Variable spending strategies paper by Pfau

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills.
Post Reply
Topic Author
Beliavsky
Posts: 1233
Joined: Sun Jun 29, 2014 10:21 am

Variable spending strategies paper by Pfau

Post by Beliavsky »

Looking at the tables in the paper, no spending strategy stands out as the best to me. Maybe that is the point -- there are trade-offs.

Making Sense Out of Variable Spending Strategies for Retirees
Wade D. Pfau
The American College
March 16, 2015
Abstract:
Variable spending strategies can be situated on a continuum between two extremes: spending a constant amount from the portfolio each year without regard for the remaining portfolio balance, and spending a fixed percentage of the remaining portfolio balance. Variable spending strategies seek compromise between these extremes by avoiding too many spending cuts while also protecting against the risk that spending must subsequently fall to uncomfortably low levels. Two basic categories for variable spending rules explored include decision rule methods and actuarial methods. Ten strategies will be compared using a consistent set of portfolio return and fee assumptions, and using an XYZ formula to calibrate initial spending: the client willingly accepts an X% probability that spending falls below a threshold of $Y (in inflation-adjusted terms) by year Z of retirement. Presenting the distribution of spending and wealth outcomes for different strategies in which the initial spending rate is calibrated with the XYZ formula will allow for a more meaningful comparison of strategies. The article provides a framework for identifying appropriate spending strategies based on client preferences.

Number of Pages in PDF File: 18
Keywords: retirement planning, retirement income modeling, systematic withdrawals, variable spending in retirement
pkcrafter
Posts: 15461
Joined: Sun Mar 04, 2007 11:19 am
Location: CA
Contact:

Re: Variable spending strategies paper by Pfau

Post by pkcrafter »

Beliavsky wrote:
Looking at the tables in the paper, no spending strategy stands out as the best to me. Maybe that is the point -- there are trade-offs.
If you think about it, you will realize every investing/withdrawing decision involves trade-offs and compromise. You cannot know what optimal will be in the future.


Paul
When times are good, investors tend to forget about risk and focus on opportunity. When times are bad, investors tend to forget about opportunity and focus on risk.
User avatar
NoRoboGuy
Posts: 882
Joined: Fri Apr 01, 2011 11:07 pm
Location: Alabama
Contact:

Re: Variable spending strategies paper by Pfau

Post by NoRoboGuy »

Thanks for sharing this. I liked the focus on the different variables a retiree needs to consider before selecting a withdrawal strategy. One way to simplify this is to answer questions like these examples, before selecting a method:

How long is my expected retirement in years?
To what extent do pensions/social security cover my minimum spending needs?
Do I own a home (as an added "insurance policy" against portfolio depletion)?
Do I wish to leave a legacy (leave a meaningful inheritance)?
Do I have catastrophic/long term care health insurance?
How much of a decrease in discretionary spending can I accept if markets perform poorly?

I cannot overemphasize the other studies that show spending in retirement naturally declines with age regardless of remaining wealth available. Real spending does not need to be constant, particularly when health insurance needs have been fully addressed.
There is no free lunch.
etarini
Posts: 615
Joined: Sun Jan 24, 2010 5:46 pm

Wade Pfau's latest on Variable Spending Strategies

Post by etarini »

[Thread merged into here, see below. --admin LadyGeek]

Just got a Wade Pfau email that he has a new article, Making Sense Out of Variable Spending Strategies for Retirees

VPW and Guyton-Klinger have been hot topics lately, and I'll admit that I haven't followed them closely, but I hope to catch up soon.

Here's the link to his website featuring the article:

http://retirementresearcher.com/making- ... -retirees/


Eric
User avatar
JDCarpenter
Posts: 1800
Joined: Tue Sep 09, 2014 2:42 pm

Re: Wade Pfau's latest on Variable Spending Strategies

Post by JDCarpenter »

FWIW, he includes VPW among the approaches examined (and a References note directing the reader to the wiki)

Short paper, only 18 pages, including 2 page appendix on capital market expectations.
Our personal blog (no ads) of why we saved/invested: https://www.lisajtravels.com/
DFrank
Posts: 474
Joined: Thu Mar 20, 2014 10:44 pm
Location: North Idaho

Re: Wade Pfau's latest on Variable Spending Strategies

Post by DFrank »

Just finished a quick read of this interesting paper. It's interesting in light of the recent discussion here on Bogleheads about VPW and G-K withdrawal strategies. The paper seems to do a good job of comparing various approaches under a common set of assumptions, so comparisons of how they perform can be made more easily.
Dave
User avatar
ResearchMed
Posts: 16767
Joined: Fri Dec 26, 2008 10:25 pm

Re: Wade Pfau's latest on Variable Spending Strategies

Post by ResearchMed »

I especially liked his statement early on that

"Clients will not play the implied game of chicken by keeping their spending constant as their portfolios plummet toward zero."

This is usually overlooked in most discussions and graphs, I think. (Is this correct?)

Obviously, the closer to the "edge" one is living - and although that's not many here on BH, there are plenty of others who are not as fortunate, who didn't know how to plan, etc. - the less flexibility there is.

RM
This signature is a placebo. You are in the control group.
User avatar
ResearchMed
Posts: 16767
Joined: Fri Dec 26, 2008 10:25 pm

Fee of 0.5% for all strategies in simulations

Post by ResearchMed »

In reading further, I noticed that Pfau seems to add a 0.5% "fee" for all strategy simulations.

For those who don't pay for "advice" (such as many of us here on BH), would this mean we could/should add 0.5% to any withdrawal rate, to get the equivalent "outcome"?

The "fees" that are charge by mutual funds are already accounted for in the returns, so there shouldn't be a general "fee" for those of us who aren't in (or who can get out of) 401k/etc., accounts that have high fees forced into them.

Or am I misunderstanding how he uses the term "fees"?

Thanks.

RM
This signature is a placebo. You are in the control group.
dbr
Posts: 46137
Joined: Sun Mar 04, 2007 8:50 am

Re: Fee of 0.5% for all strategies in simulations

Post by dbr »

ResearchMed wrote:In reading further, I noticed that Pfau seems to add a 0.5% "fee" for all strategy simulations.

For those who don't pay for "advice" (such as many of us here on BH), would this mean we could/should add 0.5% to any withdrawal rate, to get the equivalent "outcome"?

The "fees" that are charge by mutual funds are already accounted for in the returns, so there shouldn't be a general "fee" for those of us who aren't in (or who can get out of) 401k/etc., accounts that have high fees forced into them.

Or am I misunderstanding how he uses the term "fees"?

Thanks.

