SWR for basic vs discretionary spending
 02-06-2005, 09:41 PM #1 Thinks s/he gets paid by the post   Join Date: Feb 2003 Location: Mesa Posts: 3,588 SWR for basic vs discretionary spending I've been running some simulations again. I read the following article (I think ESRBob posted the link) a few weeks ago: "A Choice of Risks When Spending in Retirement" http://zunna.com/Research/ChoiceOfRisksInRetirement.pdf The article uses historical simulations (like FIRECALC or SWR Calculator) and considers two types of withdrawal strategies: 1) spend at a pace to maintain standard of living, 2) spend a fixed percentage of the value of the account each year. The first strategy (conventional historical analysis) maintains standard of living, but runs the risk of depleting the account. During most retirement periods throughout history this strategy would lead to either large final account values at the death of the retiree, or increased spending in later years. The second strategy the retiree is guaranteed to never run out of money, but the established withdrawal percentage can become vanishingly small at the end of the withdrawal period. During most retirement periods throughout history this strategy would lead to decreasing standard of living in later years. But a number of posters on this board have talked about two aspects of their retirement spending: 1) Basic required budget (that needs to be adjusted for inflation), and 2) Discretionary budget (that can be reduced or eliminated during times of poor portfolio performance). So based on some ideas that JWR1945 posted, I decided to try to simulate SWR based on the above division of budget. Here are the basic inputs: 30 year retirement, CPI adjustment for inflation, commercial paper for bond investments, minimum expense ratio =0.18%, stock/bond ratio =50/50. I used the Basic required budget to be the initial withdrawal rate, then assumed that withdrawals beyond that budget are taken as a percent of the portfolio value (This is done by increaseing the expense ratio until failure of the portfolio is observed). Here are the results with the maximum survivable initial withdrawal rates specified as a percent of the initial portfolio value. Basic............................. Required.........................Total.. Budget.....Discretionary.....Initial SWR........Budget.............SWR.... 2.0%.......4.92%.............6.92% 2.7%.......3.03%.............5.73% 3.0%.......2.22%.............5.22% 3.3%.......1.55%.............4.85% 3.6%.......0.82%.............4.42% 3.8%.......0.43%.............4.23% 4.0%.......0.00%.............4.00% The Basic Required Budget of column 1 is adjusted to match inflation throughout the simulated retirement. The column 2 withdrawal fluctuates based on the overall value of the portfolio. This value can increase or decrease in real terms depending on portfolio performance. So from the table, if your basic budget represents a 3% initial withdrawal rate, you could withdraw another 2.22% of your total portfolio value for a total initial withdrawal rate of 5.22% and achieve the same survival rate as a 4% basic budget adjusted for inflation for 30 years. __________________ __________________
 Join the #1 Early Retirement and Financial Independence Forum Today - It's Totally Free! Are you planning to be financially independent as early as possible so you can live life on your own terms? Discuss successful investing strategies, asset allocation models, tax strategies and other related topics in our online forum community. Our members range from young folks just starting their journey to financial independence, military retirees and even multimillionaires. No matter where you fit in you'll find that Early-Retirement.org is a great community to join. Best of all it's totally FREE! You are currently viewing our boards as a guest so you have limited access to our community. Please take the time to register and you will gain a lot of great new features including; the ability to participate in discussions, network with our members, see fewer ads, upload photographs, create a retirement blog, send private messages and so much, much more! Join Early-Retirment.org For Free - Click Here
Re: SWR for basic vs discretionary spending
 02-07-2005, 03:09 AM #2 John Galt   Posts: n/a Re: SWR for basic vs discretionary spending I like this! If I was so inclined, I could use this data as a segue into other possible scenarios, SWR-wise. Instead, I am going back to bed. Been up an hour or so. Time for a nap JG __________________
Re: SWR for basic vs discretionary spending
 02-07-2005, 10:45 AM #3 Recycles dryer sheets   Join Date: Dec 2002 Posts: 290 Re: SWR for basic vs discretionary spending Thanks! I liked what you did. It is interesting. Have fun. John R. __________________
Re: SWR for basic vs discretionary spending
 02-07-2005, 01:07 PM #4 Thinks s/he gets paid by the post   Join Date: Mar 2004 Posts: 1,318 Re: SWR for basic vs discretionary spending SG, This is great. Does this mean that inflation-adjusting just half of a normal 4% SWR, then taking a fixed 4.22% of portfolio every year thereafter gives you the same survivability as inflation-adjusting the full 4% SWR? If so, that is mighty interesting-- a nice blend of safety of spending and max possible dollar amount. But I was also intrigued/confused by another part of your post earlier: that going with the straight withdraw-a-percentage can result in 'established withdrawal percentage (dollar amount?) becoming vanishingly small at the end of the withdrawal period" -- that seems counter-intuitive to me. (I'd love to see your simulation results, btw). My read on Marbach's results were that a 4% percent-of-portfolio value gave you a 90% chance of having the _real_ value of the portfolio intact over 30 and 40 year periods. (And I believe the shortfall in the other 10% of the cases was on the order of 10% -- in other words that the portfolio never really fell below 90% of its original inflation-adjusted value. Maybe your vanishingly small withdrawals were the result of withdrawal rates much bigger than 4%? Or could it be that the dollar value of the 4% straight withdrawal got up so high during some middle years that it far exceeded the amount a normal inflation-adjusted SWR would have been, thus depleting the portfolio for later years? The whole intuitive point for me was that the "Straight Percentage" method _reduced_ the dollar amount of withdrawals during bad years (periods of low asset prices) relative to the normal SWR method, thus ensuring a _more_ robust portfolio in the later years, at the cost of a bit of belt-tightening during lean times. So I am surprised that the historicals simulations showed what you saw, can you shed any more light on it? As a side note, I use the method as "withdraw _up to _ 4% of your portfolio value each year". If that results in a withdrawal bigger than you need, then by all means reduce it-- perhaps the case when portfolio growth has exceeded inflation. (I've only been doing this for 4 years now and at the end of that four years my withdrawal is basically the same dollar amount under either scenario, but a spate of good years could see it rise well ahead of inflation, and I would not expect to take all that as a 'raise'. ) Thanks, SG -- looking forward to sharing data and getting to the bottom of this. __________________ ER for 10 years; living off 4.3% of savings (and a few book royalties ;-)
Re: SWR for basic vs discretionary spending
02-07-2005, 06:40 PM   #5
Thinks s/he gets paid by the post

Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: SWR for basic vs discretionary spending

Quote:
 . . .Does this mean that inflation-adjusting just half of a normal 4% SWR, then taking a fixed 4.22% of portfolio every year thereafter gives you the same survivability as inflation-adjusting the full 4% SWR?
From the simulations, a 2% inflation-adjusted withdrawal plus an annual withdrawal of 4.92% of the portfolio has the same 100% 30 year survivability as a single 4% inflation-adjusted withdrawal.

Quote:
 But I was also intrigued/confused by another part of your post earlier: that going with the straight withdraw-a-percentage can result in 'established withdrawal percentage (dollar amount?) becoming vanishingly small at the end of the withdrawal period" -- that seems counter-intuitive to me. (I'd love to see your simulation results, btw). My read on Marbach's results were that a 4% percent-of-portfolio value gave you a 90% chance of having the _real_ value of the portfolio intact over 30 and 40 year periods. (And I believe the shortfall in the other 10% of the cases was on the order of 10% -- in other words that the portfolio never really fell below 90% of its original inflation-adjusted value.
That conclusion comes directly from the zunna research paper I gave the link for. They use a 7.81% annual withdrawal rate as the highest sustainable 30 year withdrawal percentage possible. This withdrawal rate resulted in a \$1 ending value during the worst case 30 year simulation. Clearly, the withdrawal amounts during the last several years would have been very low. Obviously, you could still take an 8 cent withdrawal the following year, so the 1\$ end result is kind of arbitrary and misleading. This result sets a hard upper limit, but is probably not comparable to the standard inflation-adjusted result.

