7.64%
2       10.22%
If we're expecting about 3% real return for a 50/50 portfolio, that would imply a savings rate of 20 to 25 percent. This is a bit more than your 16.62% MSR from the paper.

Maybe we can knock this 25% savings rate down further if you assume that there will be some other source of income like pension or social security. But that just means that some else, like your employer or Uncle Sam, did some of your saving for you. Maybe the company pension and social security saved another 10% beyond the your 16.62%. So the overall savings rate is not really reduced. Otherwise, a lower savings rate would imply supporting you either for fewer years or at a reduced level of consumption.

:arrow: Anyway, to summarize, if there is zero real return, then somebody somewhere has to save 50% of what you produce. If you can be assured of a real return, you can increase consumption and lower the savings rate. The lower the savings rate, the higher the required return.

Here is a question to ponder. What is the source of "magical" real return that makes the pile grow without adding to it?
Real things like grain in an elevator or gold in a vault don't increase. What about bonds? I know it is interest payments, but where does interest come from? How is it that you can put in $100 and get back $105 one year later? Isn't someone else going to be $5 poorer?

About the only organic growth I understand a fundamental reason behind it is for timberland.

Critical Mass in Retirement
How much to save?
User avatar
Topic Author
Majormajor78
Posts: 910
Joined: Mon Jan 31, 2011 8:13 pm

Post by Majormajor78 »

grayfox wrote:Here is a question to ponder. What is the source of "magical" real return that makes the pile grow without adding to it?
Real things like grain in an elevator or gold in a vault don't increase. What about bonds? I know it is interest payments, but where does interest come from? How is it that you can put in $100 and get back $105 one year later? Isn't someone else going to be $5 poorer?

About the only organic growth I understand a fundamental reason behind it is for timberland.

Critical Mass in Retirement
How much to save?
A very intutive analogy of the discussion. As for the "magical" real return I would attribute this to increases in efficiency and productivity. In essence a corporation "consumes" less resources and labor to generate an equal or greater amount of goods or services. This would account for the "extra $5" you mentioned.
"Oh, M. le Comte, it is only a loss of money which I have sustained... nothing worth mentioning, I assure you."
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox »

Majormajor78 wrote:
grayfox wrote:Here is a question to ponder. What is the source of "magical" real return that makes the pile grow without adding to it?
Real things like grain in an elevator or gold in a vault don't increase. What about bonds? I know it is interest payments, but where does interest come from? How is it that you can put in $100 and get back $105 one year later? Isn't someone else going to be $5 poorer?

About the only organic growth I understand a fundamental reason behind it is for timberland.

Critical Mass in Retirement
How much to save?
A very intutive analogy of the discussion. As for the "magical" real return I would attribute this to increases in efficiency and productivity. In essence a corporation "consumes" less resources and labor to generate an equal or greater amount of goods or services. This would account for the "extra $5" you mentioned.
It seems like there would be diminishing returns from productivity increases. 150 years ago 80% of the population worked on farms. Doubling productivity freed up 40% of the workers. The next doubling only freed up 20% of the workers. Now maybe 2% work on farms and produce enough food for everyone and doubling efficieny now would only free up another 1% of the workforce.

Think about where the goverment got $105 to pay me off at the end of the year. Don't they just roll over short term T-bills? So they borrow $105 from some new investor to pay off the old investor. Sounds like Ponzi.

The reason I ask where does real return come from is because I'm trying to figure out if it is something that can be counted on. As I reasoned above, if you have zero real return, then the Safe Savings Rate is 50%.

Now I understand that stocks and bonds had a real return in the past, but can you count on it going forward? Sure, today you can get 2% real in a 30-year TIPS, so for now, at least, there is 2% you can count on. That would put SSR around 30%.

Here is something I find interesting.
American savings rate is about 5%, so the required real return would be 7.6% by my simple model. I think this is close to what U.S. stocks have returned since 1802 according to Jeremy "Long Run" Siegel.
I read China's gross saving rate was 53% in 2008. So they require zero real return. No wonder they're willing to buy some many U.S. Treasuries at such low yields.
User avatar
Topic Author
Majormajor78
Posts: 910
Joined: Mon Jan 31, 2011 8:13 pm

Post by Majormajor78 »

grayfox wrote:
Majormajor78 wrote:
grayfox wrote:Here is a question to ponder. What is the source of "magical" real return that makes the pile grow without adding to it?
Real things like grain in an elevator or gold in a vault don't increase. What about bonds? I know it is interest payments, but where does interest come from? How is it that you can put in $100 and get back $105 one year later? Isn't someone else going to be $5 poorer?

About the only organic growth I understand a fundamental reason behind it is for timberland.

Critical Mass in Retirement
How much to save?
A very intutive analogy of the discussion. As for the "magical" real return I would attribute this to increases in efficiency and productivity. In essence a corporation "consumes" less resources and labor to generate an equal or greater amount of goods or services. This would account for the "extra $5" you mentioned.
It seems like there would be diminishing returns from productivity increases. 150 years ago 80% of the population worked on farms. Doubling productivity freed up 40% of the workers. The next doubling only freed up 20% of the workers. Now maybe 2% work on farms and produce enough food for everyone and doubling efficieny now would only free up another 1% of the workforce.