RM
Not sure exactly what is meant by fee, but my understanding of returns in these studies is that they are the cost/expense free returns of asset classes not of mutual funds one actually buys. In that case all costs associated with owning mutual funds need to be deducted and he is allowing 0.5% for that.
DFrank
Posts: 474
Joined: Thu Mar 20, 2014 10:44 pm
Location: North Idaho

Re: Fee of 0.5% for all strategies in simulations

Post by DFrank »

dbr wrote:Not sure exactly what is meant by fee, but my understanding of returns in these studies is that they are the cost/expense free returns of asset classes not of mutual funds one actually buys. In that case all costs associated with owning mutual funds need to be deducted and he is allowing 0.5% for that.
I believe this is the correct interpretation. He seems to always use 0.5% costs in his studies, so I think that is just his default assumption. Obviously, if you run a lower ER your experience will be that much better.

As I said above, I think the most interesting aspect of this latest paper is that all the various withdrawal strategies are analyzed under identical conditions, so it gives you good insight into how they perform compared to one another in identical circumstances.
Last edited by DFrank on Tue Mar 17, 2015 5:04 pm, edited 1 time in total.
Dave
User avatar
LadyGeek
Site Admin
Posts: 95466
Joined: Sat Dec 20, 2008 4:34 pm
Location: Philadelphia
Contact:

Re: Variable spending strategies paper by Pfau

Post by LadyGeek »

I merged etarini's thread into here, which is now in the Personal Finance (Not Investing) forum (retirement spending).

The wiki has some background info: Wade Pfau
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
FinancialDave
Posts: 1819
Joined: Thu May 26, 2011 9:36 pm

Re: Variable spending strategies paper by Pfau

Post by FinancialDave »

pkcrafter wrote:Beliavsky wrote:
Looking at the tables in the paper, no spending strategy stands out as the best to me. Maybe that is the point -- there are trade-offs.
If you think about it, you will realize every investing/withdrawing decision involves trade-offs and compromise. You cannot know what optimal will be in the future.


Paul
I guess I am a little biased with my own analysis:

viewtopic.php?f=10&t=152805&p=2292866#p2292866

When I look at Wade's Table 3 what I see is the RMD as the clear winner, which is the conclusion I came to 3 months ago, if your goal is to extract the most money in retirement without going broke. If your goal is other than the above then the RMD strategy maybe not for you.

Now most people on this forum don't appear to usually be "strapped for cash" but at least IMO this is not the case for at least the average retiree today and in the near future. It appears to me that maximizing their withdrawals may come into play at least sometime during their retirement, even if they don't realize it now.

To combat the low starting rate for the RMD withdrawal I chose to use 3.65% as the minimum rate to use until the RMD tables override that -- no matter what your starting date.

Dave
I love simulated data. It turns the impossible into the possible!
ourbrooks
Posts: 1575
Joined: Fri Nov 13, 2009 3:56 pm

Re: Variable spending strategies paper by Pfau

Post by ourbrooks »

Sigh. I had hoped for a lot better. What this article and others neglect is to provide any measure of the year to year volatility in spending.

It's all well and good to assume that if they have extra money people will find something to spend it on, but that's only true to a very limited extent for people (Bogleheads) with good financial habits. Suppose we have a couple who love gardening, cooking, and listening to classical music. The don't spend much on new plants since their garden is already too crowded; they cook what they grow; and the Internet has far more classical music stations than they could ever listen to. They don't go on vacations much, since they want to look after their garden.

In a bad year, they need all of their investment returns to cover their medical care expenses, taxes, insurance, basic food expenses and utilities. What could they do with the surplus in a good year? Probably, they would "save" it, e.g., not withdraw it. Withdrawal schemes with higher variability, even with higher total spending, don't do them any good; they won't spend it.

They could, of course, follow the 9th strategy and guarantee critical expenses through Social Security, pensions, SPIAs, etc. and just use portfolio withdrawals for optional expenditures, but, if that's the case, the motivation for making regular annual withdrawals is reduced. Why withdraw every year? Why not wait for five years and then do the big kitchen remodel?

It would be really useful if the article characterized each strategy in term of the distribution of year to year variability. A plan with higher total spending may not sound nearly as good if followers must be prepared to deal with 23% year to year swings in the withdrawal amount.
DFrank
Posts: 474
Joined: Thu Mar 20, 2014 10:44 pm
Location: North Idaho

Re: Variable spending strategies paper by Pfau

Post by DFrank »

ourbrooks wrote:Sigh. I had hoped for a lot better. What this article and others neglect is to provide any measure of the year to year volatility in spending.
I think if you look at the 20 and 30 year real spending at the 10th percentile level you will get an idea of spending volatility. Not perfect perhaps, but a useful metric to measure one strategy against another under identical assumptions, which is what is missing in most other analyses.

For a good look at year to year spending volatility you can use a tool such as cFireSim.
Dave
FinancialDave
Posts: 1819
Joined: Thu May 26, 2011 9:36 pm

Re: Variable spending strategies paper by Pfau

Post by FinancialDave »

ourbrooks wrote:Sigh. I had hoped for a lot better. What this article and others neglect is to provide any measure of the year to year volatility in spending.


In a bad year, they need all of their investment returns to cover their medical care expenses, taxes, insurance, basic food expenses and utilities. What could they do with the surplus in a good year? Probably, they would "save" it, e.g., not withdraw it. Withdrawal schemes with higher variability, even with higher total spending, don't do them any good; they won't spend it.

It would be really useful if the article characterized each strategy in term of the distribution of year to year variability. A plan with higher total spending may not sound nearly as good if followers must be prepared to deal with 23% year to year swings in the withdrawal amount.
Year to year variability using a Monte Carlo simulation? Now just how useful would that be? Furthermore, if I could predict the future I could calculate the SD using my analysis, but is that really very useful?

Are you saying your couple is so unskilled after surviving in the world on a low income for 45 years that they don't know how to budget and can't possibly save money in the good years for the bad years. If their starting budget can stand up to a 3.65% withdrawal rate and they have saved the standard 6 months to a year of expenses, they can be in pretty good shape. With the appropriate asset allocation they are most likely not going to suffer 23% swings in their asset valuation - and who cares if its down 23% one year and up 35% the next. Take the RMD, if you don't need the money put it in your cash emergency account, then you have it for the next downturn.

fd
I love simulated data. It turns the impossible into the possible!
ourbrooks
Posts: 1575
Joined: Fri Nov 13, 2009 3:56 pm

Re: Variable spending strategies paper by Pfau

Post by ourbrooks »

FinancialDave wrote:
ourbrooks wrote:Sigh. I had hoped for a lot better. What this article and others neglect is to provide any measure of the year to year volatility in spending.


In a bad year, they need all of their investment returns to cover their medical care expenses, taxes, insurance, basic food expenses and utilities. What could they do with the surplus in a good year? Probably, they would "save" it, e.g., not withdraw it. Withdrawal schemes with higher variability, even with higher total spending, don't do them any good; they won't spend it.