The authors also look at using the fixed percentage withdrawal applying the constraint that the end portfolio value must result in a final portfolio value with the same buying power as the original portfolio value. For this case, the highest sustainable withdrawal percentage was only 2.98%. This result seem too pessimistic compared to the inflation-adjusted 4% value.

Quote:
 The whole intuitive point for me was that the "Straight Percentage" method _reduced_ the dollar amount of withdrawals during bad years (periods of low asset prices) relative to the normal SWR method, thus ensuring a _more_ robust portfolio in the later years, at the cost of a bit of belt-tightening during lean times.
Yeah. Your earlier posts and the research paper are what made me look for a way to simulate this senario. I think this hybrid simulation makes a lot of sense. You can lump your food, shelter, clothing, medical insurance costs into the Required budget and then increase the expense ratio till the simulation fails. That expense ratio value (minus your actual expense ratio) is the safe fixed percentage withdrawal you could have survived with historically.

__________________

Re: SWR for basic vs discretionary spending
Re: SWR for basic vs discretionary spending
 02-08-2005, 07:07 AM #7 Richard Rosenthal   Posts: n/a Re: SWR for basic vs discretionary spending Interesting material. How does the required withdrawls from qualified accounts (401s/403s/457s, etc) play into this equation? __________________
Re: SWR for basic vs discretionary spending
 02-08-2005, 08:30 AM #8 Recycles dryer sheets   Join Date: Nov 2002 Posts: 373 Re: SWR for basic vs discretionary spending I think that another way of looking at what you are modelling can be done by looking at a split portfolio. Portfolio Core is funded with enough money to have the desired survival characteristics (30 yr, 95%, or longer/higher) for the desired basic standard of living. The remaining money is in Portfolio Fun, and can have a % of portfolio withdrawal scheme or any other aggressive withdrawal scheme. Portfolio Core provides the basic you are modeling, and portfolio Fun provides the extras. Having seperate "pots" of money may make it easier to visualize your strategy over time. Just a different way to look at what is essentially the same thing, if I am reading your post correctly. However, such a view has the advantage of being able to specify different time horizons for the seperate pots of money, such as 50yr for basic and 20yr for fun. Wayne __________________
Re: SWR for basic vs discretionary spending
02-08-2005, 08:41 AM   #9
Thinks s/he gets paid by the post

Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: SWR for basic vs discretionary spending

Quote:
 SG, A couple points: *(If I am understanding your earlier post correctly), I think anticipating your point about the vanishingly small withdrawals, Marbach actually doesn't run that study, where you'd spend a fixed percent and let the portfolio run into the ground. *He only does the fixed percent on two strategies: *1) keep dollar value of portfolio same (in this case 1 million- Table 3) or 2) keep real value of portfolio the same and spend a fixed percentage of portfolio (Table 5). *Failures in either case are when the ending value ends up below nominal or real 1 million respectively.
You are correct. I misread his table 3 end value and assumed that he had run the portfolio completely in the ground. In fact, he had only run it to the point where the final nominal value dipped below the nominal starting value. Still, considering 30 years worth of inflation at 3%, the final years of retirement would be at a lifestyle that is about 40% of the original lifestyle. That's what he discusses in the paper.

Also, it is not clear to me from the paper, but it appears that he imposes the simulation ending constraint only in the final year of the simulation. I'm not sure whether he requires that the portfolio maintain a value above the constraint leve thoughout the entire simulation or not.

None of this affects the simulations I've done, however. My end constraint is that there is money for the required budget throughout the entire simulation and that additional spending is kept to a percentage of the portfolio value. That additional spending may dwindle to pennies at any time during the simulation, or it may grow to many times the original amount. There are no constraints on this part of the budget.