Think about where the goverment got $105 to pay me off at the end of the year. Don't they just roll over short term T-bills? So they borrow $105 from some new investor to pay off the old investor. Sounds like Ponzi.

The reason I ask where does real return come from is because I'm trying to figure out if it is something that can be counted on. As I reasoned above, if you have zero real return, then the Safe Savings Rate is 50%.

Now I understand that stocks and bonds had a real return in the past, but can you count on it going forward? Sure, today you can get 2% real in a 30-year TIPS, so for now, at least, there is 2% you can count on. That would put SSR around 30%.

Here is something I find interesting.
American savings rate is about 5%, so the required real return would be 7.6% by my simple model. I think this is close to what U.S. stocks have returned since 1802 according to Jeremy "Long Run" Siegel.
I read China's gross saving rate was 53% in 2008. So they require zero real return. No wonder they're willing to buy some many U.S. Treasuries at such low yields.
I'm not entirely sure what you're driving at is completely reasonable but I could be interpreting it incorrectly so please bear with me. Most of the productivity increases we have seen stem not from labor being freed from agrarian employment to industrial employment. Most of the productivity increases recently come from computers, logistics, mechanical systems, ect. Admitedly these improvements cause localized pressures as some people are laid off due to robots and the like but the growth the effcienies create open up opportuniites elsewhere... on a macro scale things balance out. The U.S. bureau of labor statistics keeps very detailed records of productivity ( http://www.bls.gov/news.release/prod2.nr0.htm ). You are assuming that productivity is a zero sum game where one unit of labor today equals one unit of labor ten years from now and it surely is not.

You stated "Now I understand that stocks and bonds had a real return in the past, but can you count on it going forward? Sure, today you can get 2% real in a 30-year TIPS, so for now, at least, there is 2% you can count on. That would put SSR around 30%."

Of course you can not guarantee that stocks and bonds will have a real return going forward. As Jack Bogle said in the first paragraph of chapter 1 (p.3) in his book Common Sense on Mutual Funds 10th anniversary edition he stated, "Investing is an act of faith. We entrust our capital to corporate stewards in the faith - at least the hope - that their efforts will generate high rates of return on our investments. When we purchase corporate America's stocks and bonds, we are professing our faith that the long-term success of the U.S. economy and the nation's financial markets will continue in the future." Could this faith be misplaced? Could be. Personally I believe that over the next 30 years our economy will have it's ups and downs but will still exist when I come to retirement age. And if it doesn't? If the global economy collapses into chaos leaving everybody to fend for themselves is some kind of Thunderdome scenario then I fail to see what actions I can take today to protect myself in this environment. Be frugle, save agressivly, invest wisely... these choices can raise the probability that I'll be able to retire comfortably to a near certaintude, but I will not base my decisions on a rock-bottom, absolute worst case scenario. There is no such thing as a guarantee... only probabilities.
"Oh, M. le Comte, it is only a loss of money which I have sustained... nothing worth mentioning, I assure you."
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

grayfox wrote:I think you are on the right track looking at savings rates. This confirms what I have been thinking, that it really all comes down to how much you consume vs. how much you produce. I think these principles are universal and apply to everyone and everything, beyond stocks, bonds, money. It is applicable to workers today, ancient civilizations, reservoirs and the water supply, bees making honey, animals storing food for the winter, bears storing fat to consume while they hibernate.

I've mentioned before that I am not a huge believer in proving things by backtesting. You can really get caught up in the minutia of interest rates, historical PE ratios, dividend payout ratios, etc. and lose track of the underlying principles of production, consumption and saving. My philosophy is to derive principles from simple logic and basic arithmetic that a 2nd grade can understand. Later, it makes sense to examine past data to see if the ideas held up in the past.

So I concluded from simple arithmetic that a baseline savings rate is 50%. Produce 100 bushels of grain per year, consume 50 bushels, and stockpile 50 bushels each year in the silo. If you do this for 7 fat years, then you can live off them for 7 lean years. Or work for 30 years saving 50% and retire for 30 years on the accumulated savings.

This assumes zero real return, meaning that the amount of grain in the silo or gold in the vault doesn't magically increase, but you also don't lose any grain to rot.

Now if you can magically get some real return, your savings rate decreases. I created a spreadsheet to model production, consumption and savings and came up with this simple table.

Code: Select all

Savings Real Return
Rate    Needed
50      0%
40      1.02%
30      2.14%
25      2.78%
20      3.53%
15      4.43%
10      5.65%
5       7.64%
2       10.22%
If we're expecting about 3% real return for a 50/50 portfolio, that would imply a savings rate of 20 to 25 percent. This is a bit more than your 16.62% MSR from the paper.

Maybe we can knock this 25% savings rate down further if you assume that there will be some other source of income like pension or social security. But that just means that some else, like your employer or Uncle Sam, did some of your saving for you. Maybe the company pension and social security saved another 10% beyond the your 16.62%. So the overall savings rate is not really reduced. Otherwise, a lower savings rate would imply supporting you either for fewer years or at a reduced level of consumption.