It would be really useful if the article characterized each strategy in term of the distribution of year to year variability. A plan with higher total spending may not sound nearly as good if followers must be prepared to deal with 23% year to year swings in the withdrawal amount.
Year to year variability using a Monte Carlo simulation? Now just how useful would that be? Furthermore, if I could predict the future I could calculate the SD using my analysis, but is that really very useful?

Are you saying your couple is so unskilled after surviving in the world on a low income for 45 years that they don't know how to budget and can't possibly save money in the good years for the bad years. If their starting budget can stand up to a 3.65% withdrawal rate and they have saved the standard 6 months to a year of expenses, they can be in pretty good shape. With the appropriate asset allocation they are most likely not going to suffer 23% swings in their asset valuation - and who cares if its down 23% one year and up 35% the next. Take the RMD, if you don't need the money put it in your cash emergency account, then you have it for the next downturn.

fd
Certainly, you can calculate year to year variability using a Monte Carlo simulation. In fact, if all you're looking at are means in the simulation, you're probably making serious interpretation errors. It would be nice to have a statement for each strategy that, based on 10 zillion simulation runs, the average year to year change in withdrawals for strategy X was W.

If my hypothetical couple knew how much money to save in the good years for the bad years, they wouldn't need to read Wade Pfau's paper because they'd already have a perfect withdrawal strategy! In fact, the whole idea of "saving" from retirement withdrawals is a bit of a fallacy. You already have the money in your retirement savings so you can't be saving it again. What you're really doing by "saving" during the good years is (a) premature withdrawals from tax deferred accounts and (b) changing your asset allocation to increase the cash percentage. There are some other threads recently active that show that getting up much above 5% cash has serious negative effects on portfolio survivability so the net result of "saving" is probably negative. A better strategy would be not to withdraw the money at all until they needed it but then they'd have to have some algorithm for determining how much they could safely take out. (Another paper, Prof. Pfau?)

What's an "appropriate" asset allocation for not seeing large withdrawal swings? Suppose that you had a conservative, 60/40 portfolio. In 2008, the stock market went down by 39%; that results in a 23% drop in any withdrawal strategy that takes a percentage of current assets. A 50/50 portfolio still results in a 20% drop.

When all is said and done, my hypothetical couple are probably best off with the old fashioned, constant dollar, "safe withdrawal" strategy approach. It actually gives them the most spendable money since it gives them an amount on which to do longer term planning ("yes, we can afford the house with the bigger garden.") My hypothetical couple is probably a bit of an exaggeration, but they're close enough to real people I know to make me believe that year to year variability is an important consideration in choosing a withdrawal scheme.
DFrank
Posts: 474
Joined: Thu Mar 20, 2014 10:44 pm
Location: North Idaho

Re: Variable spending strategies paper by Pfau

Post by DFrank »

ourbrooks wrote:If my hypothetical couple knew how much money to save in the good years for the bad years, they wouldn't need to read Wade Pfau's paper because they'd already have a perfect withdrawal strategy! In fact, the whole idea of "saving" from retirement withdrawals is a bit of a fallacy. You already have the money in your retirement savings so you can't be saving it again. What you're really doing by "saving" during the good years is (a) premature withdrawals from tax deferred accounts and (b) changing your asset allocation to increase the cash percentage. There are some other threads recently active that show that getting up much above 5% cash has serious negative effects on portfolio survivability so the net result of "saving" is probably negative. A better strategy would be not to withdraw the money at all until they needed it but then they'd have to have some algorithm for determining how much they could safely take out. (Another paper, Prof. Pfau?)
I think there is confusion here over what it means to save during retirement. During retirement, saving means not spending. The only reasons I would sell an investment and move it to cash are: 1) If I planned to spend it or 2) If I needed to move it to cash to achieve my target asset allocation. I may need to sell an investment above and beyond my spending level to meet an RMD requirement, but in that case the proceeds will be reinvested in an after tax account in accordance with my asset allocation target, not put into cash. I may be following a variable spending strategy that tells me that based on the great market returns I can spend more this year, but if I choose to not spend that money that just means it sits in my investment accounts for another time.
ourbrooks wrote:What's an "appropriate" asset allocation for not seeing large withdrawal swings? Suppose that you had a conservative, 60/40 portfolio. In 2008, the stock market went down by 39%; that results in a 23% drop in any withdrawal strategy that takes a percentage of current assets. A 50/50 portfolio still results in a 20% drop.
In this paper Dr. Pfau specifically focused on comparing the outcomes of various withdrawal strategies under identical conditions. For the purposes of this analysis he chose a 50/50 asset allocation, which I think is close enough to an average retiree's allocation to produce meaningful results. He has written other papers in which he varies asset allocation and measures the impact of different allocations on spending outcomes. I think if he had added asset allocation as a variable across 10 different withdrawal strategies he probably would have produced more confusion than clarity.
ourbrooks wrote:When all is said and done, my hypothetical couple are probably best off with the old fashioned, constant dollar, "safe withdrawal" strategy approach. It actually gives them the most spendable money since it gives them an amount on which to do longer term planning ("yes, we can afford the house with the bigger garden.") My hypothetical couple is probably a bit of an exaggeration, but they're close enough to real people I know to make me believe that year to year variability is an important consideration in choosing a withdrawal scheme.
Maybe yes, maybe no. What the analysis shows is that the constant dollar safe withdrawal strategy allows an initial withdrawal rate of 2.85% under the condition that you want a 90% probability that the withdrawal will never go below 1.5% of the initial portfolio value ($1500 from a $100k portfolio). The result is constant spending in real dollars of $2850 per year for 30 years. If that spending level meets your couple's spending requirement then that may be the right approach for them.

However, what the paper also shows is that if they are willing to have some spending variability, they could spend as much as 4.95% of the initial portfolio value while maintaining the same probability that spending will fall below $1500 in 30 years. The spending results at 10, 20 and 30 years give you some idea what the volatility might be. As an example, at 30 years, 50% of the time spending will be $3200 real or greater, but 10% of the time it may be below $1580. I think it's a useful metric to understand how the different withdrawal strategies compare to each other under identical conditions.

Since most people will have a component of their desired spending that is discretionary (e.g. we can wait to build the greenhouse until we have a better year), they may be willing to deal with some variability in return for higher initial/average/overall spending during their retirement. Many variable spending strategies allow you to set a spending floor to ensure at least non-discretionary spending needs are met. One can also design hybrid strategies where a floor is met through pensions/social security/annuities etc., and a variable withdrawal strategy is used to manage the balance of the portfolio.

As suggested earlier, if you want to look at year to year spending variability in detail head over to cFireSim and set up a scenario of your choosing and the program will show you a graph of how spending changes over time for all historical market return/inflation sequences that they use in their analysis. It also will allow you to set different asset allocations, to set spending floor/ceiling and so forth to reflect your individual situation. In addition to reports on spending outcomes the report generates statistics about min/average/max spending levels, biggest change in spending level, biggest change in portfolio value, and other statistics. It's quite a powerful tool.