You asked several questions, so I'll split up my answers in a couple of posts.
__________________

Re: SWR for basic vs discretionary spending
02-08-2005, 08:48 AM   #10
Thinks s/he gets paid by the post

Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: SWR for basic vs discretionary spending

Quote:
 . . . He came up with the following success rates, somewhat better than his Table 5 100% SP500 portfolio results, with presumably far less stomach churning over the years for the early retiree: with 4% pecent of the portfolio withdrawn each year, real value of the portfolio was maintained: For 10 or 20 years: *82.5% of the time For 30 years: *95.7% success For 40 years: *100% success and with a 4.5% withdrawal of portfolio value every year, real value of the portfolio was maintained: For 10 or 20 years: *79% of the time For 30 years: *95% success For 40 years: 100% success
Is this correct? Why would success rate go up for longer retirement periods? If this is true, then it indicates that while you could use this strategy and be pretty sure you would be in great shape in 40 years, there is a high probability that you will have to live through some periods of low income during the middle of your retirement. Or is this a typo?
__________________

Re: SWR for basic vs discretionary spending
02-08-2005, 09:00 AM   #11
Thinks s/he gets paid by the post

Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: SWR for basic vs discretionary spending

Quote:
 . . . Still, your new analysis uses the 7.81% bogey, and the 'maintain original nominal value of portfolio' approach which given inflation over long periods is sort of a low standard. *Would you consider using the Table 5 data and running your analysis for us again? *Or if it makes a difference, maybe using the Bernstein/CoffeHouse style portfolio I put together and the 4 or 4.5% SWRs, instead of Marbach's 100% SP500? If you are game, here is the portfolio I asked Marbach to run, and I can send you the 75-year data files (they came from Ibbotson and DFA,'s Data Book, but I hand-entered them -- not sure where else to get them electronically. *I would have wanted international stocks, but the 75 year data just doesn't exist, so I hoped riskier US small stocks would somehow be a proxy for these riskier international equities). S&P 500 11% US Large Value 9% US Small Value 9% US Small *10% US Micro Cap *11% Intermediate Government Bond *18% 1 Month Treasury * 2% LT Corporate Bonds *30%
I think you misunderstand what I've done here. I don't use any of the results from the Marbach paper. His simulations simply started my train of thought. I have not assumed any percentage withdrawal for the discretionary budget. The withdrawal percentage is the result of my simulations. It is the highest withdrawal amount that can be tolerated to insure that the inflation-adjusted withdrawal (the required budget) does not decline in real terms for the entire retirement period.

Regarding using your portfolio allocation, I have used FIRECALC for these simulations with a 50% stock - 50% bond (commercial paper) allocation. While I intend to run some additional simulations with various allocations, FIRECALC only provides me with an opportunity to use S&P500 stock performance and one bond choice.
__________________

Re: SWR for basic vs discretionary spending
02-08-2005, 09:39 AM   #12
Thinks s/he gets paid by the post

Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: SWR for basic vs discretionary spending

Quote:
 I think that another way of looking at what you are modelling can be done by looking at a split portfolio. *Portfolio Core is funded with enough money to have the desired survival characteristics (30 yr, 95%, or longer/higher) for the desired basic standard of living. * The remaining money is in Portfolio Fun, and can have a % of portfolio withdrawal scheme or any other aggressive withdrawal scheme. *Portfolio Core provides the basic you are modeling, and portfolio Fun provides the extras. Having seperate "pots" of money may make it easier to visualize your strategy over time. *Just a different way to look at what is essentially the same thing, if I am reading your post correctly. *However, such a view has the advantage of being able to specify different time horizons for the seperate pots of money, such as 50yr for basic and 20yr for fun. * Wayne
Hi Wayne,

That's how I started looking at the problem initially. As you say, the separate pots of money approach offers the advantage that you could use different time frames or even different allocations for the two pots. I think there is some value to looking at the problem that way. But I envisioned some issues that the separate pot approach would not easily address.