:arrow: Anyway, to summarize, if there is zero real return, then somebody somewhere has to save 50% of what you produce. If you can be assured of a real return, you can increase consumption and lower the savings rate. The lower the savings rate, the higher the required return.

You are bringing up lots of good points. I think Majormajor78 is making good comments about the productivity issue. But let me try to address your other points.

First, I agree about the 50% savings rate with zero percent returns. But we are going to have some slight differences when it comes to your table of savings rates and required returns, because I assume:

-investments are made at the end of each year during accumulation phase
-withdrawals are made at start of each year during retirement phase

I think your assumption here is a bit different, but I didn't check to see which way.

But this is still very interesting. I made a little Excel spreadsheet about this. The 1918 retiree is the one who defined the 16.62 savings rate as the "safe savings rate" because he had the most unfortunate situation. The geometric return on his investments in the 30 years before and after retirement was 4.3%. So I compared a hypothetical 60 year lifecycle with a constant 4.3% real return, against the lifecycle of the 1918 retiree.

Despite having the same geometric average return, the hypothetical guy retired with more wealth (980.2 vs. 687.2) and ended a 30 year retirement with more wealth (390.5 vs. -1).

This points to the important issue of sequence of returns risk, which is what your post has me thinking about. Why did things go so bad for the 1918 retiree compared with others?

Also, as an extra note. With a 4.3 percent constant real return assumption, I find that the safe savings rate is 14.75%. The actual safe savings rate is higher than this due to the sequence of returns!

Image
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Post by grayfox »

That is very interesting. The 1918 retiree was sailing along smoothly until 1916 - 1920 when he hit a string of five bad years. Must have had something to do with WWI. Reminds me of the three bad years 2000-2002.

The worst time to have a big losses is when your balance is largest.
Bongleur
Posts: 2276
Joined: Fri Dec 03, 2010 9:36 am

Post by Bongleur »

WRT the sequence of returns (SoR) issue:

If you analyze the effect of the average of the set of completely random SoRs...

...as a factor that changes the result of a steady average return...

...what happens?

I guess you get the same result as a steady average return?

Is it possible to find the "average" change caused by random SoRs and calculate its standard deviation?

If so, you can apply one or two of the SoR standard deviations to any average return, and calculate a safe withdrawal rate for that degree of "bad randomness" in sequence of returns.

***

Must we assume that "bad sequences" are completely random, or can we look at historical data and discover the number of "bad sequences" (defined in some way) and their length and other qualities (such as their internal pattern of change) ?

If so, you can insert those patterns of "bad sequences" into the SoR generator to create results that are average, or worse by some measurable standard deviation.

FireCalc has you repeating history; the above would give random results within a framework based on history. Not the same as a Monte Carlo.
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

Bongleur wrote:Must we assume that "bad sequences" are completely random, or can we look at historical data and discover the number of "bad sequences" (defined in some way) and their length and other qualities (such as their internal pattern of change) ?
You have some good points here. And you can do what you are suggesting, but I am cautious about that. The more parameters you are trying set based on the historical data, such as length or qualities of bad sequences, then the less reliable of simulations will be. That is why for the topic of this thread, I am only comfortable basing any decisions using historical data. But what you are saying could be used if you want to try to develop a simulated "safe savings rate" that could be higher than what historical data has shown to be necessary.
JW-Retired
Posts: 7189
Joined: Sun Dec 16, 2007 11:25 am

Post by JW-Retired »

wade,
Appreciate your good work. I read your paper and really like the idea of a safe savings rate. That's a useful concept for starting out savers. I will be sending a hard copy to my 30-something kids. If I email it they would be less likely to read it.
JW
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Post by wade »

I revised this article after receiving reviewer comments, and it has been accepted to appear in the May 2011 issue of Journal of Financial Planning. I thanked some of you for your helpful comments in my acknowledgments section. I've posted a revised version online, but the link is still not ready. I will post a link to the revised paper when I can.

There was nothing too major in my revisions, except for one startling thing.

I knew that administrative fees would be important, but I didn't include them in order to be consistent with existing research. But when I tried just adding in a 1% account fee that would be paid at the end of each year, it caused the "safe savings rate" in my baseline case to jump from 16.62% to 22.15%! :shock:

Wow! I didn't realize the effects of fees would be that big. If ever there is a reason to use low cost index funds for your investments, well that right there is a major reason!
User avatar
DRiP Guy
Posts: 2241
Joined: Tue Feb 20, 2007 3:54 pm

Post by DRiP Guy »

wade wrote:There was nothing too major in my revisions, except for one startling thing.

I knew that administrative fees would be important, but I didn't include them in order to be consistent with existing research. But when I tried just adding in a 1% account fee that would be paid at the end of each year, it caused the "safe savings rate" in my baseline case to jump from 16.62% to 22.15%! :shock:

Wow! I didn't realize the effects of fees would be that big. If ever there is a reason to use low cost index funds for your investments, well that right there is a major reason!
There's hope for you to become a Boglehead, yet Wade!

On a personal note, glad to hear you and the family will be returning stateside for a bit. Sounds prudent at this point.



8) :lol:
Post Reply