None of these papers is ever going to answer every question, particularly when the topic is as complex as variable spending strategies. I think the paper provides a useful comparison of different withdrawal strategies, and serves as a basis for identifying those that might meet your needs so that you can go on to do your own analysis based on your individual objectives and situation.
Dave
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: Variable spending strategies paper by Pfau

Post by siamond »

DFrank wrote:
ourbrooks wrote:Sigh. I had hoped for a lot better. What this article and others neglect is to provide any measure of the year to year volatility in spending.
I think if you look at the 20 and 30 year real spending at the 10th percentile level you will get an idea of spending volatility. Not perfect perhaps, but a useful metric to measure one strategy against another under identical assumptions, which is what is missing in most other analyses.

For a good look at year to year spending volatility you can use a tool such as cFireSim.
Yes, I agree that std-deviation was sorely missing from the analysis. And also the slope of the withdrawal trajectory. Although I was pleased to see the 3x3 matrix on the withdrawal trajectory (percentiles, 10/20/30 years), this does provide a lot of information that we mostly miss by just looking at the average or the minimal spend overall. Also, some analysis on how the withdrawal method adjusts itself depending on how good or bad your assumption about expected returns turns out to be (a key input parameter for most methods, hidden or explicit) would have been welcome.

As to VPW/PMT, what the author analyzed is something similar and yet different. He used a PMT formula driven the 10yrs treasury rate as expected returns rate. Which really makes little sense until you're 100% bonds! I know he was reacting to another research paper, but still, it would have been good to see something more realistic.

Finally, I would love to see more conclusions being drawn, some mapping between goals and a choice of withdrawal method. There is no one-size-fit-all, that we know, but having some kind of decision-tree based on goals would have been terrific! There is no shortage of follow-up work, in other words. Still, this is a great start, I'm glad to see a researcher as good as Pr Pfau to take on this important topic, and to define a framework to try to compare apples to apples. I'm not so sure I would use this XYZ approach for such comparison though - need to think more about it!
FinancialDave
Posts: 1819
Joined: Thu May 26, 2011 9:36 pm

Re: Variable spending strategies paper by Pfau

Post by FinancialDave »

When all is said and done, my hypothetical couple are probably best off with the old fashioned, constant dollar, "safe withdrawal" strategy approach. It actually gives them the most spendable money since it gives them an amount on which to do longer term planning ("yes, we can afford the house with the bigger garden.") My hypothetical couple is probably a bit of an exaggeration, but they're close enough to real people I know to make me believe that year to year variability is an important consideration in choosing a withdrawal scheme.
The most spendable money during retirement cannot certainly come from a real constant dollar approach, unless you figure the couple is going to die early and you use a rate above the 4% norm. All the studies clearly show that more money is withdrawn from a variable strategy. But in most cases a starting plan would be based on constant dollars, as I'll explain below.

Just because your portfolio goes down 23% does not mean you are eating rice and beans if you have a sound retirement plan. In the first place close to half of your retirement income should be coming from SS + either pensions or a guaranteed annuity - these income streams will not drop. That leaves you with a mere 12% drop in income which most anyone could handle without resorting to drastic measures. In fact if you are using some type of bucket approach like I do, then you have a year or two of essentially cash in this bucket to ride through the above with no problem.

As far as the excuse that you don't know how much to save for the down years - it does not need to be a fixed number, you save the excess beyond your retirement plan, which you must have or you are just defeated before you start.

William Bernstein in his book the "The Investor's Manifesto" (thank you Taylor!) put's it this way (I'm paraphrasing a bit) -- if you try and withdraw 4% inflation adjusted you are working on the edge, 3% gives you are decent cushion, and if you only need 2% you can do just about anything. That's why in my RMD plan the withdrawal rate stays a constant 3.65% until age 71.

fd
I love simulated data. It turns the impossible into the possible!
ourbrooks
Posts: 1575
Joined: Fri Nov 13, 2009 3:56 pm

Re: Variable spending strategies paper by Pfau

Post by ourbrooks »

So let me re-assert my previous claim: the most spendable money for my hypothetical couple would be from constant dollar withdrawals.

Yes, in theory, variable withdrawal schemes would permit higher total dollar withdrawals. In practice, though, that won't happen. To reduce volatility, the couple must invest half of their retirement assets in non-volatile vehicles. That's money they definitely won't have available to invest in higher return vehicles such as stocks so the higher withdrawals must start with half as much money.

Second, on what amounts should the couple base their retirement plans? Certainly, not on the potentially higher amounts available through variable withdrawals, because those extra amount might not be available. Suppose they follow William Bernstein's advice and base their budget on 3% constant dollar withdrawals. Remember that that 3% rate is based on both up and down years. Suppose that the market is up this year by 7%. Does it mean that this year they've got excess or was that up year already baked into the 3% rate? Without some guidance, many people will simply not spend the money and leave it in a money market account or cash.

"Saving" the money that way is, as I mentioned above, not a good idea. What you're withdrawing from already is savings. You're just changing your asset allocation and, maybe, incurring an unnecessary tax penalty on the way. People would be better off not withdrawing excess money until they actually needed it.

What people could use more guidance with is determining the amounts available for lump sum withdrawals. Many retirees have large expenses such as new roofs, new furnaces, etc. These large expenses often can include a large discretionary component - metal versus composition roof, new kitchen cabinets versus just the counter tops, etc. Even people following constant dollar withdrawal strategies usually find themselves ahead of plan; after 30 years, people following a constant withdrawal plan have a median value of 80% of their original amount left. If five years have gone by and it's time for the new roof, how much can they afford to splurge? What's the formula?
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: Variable spending strategies paper by Pfau

Post by siamond »

I have troubles to understand the Guyton-Klinger numbers (as I just played at length with this withdrawal method myself), my own findings are that it is really quite hard to make it fail (i.e get the spending and/or portfolio to go under unacceptable levels) when running historical backtesting. It is possible to make it fail if you pick a very unreasonable IWR (Initial Withdrawal Rate), e.g. 7%, but then the paper didn't show such high number in association with the dire case (10% chance of low spend), more an IWR around 5%. Which is not historically unreasonable for a 50/50 US-only AA.

Note that, if I got it right, the general idea of the model is to increase the initial rate until the Monte-Carlo simulation reaches a state where 10% of the outcomes are bad enough (e.g. $1500 spend by year 30). I have significant reservations about such methodology, this seems quite misleading, but let's park that for now.