1st: FIRECALC would not recognize a set of constraints that would cause the discretionary pot to ever fail. Your spending in FIRECALC for this part of the simulation would be \$0 and you could set the expense ratio to anything. FIRECALC would never recognize a failure. You can always withdraw x% of a portfolio as long as x is less than 100. So the percentage withdrawal on the discretionary pot could be very high and you would not see failure. You could concievably print out the detailed FIRECALC results and look at the 101 tabulated 30 year histories on a year by year basis to test the discretionary withdrawal against some constraint, but that would be tedious and make it very difficult to optimize.

2nd: It isn't clear to me that, if you could put a realistic failure constraint on the discretionary pot, you would observe failure during the same time period as you would observe with the required budget. Nor is it obvious to me that you would ever want to accept failure of your required budget if discretionary funds could fix the temporary shortfall. In other words, mingling the funds offers potential short-term advantages.

If I get a chance, I'll try to run the same basic simulations as a set of two (required pot and discretionary pot) and see if any significant differences emerge.
__________________

Re: SWR for basic vs discretionary spending
02-08-2005, 09:46 AM   #13
Recycles dryer sheets

Join Date: Nov 2002
Posts: 373
Re: SWR for basic vs discretionary spending

Quote:
 Is this correct? Why would success rate go up for longer retirement periods? If this is true, then it indicates that while you could use this strategy and be pretty sure you would be in great shape in 40 years, there is a high probability that you will have to live through some periods of low income during the middle of your retirement. Or is this a typo?
I think we have seen this in the past on various studies also. Over a 10 or 20 year period, the portfolio dropped below the inflation adjusted beginning value. Over longer time frames, the portfolio recovered to exceed the inflation adjusted beginning value.

Wayne
__________________

The Re: SWR for basic vs discretionary spending
 02-08-2005, 09:51 AM #14 Thinks s/he gets paid by the post   Join Date: Dec 2003 Posts: 1,375 The Re: SWR for basic vs discretionary spending [quote] Is this correct? *Why would success rate go up for longer retirement periods? *If this is true, then it *indicates that while you could use this strategy and be pretty sure you would be in great shape in 40 years, there is a high probability that you will have to live through some periods of low income during the middle of your retirement. * Salaryguru: Given the fact that stocks in the long haul (historically), have outperformed most other passive investments, it makes sense to me that his study is correct. That being said, (other than the fact that I would like to leave my kids something), I am more interested in my spendable for the next 15 to 20 years, then ending up with an overkill after that period of time. What i have personally done is reduce my exposure to equities to 30%, and allow them time to run. After a lot of soul-searching, I figure this gives us the best odds of the maximum "draw" while we are able to enjoy the additional income. Distribution is much different than accumulation Jarhead __________________
Re: SWR for basic vs discretionary spending
02-08-2005, 10:02 AM   #15
Recycles dryer sheets

Join Date: Nov 2002
Posts: 373
Re: SWR for basic vs discretionary spending

Quote:
 1st: FIRECALC would not recognize a set of constraints that would cause the discretionary pot to ever fail.... 2nd: ..., In other words, mingling the funds offers potential short-term advantages.
Yes, I agree with statements. I do think that Gummy had some spreadsheets which modelled variations like this, including a % of extras strategy. I think one of them included the percent of portfolio method. In anycase, the seperate pots is so that you can spend one, i.e. failure of the pot is in a manner of thinking, a desired outcome. It's the ER fun pot, for travel, etc. I see the lifestyles of both my parents, and my wife's parents and both of them spend a lot less, don't golf, etc. much anymore, and travel very little as they find travel hard. So spending the fun pot at the highest rate that it can fail in 20 years (or even less) is one goal of this strategy.

As to having the fun pot bail out the basic pot in bad years, maybe combining them gives a slightly better survival case for the basic pot. I can see how that advantage might help out the numbers a little bit.

I do think that placing different probablilites and time frames on the two pots is an important addition. I could see 50yr, 100% as the basic pot criteria, and 20yr, 50% as the fun pot criteria. Ok, maybe not that extreme for everyone, but I'm thinking about it.