Let's just focus on the XYZ methodology in combination with Monte-Carlo testing and today's situation on future returns (cf. Appendix A). Something looks amiss to me. Yes, today's situation isn't pretty for expected returns, notably for a US-only portfolio, given the high valuation of equities and the historically low bond yields. But then, it seems that the methodology kind of double-counts the issue. First, by assuming a trajectory of returns starting low (Appendix A median values), then slowly back to historical numbers. Next, by adding a Monte-Carlo layer centered on such median trajectory, which means that bad outcomes (the 10% chance of the XYZ methodology) will get, well, REALLY BAD. Actually, unrealistically bad. Fact is today's situation is probably ALREADY in the bottom 10% to 20% cases, so it doesn't seem to make sense to calibrate Monte-Carlo assuming this is a median, it is simply NOT.

Am I missing something? If not, this would probably explain the strange G-K numbers (and the same is true for all withdrawal methods).
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Re: Variable spending strategies paper by Pfau

Post by wade »

Hi all, I'm glad to see this thread come alive with lots of interest comments. Let me add my 2 cents on some of these issues:

Fees: I assume a 0.5% annual fee. I always struggle with that. Bogleheads can spend less, but many will spend more. Practically, this can be viewed as underperformance relative to the underlying indices. As an estimate, if you think a 0% fee is reasonable, you can probably add about half of the fee (0.25%) to the withdrawal rates.

Volatility of Spending: Thank you FinancialDave, I agree with your explanations. I played around with adding a spending volatility measure, but ultimately decided against it. Basically, strategies which allow greater spending on the upside will have a bigger standard deviation. That was all the measure was picking up. I did stick with one asset allocation (50/50) for the paper just because it would be too much to try add more asset allocation scenarios when there is already so much going on with the variable spending. Naturally, you do not have to spend as much as the strategy suggests, but the amount indicated is what the strategy determines as feasible. If you have no plan to spend upside and really want spending to be very stable on the downside, you should really consider a bigger role for individual bonds and income annuities. Constant spending from a volatile portfolio creates sequence risk. It offers more upside potential, but also more downside risk. That is why spending needs to be conservative... to self-manage the market and longevity risks. If you really want stable spending, don't try to completely self-manage those risks.

Big Lump-Sum payments: Ourbrooks, on the issue of wanting to take out occasional large amounts to cover big expenses, that is really an issue which the PMT methods are designed to help out. If you want to know the impact of an extra large withdrawal, run the PMT formula on what will then be the lower remaining account balance. It can give you a good idea about the impact of a large withdrawal on subsequent sustainable spending rates. Decision rule methods aren't really designed to deal with those contingencies, other than to do a start-over with the strategy from that point in retirement after the large expense is made.

Monte Carlo simulations: siamond, I do not think I am double-counting any sort of bad luck by assuming lower median returns in the near-term horizon. This is a function of the lower interest rates today. Interest rates are already low, so the "expected returns" are low. That's the reality. Monte Carlo simulations fluctuate from where we actually are, starting out today.

I'm about to board a plane, and I need to review the thread again to see if there are more comments to address, but I hope this can provide some responses to keep the conversation going. Siamond, about your concerns with the Guyton-Klinger method, there could be some possibility that I've made a calculation error, but I hope not.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Re: Variable spending strategies paper by Pfau

Post by wade »

I've got a few more minutes...

Siamond on the issue of the spending trajectory implied by the strategy, this really was my purpose in showing the distribution of spending at 10, 20, and 30 years into retirement. You can see the slope from those numbers. And it is important to realize there is not a fixed slope, as it depends on where you are in the distribution. For the purposes of understanding the basic slope, i.e. will spending tend to increase or decrease over retirement, you'd probably want to observe the trend with the median spending numbers.

About using 10-year Treasuries for the PMT example. Indeed, there are a million ways just to use the PMT formula. The higher the interest rate you use, the more you spend today, but the more likely you are to subsequently experience a decline in future spending. A lower discount rate is more conservative, and it isn't incompatible with actually using a different asset allocation. They are really two different decisions, and again, the interaction of those decisions will impact the spending trajectory over retirement.

On that issue of Guyton and Klinger again, there definitely can be failure with this strategy. Even in their initial research articles they were getting failure. The portfolio can decline faster than what those 10% spending cuts can keep up with.
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: Variable spending strategies paper by Pfau

Post by siamond »

wade wrote:Monte Carlo simulations: siamond, I do not think I am double-counting any sort of bad luck by assuming lower median returns in the near-term horizon. This is a function of the lower interest rates today. Interest rates are already low, so the "expected returns" are low. That's the reality. Monte Carlo simulations fluctuate from where we actually are, starting out today.

I'm about to board a plane, and I need to review the thread again to see if there are more comments to address, but I hope this can provide some responses to keep the conversation going. Siamond, about your concerns with the Guyton-Klinger method, there could be some possibility that I've made a calculation error, but I hope not.
It seems quite likely that the mismatch on the numbers is due to this combination of today's dire situation, plus Monte-Carlo and 10% worst outcomes. I wasn't suggesting a calculation error, although it is always good to see such math performed by two ways. My own math was based on historical returns (possibly weighed by a drag parameter).

Sure, today's reality is dire for expected returns, this is the reality, we just had a long thread on Dr Bernstein latest assessment of a 2% return on 50/50 portfolio for the coming decade (or maybe 20 years). But using this situation as the new median really doesn't seem quite right. There MAY be a 'new normal' where average returns are lower than in the past century, this is really hard to say, although I am ready to believe some of that. But I am not ready to believe that this 'new normal' should be equated to today's situation as a median, with the full historical std-deviation around it. We are in a rather extreme (and historically unique) situation. I suspect you may be conflating two things. A possible 'new normal' vs today's dire expected returns. Not the same thing.

Personally, I would approach things a tad differently. Forget where we are today, anyway, it will be different tomorrow. Take historical returns and feed relevant parameters in Monte-Carlo and the XYZ methodology. See what goes with the various withdrawal methods. Then subtract a drag to the average historical returns (say by 0.5% steps; separate steps on equities from steps on bonds). Run Monte-Carlo again. Optionally, do the same with rosier returns (hey, who knows?). In other words, perform a sensibility analysis to test some possible 'new normals'. Which would have a lot of value per se on how well various withdrawal methods react to such possible futures. Then coming back to today's dire expected returns is simple, one could assume we may be in the 10% to 20% worst cases, so this maps well to your XYZ methodology. Then not only will you have results that remain valid a few years from now, but also better information on the dynamics of variable withdrawal methods, and... no double-counting! :wink:

PS. and yes, G-K can fail, that I agree. It just takes more than a 5% IWR to do so. I had to go to 7% with historical returns to make it happen. Which is of course a totally unreasonable IWR for a 50/50 portfolio.
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: Variable spending strategies paper by Pfau

Post by siamond »

wade wrote:Siamond on the issue of the spending trajectory implied by the strategy, this really was my purpose in showing the distribution of spending at 10, 20, and 30 years into retirement. You can see the slope from those numbers. And it is important to realize there is not a fixed slope, as it depends on where you are in the distribution. For the purposes of understanding the basic slope, i.e. will spending tend to increase or decrease over retirement, you'd probably want to observe the trend with the median spending numbers.
Yes, agreed. It was really good to see you display this 10/20/30 years decomposition. I know I learned a lot by looking at numbers with such tiers with my own modeling. You still need to take a much closer look at std-deviation though. Things can get remarkably hectic with some variable methods, and averages tend to hide the matter. By the way, is your 10 years number the average of the first decade, or the spend on year#10? The former, I hope? Then what about the 20 years number? The average of year 11 to 20? Or else?
wade wrote:About using 10-year Treasuries for the PMT example. Indeed, there are a million ways just to use the PMT formula. The higher the interest rate you use, the more you spend today, but the more likely you are to subsequently experience a decline in future spending. A lower discount rate is more conservative, and it isn't incompatible with actually using a different asset allocation. They are really two different decisions, and again, the interaction of those decisions will impact the spending trajectory over retirement.
Yes, agreed, using a PMT formula with a rate higher or lower than the expected return from your AA is a perfectly reasonable approach, depending on your goals. It's just that using the treasury interest rate would be rather extreme. And then, it should display quite the upwards slope in the results. By the way, how does the PMT test in Table 2 & 3 work? Did you use the initial rate as a fixed rate in the PMT formula? Or did you factor in treasury rates somehow?
User avatar
LadyGeek
Site Admin
Posts: 95466
Joined: Sat Dec 20, 2008 4:34 pm
Location: Philadelphia
Contact:

Re: Variable spending strategies paper by Pfau

Post by LadyGeek »

FYI - Forum member wade is Wade Pfau, the paper's author.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
FinancialDave
Posts: 1819
Joined: Thu May 26, 2011 9:36 pm

Re: Variable spending strategies paper by Pfau

Post by FinancialDave »

ourbrooks wrote:So let me re-assert my previous claim: the most spendable money for my hypothetical couple would be from constant dollar withdrawals.

Second, on what amounts should the couple base their retirement plans? Certainly, not on the potentially higher amounts available through variable withdrawals, because those extra amount might not be available. Suppose they follow William Bernstein's advice and base their budget on 3% constant dollar withdrawals. Remember that that 3% rate is based on both up and down years. Suppose that the market is up this year by 7%. Does it mean that this year they've got excess or was that up year already baked into the 3% rate? Without some guidance, many people will simply not spend the money and leave it in a money market account or cash.
Just like the 4% rule, which is designed to survive most 30 year up and down markets, your couples 3% rule would (IMHO) not have a problem surviving in most any asset allocation that has at least 40% bonds, and maybe even less. The up years ARE "baked into the cake" as you say, if you design a plan based on 3% withdrawal, then it does not mean you take excess money out just because the market is up, nor do you need to not withdraw the 3% when the market is down.

Most people start BEFORE retirement so they can have enough money to satisfy their budget - that is the proper way to do it. But as we all know, some people get to retirement and say "how much can I spend." This is where you have to assess their risk tolerance and decide, what will this risk tolerance support. Depends on how much of the "3 legs" of the retirement stool they have (SS, pensions & annuities, & other savings such as IRA, Roth, & taxable accts.) Of the first two legs, these are pretty much guaranteed, but not all in constant, inflation adjusted, dollars. The size of the 3rd leg (personal investments) compared to the other two, determines the riskiness of their investments. Suppose a couple has $60k in combined SS & inflation adjusted pensions from the government and only $150k in retirement savings accounts. As a worst case I use the 4% rule as a conversion between lump sum and yearly income. So a $150k lump sum retirement will produce roughly $6000 a year, while they are also getting $60k guaranteed. So if the market crashes they have little chance of needing to eat "rice and beans," because they have lost less than 10% of their income, so no matter how much risk they add to their personal investments it won't materially affect their retirement. However, as you should be able to see the more the size of personal investments grow, the better off they are, but the risk increases, so you may want to dial-down the equity allocation. Many people have written books on the process (including the Bogleheads) but I just start with these numbers to know if someone is "out in left field" thinking they can withdraw $50k per year from an account of say $300k when they should be only withdrawing $12k. I don't want to over-simplify the process, but it is not that complex either.

What people could use more guidance with is determining the amounts available for lump sum withdrawals. Many retirees have large expenses such as new roofs, new furnaces, etc. These large expenses often can include a large discretionary component - metal versus composition roof, new kitchen cabinets versus just the counter tops, etc. Even people following constant dollar withdrawal strategies usually find themselves ahead of plan; after 30 years, people following a constant withdrawal plan have a median value of 80% of their original amount left. If five years have gone by and it's time for the new roof, how much can they afford to splurge? What's the formula?
The answer to this is very simple, which I mention in my first post - it is a budget! Other people use fancy names like "sinking funds" which just means you have a savings account that you fund monthy (or you pre-fund it from some lump sum) that will cover "maintenance" and one-time expenses. You do NOT put this money in investment accounts that are subject to market losses, you put this money in your money market earning 1% PERIOD. Other people keep two years of cash in their accounts and use this for what they call emergencies (which really aren't because most are predictable.) It's all a matter of personal preference on what works for you.

fd
I love simulated data. It turns the impossible into the possible!
DFrank
Posts: 474
Joined: Thu Mar 20, 2014 10:44 pm
Location: North Idaho

Re: Variable spending strategies paper by Pfau

Post by DFrank »

ourbrooks wrote:What people could use more guidance with is determining the amounts available for lump sum withdrawals. Many retirees have large expenses such as new roofs, new furnaces, etc. These large expenses often can include a large discretionary component - metal versus composition roof, new kitchen cabinets versus just the counter tops, etc. Even people following constant dollar withdrawal strategies usually find themselves ahead of plan; after 30 years, people following a constant withdrawal plan have a median value of 80% of their original amount left. If five years have gone by and it's time for the new roof, how much can they afford to splurge? What's the formula?
Regardless of your withdrawal strategy, I think the answer here is to put such expenses in your retirement budget. As an example, if you plan to buy a car every 10 years, include 1/10th of what you typically spend on a car in your retirement spending forecast. Obviously, you wouldn't spend that money each year, but you could set it aside in a different account if you wanted. Similar things can be done for other infrequent but significant expenses. There are tools available that can give you some guidance on this subject, particularly with respect to lifetime of major home maintenance items.

Another approach is found in the Analyze Now retirement planner developed by Bud Hebeler. One of his worksheets allows you to forecast these sorts of expenses by year through your entire retirement period. These expenses are accounted for in the analysis, but not included in your 'routine' annual spending.
Dave
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: Variable spending strategies paper by Pfau

Post by siamond »

Irrespective of the calibration of the Monte-Carlo simulation, I still have some troubles to wrap up my mind around the XYZ methodology.

The author stated:
XYZ Formula = Client Willingly Accepts an X% probability that spending falls below a threshold of $Y (in inflation-adjusted terms) by year Z of retirement.
For instance, instead of accepting a 10% chance for failure within the first 30 years of retirement, an XYZ rule could be that the client accepts a 10% chance that their spending level falls below an inflation-adjusted $60,000 by the 30th year of retirement.
Is it really assessed on T0+29 (the withdrawal for the 30th year), or is it assessed for all years over the entire duration of the retirement period? With some variable withdrawal methods, one can suffer from quite an income drop on a bad year, which could be troublesome, but this isn't necessarily the 30th year, the market (and portfolio balance) may have recovered by then, and the withdrawal be back in shape.
User avatar
wade
Posts: 612
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Re: Variable spending strategies paper by Pfau

Post by wade »

Siamond, thanks for the good comments.

About the XYZ rule, I'm specifically calculating it with regard to the year 30 spending. You are right that spending could dip lower in some years and then come back up by year 30. I thought about considering whether spending fell below the threshold in any year up to year 30. I didn't ignore this issue. But I don't think there are many cases where spending dips lower and then comes back up, and decided to just go with the more straightforward way to calculate and explain things. Explaining the alternative is a bit tougher. It would be something like: the retiree accepts that spending might fall below the threshold up to 10% of the time in the first 30 years of retirement. I think that's less clear for readers.

About the year 10 spending, year 20, etc. That is just for those years. It's not an average of all the years up to that point. Your suggestion is another viable way to do things, but it's not what I did.

About the issue of median returns. One could debate all day about whether stock returns will be lower in the future or not. But there isn't much to debate about bonds. When you buy a bond with a low yield and hold it to maturity, your return will be low. Today's yield serves as something very close to it's median return. More generally, I am actually simulating future bond yields now. And bond and stock returns are driven from those bond yield simulations. Because I have future simulated bond yields, when I use the PMT formula, I am able to plug in future bond yields for each year of each simulation.
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: Variable spending strategies paper by Pfau

Post by siamond »

Hi Wade, thanks for coming back to me. I appreciate your patience and eagerness to discuss.

I looked at PMT-based and G-K methods quite in depth, using historical backtesting under various conditions, and looking at many individual cycles. And it is very clear that those methods can go through up and downs that the very last year of the retirement period is NOT representative of. So yes, I would strongly encourage you to look at metrics (average, min, slope and std-deviation) being more representative of the full trajectory (as a whole and per tier), not just points in time like year#10, #20, #30. I appreciate your concern for simplicity, but here we're just losing too much of the important data. Please take a look at various exemplary scenarios I posted on this thread to get my point:
http://www.bogleheads.org/forum/viewtop ... &start=100

About the calibration of future returns, I don't know, it seems that I have troubles explaining my view. I do agree that the bond yield is a good predictor of expected returns in the coming decade, and this isn't pretty nowadays. But it's no median that Monte-Carlo should be calibrated around, and then a 10% chance of failure be applied to. Adding another full layer of historical std-deviation around today's situation just doesn't seem to add up - I mean, do you seriously entertain the possibility of bond yields to go down 2% to 3% (a single SD unit) from where they are now? I am repeating myself, sorry, I just don't know how to explain the issue better. And maybe it's just me not fully understanding your way of modeling. I just don't see a way around doing sensitivity analysis on future returns assumptions.

Finally, something else nags me about this XYZ methodology. Maybe it's just the examples that you took in Table 2 and 3. I would never go for a 10% chance for my spending to go that low, and/or for my portfolio balance to go close to zero. 10% is a big number. And an even bigger number than one might think, because one might often decide to retire when his/her portfolio looks good (I reached my number!), hence at high valuation, hence at higher risk of troubles to come early in retirement. Also, this 10% number calibrates everything else, and notably the initial withdrawal rate parameter, and this seems somewhat skewed logic. Maybe I have troubles getting out of my sensitivity analysis mindset, but here again, I would take various initial rates as AN INPUT, and see what goes, looking at all percentiles (good or bad) in the results. This being said, admittedly, maybe (not sure!) I am simply looking for lower value of "X", a higher value of "Y" for the spend, combined with a higher lower cap for the portfolio balance than zero. Maybe we need an X/YS/YP/Z method if you see what I mean? As to the "Z" parameter, do this mean more than the max duration of the retirement period? I suspect so, but I find hard to see a practical use otherwise.
Dandy
Posts: 6701
Joined: Sun Apr 25, 2010 7:42 pm

Re: Variable spending strategies paper by Pfau

Post by Dandy »

All withdrawal strategies are stating points. For most it would be foolish to think you could pick a strategy that would serve you best and not adjust it for 25 or30 years - or that you would be competent to implement it for all that time. It could serve you well but that is very unlikely in my opinion. There are just too many variables to feel you have a set it and forget it strategy for life.

Every few years you need to look at your health, portfolio, expenses, etc and see if an adjustment is warranted. While it would be foolish to make panic moves it might also be foolish to just trust back tested theories when facing a harsh reality. Minor adjustments in spending reflected in withdrawals, made early, can make a big difference. Prudence vs panic.
FinancialDave
Posts: 1819
Joined: Thu May 26, 2011 9:36 pm

Re: Variable spending strategies paper by Pfau

Post by FinancialDave »

siamond wrote:Hi Wade, thanks for coming back to me. I appreciate your patience and eagerness to discuss.

I looked at PMT-based and G-K methods quite in depth, using historical backtesting under various conditions, and looking at many individual cycles. And it is very clear that those methods can go through up and downs that the very last year of the retirement period is NOT representative of. So yes, I would strongly encourage you to look at metrics (average, min, slope and std-deviation) being more representative of the full trajectory (as a whole and per tier), not just points in time like year#10, #20, #30. I appreciate your concern for simplicity, but here we're just losing too much of the important data. Please take a look at various exemplary scenarios I posted on this thread to get my point:
http://www.bogleheads.org/forum/viewtop ... &start=100

About the calibration of future returns, I don't know, it seems that I have troubles explaining my view. I do agree that the bond yield is a good predictor of expected returns in the coming decade, and this isn't pretty nowadays. But it's no median that Monte-Carlo should be calibrated around, and then a 10% chance of failure be applied to. Adding another full layer of historical std-deviation around today's situation just doesn't seem to add up - I mean, do you seriously entertain the possibility of bond yields to go down 2% to 3% (a single SD unit) from where they are now? I am repeating myself, sorry, I just don't know how to explain the issue better. And maybe it's just me not fully understanding your way of modeling. I just don't see a way around doing sensitivity analysis on future returns assumptions.

Finally, something else nags me about this XYZ methodology. Maybe it's just the examples that you took in Table 2 and 3. I would never go for a 10% chance for my spending to go that low, and/or for my portfolio balance to go close to zero. 10% is a big number. And an even bigger number than one might think, because one might often decide to retire when his/her portfolio looks good (I reached my number!), hence at high valuation, hence at higher risk of troubles to come early in retirement. Also, this 10% number calibrates everything else, and notably the initial withdrawal rate parameter, and this seems somewhat skewed logic. Maybe I have troubles getting out of my sensitivity analysis mindset, but here again, I would take various initial rates as AN INPUT, and see what goes, looking at all percentiles (good or bad) in the results. This being said, admittedly, maybe (not sure!) I am simply looking for lower value of "X", a higher value of "Y" for the spend, combined with a higher lower cap for the portfolio balance than zero. Maybe we need an X/YS/YP/Z method if you see what I mean? As to the "Z" parameter, do this mean more than the max duration of the retirement period? I suspect so, but I find hard to see a practical use otherwise.
siamond,
You make all very good points which I believe stem from the short comings of Monte Carlo simulations. They are not real!!!! Thus are subject to the constraints the author puts on them. What I think most don't realize is that because of the interaction of the variables, it is usually quite possible that less than 10% of the solutions could even realistically happen.

Now of course we all know that usually something will happen that we can't predict or has never happened before, but I have never been convinced it is going to fall outside the bounds of something that hasn't happened over the last 115 years, that's why I think it is better to do these simulations using real data, which is what I do, either Schiller data or the data that Jim Otar uses (being careful to note some shortcomings in the Otar retirement calculator that I have uncovered.)

fd
I love simulated data. It turns the impossible into the possible!
Professor Emeritus
Posts: 2628
Joined: Mon Aug 13, 2012 6:43 am

Re: Variable spending strategies paper by Pfau

Post by Professor Emeritus »

wade wrote:I've got a few more minutes...

Siamond on the issue of the spending trajectory implied by the strategy, this really was my purpose in showing the distribution of spending at 10, 20, and 30 years into retirement. You can see the slope from those numbers. And it is important to realize there is not a fixed slope, as it depends on where you are in the distribution. For the purposes of understanding the basic slope, i.e. will spending tend to increase or decrease over retirement, you'd probably want to observe the trend with the median spending numbers.
Dear Wade

I believer your "annuitize the floor and have more discretionary spending" is the closest to the three part retirement spending analysis I advocated in a different thread viewtopic.php?f=10&t=163545
I divided spending into

1) Routine recurring expected etc. (often called necessities but let's not get into that debate. It includes capital goods that must be replaced such as cars. ) Living in y a home you own is in this category. Any reduction in this level is painful
2) nonrecurring discretionary expenses (sometimes called "luxuries)
3) expenditures required to avoid or cope with disastrous or life altering events (medical care, LTC, local pubs closing)

Income or asset liquidation has to match these expenditures.

Sometimes you can buy insurance to shift expenditures from category 3 to category 1.
I will suggest that for a bogle retirement approach

Category 1 expenditures should be covered by SS pension or Life long annuities plus home ownership if applicable.
Category 2 expenses are rationally paid out of a modifiable fairly high SWR from a balanced portfolio
Category 3 should be "ring fenced" secure investments matched to the potential needs

I believe that risk of Category 3 expenses are as large a motivator as any other. However analysis is complicated by social programs such as medicaid. Annuities are irrational for anyone who believes they will need medicaid for Long term care. Medicaid also makes some assets "more equal " than others.

Your analysis works quite well for people who are far above the medicaid threshold or are fully insured for long term care.
heyyou
Posts: 4453
Joined: Tue Feb 20, 2007 3:58 pm

Re: Variable spending strategies paper by Pfau

Post by heyyou »

Asset spending is not a one shot, irrevocable decision. You have SS and you can choose to spend some on annuities when the rates are attractive, then use SWR on the remainder of the portfolio.

Consider diversifying WD rates by using a fixed or variable rate with portfolio performance tests while you are cognizant in the first half of retirement, then start a 15 year WD plan with a fixed percentage or the VPW for the second half.

Wade, thank you for mentioning the two David Zolt papers. My search here at BHs did not show any mention of them.

Zolt's annual comparison of the WD amount, to the remaining portfolio balance seems like an excellent test. For those who haven't read Zolt, if your proposed WD (last year's WD plus inflation) represents a larger percentage of the remaining portfolio than the initial WD rate, you do not get that annual inflation increase, you only get last year's WD amount. Zolt's premise is that a retiree does not ever want a reduction in nominal spending, but will tolerate a zero increase for the previous year's inflation. He acknowledges the risk of long term loss of real spending for the tradeoff of an extra one percent more spending in the early years of retirement. Seems like a good path since it is buffered by SS (w/COLA) and other pension or annuity income.
Professor Emeritus
Posts: 2628
Joined: Mon Aug 13, 2012 6:43 am

Re: Variable spending strategies paper by Pfau

Post by Professor Emeritus »

heyyou wrote:Asset spending is not a one shot, irrevocable decision. You have SS and you can choose to spend some on annuities when the rates are attractive, then use SWR on the remainder of the portfolio.
.
Isn't this just market timing interest rates? How is it different from Buy stocks when they are low?
heyyou
Posts: 4453
Joined: Tue Feb 20, 2007 3:58 pm

Re: Variable spending strategies paper by Pfau

Post by heyyou »

Professor Emeritus » Sun Apr 19, 2015 2:56 pm

heyyou wrote:
Asset spending is not a one shot, irrevocable decision. You have SS and you can choose to spend some on annuities when the rates are attractive, then use SWR on the remainder of the portfolio.

Isn't this just market timing interest rates? How is it different from Buy stocks when they are low?
Absolutely correct. Ten years from now, starting in my mid-seventies, the annuity mortality credits start to overwhelm the interest rate. That is timing, and one parent lived to age 92. I also bought S&P500 index shares in the 401k throughout the 1980s. Haven't sold from that account yet, so that is three decades of waiting for the right time to sell. Guilty as charged, and pleased about it.
Professor Emeritus
Posts: 2628
Joined: Mon Aug 13, 2012 6:43 am

Re: Variable spending strategies paper by Pfau

Post by Professor Emeritus »

heyyou wrote:
Professor Emeritus » Sun Apr 19, 2015 2:56 pm

heyyou wrote:
Asset spending is not a one shot, irrevocable decision. You have SS and you can choose to spend some on annuities when the rates are attractive, then use SWR on the remainder of the portfolio.

Isn't this just market timing interest rates? How is it different from Buy stocks when they are low?
Absolutely correct. Ten years from now, starting in my mid-seventies, the annuity mortality credits start to overwhelm the interest rate. That is timing, and one parent lived to age 92. I also bought S&P500 index shares in the 401k throughout the 1980s. Haven't sold from that account yet, so that is three decades of waiting for the right time to sell. Guilty as charged, and pleased about it.
:happy Heck I know people who had unprotected sex and no one got pregnant, :happy but as they say "luck is not a strategy".
Post Reply