Wayne
__________________

Re: SWR for basic vs discretionary spending
 02-08-2005, 10:25 AM #16 Thinks s/he gets paid by the post   Join Date: Feb 2003 Location: Mesa Posts: 3,588 Re: SWR for basic vs discretionary spending Wayne, Jarhead, Thanks for the feedback on the simulation results. I appreciate the critiques and input on the simulations and analysis. The original Marbach paper was very interesting and useful to me, but I couldn't find specific descriptions for how he applied his failure constraints to the fixed percentage withdrawal aproaches. Based on the tabulated results, I suspected that the portfolio value was constrained only in the final year of the simulation. And the tables ESRBob has presented in this thread would appear to confirm that. But I don't find that very comforting as a retirement withdrawal plan. You end up concluding that if you live for 40 more years you will be very comfortable, but you may have to live in a box and eat catfood for a few years along the way. Maybe I exagerate a little. __________________
Re: SWR for basic vs discretionary spending
 02-08-2005, 08:08 PM #17 Thinks s/he gets paid by the post   Join Date: Mar 2004 Posts: 1,318 Re: SWR for basic vs discretionary spending SG- Yes, I believe I can confirm Marbach's methodology -- he is only looking at ending value to decide whether one of the historical runs is a failure or a success. You get to take the money out every year in between, but you'd need to sweat out some uncomfortable times along the way, in some of these cases. By looking at the 10 and 20-year data he sent me, there are plenty of that are underwater at those milestones, but by 30 years almost everybody is home free. Although there were lots of flaws in the recent paper by Jonathan Guyton (October 2004 Journal of Financial Planning-- Nords did a good job of ripping it to shreds!) I did appreciate the fact that he was trying to model some sort of dynamic rules that you could apply each year during ER to decide whether and how much to increase your withdrawal. Paper is at: http://www.fpanet.org/journal/articl...nderforprint=1 Based on some questions from Nords last month, I am toying with a new SWr rule; I have no way to model it but maybe someone does: Withdraw 4% of Portfolio value each year. But if Portfolio Value goes down, just freeze your withdrawals at last year's levels. Continue at that level until Portfolio Value once again rises and you can start to move back up. It doesn't fully address inflation, but it also keeps your spending from going down painfully during market downturns. It softens the withdrawals from the portfolio during downturns, but also lets you get your 'raises' back when good times return. And if your portfolio is a blowout success, you get to keep consuming bigger real amounts every year if you like. (most of the inflation adjusted withdrawals keep you at the same real spending level every year even if you've become a gazillionaire.) What do you all think? Anybody have any ideas about how to model this rule in a historical study? __________________ ER for 10 years; living off 4.3% of savings (and a few book royalties ;-)
Re: SWR for basic vs discretionary spending
 02-09-2005, 08:26 PM #18 Give me a museum and I'll fill it. (Picasso)Give me a forum ...   Join Date: Dec 2003 Location: Losing my whump Posts: 22,697 Re: SWR for basic vs discretionary spending Too much analysis. I spend what needs to be spent. Past that, if I want something and it passes the "wife test", if I still really want it I look at how things are going in the old portfolio. If they're good, I go ahead and get the new toy. If they're not, I wait a while. And wait awhile. If after some scientifically undeterminate time I still REALLY want the thing and the port STILL isnt doing well, and I wont impair my financial future by getting it, I get it anyways. Half the time I dont want it anymore. Half the time I want something else more. The other half of the time I get it anyway because life is too short. Yes, I'm math impaired, and thats a good thing! I still dont get SWR analyses. Its easy...if you're spending 6+% of your nestegg on a sustained basis or planning to eat portions of your nestegg on a sustained basis as part of your overall "plan", you're not gonna make it. Period. If you're spending <3% of your egg, you either have way too frickin much money, or you're depriving yourself. Cut it out. __________________ Be fearful when others are greedy, and greedy when others are fearful. Just another form of "buy low, sell high" for those who have trouble with things. This rule is not universal. Do not buy a 1973 Pinto because everyone else is afraid of it.
Re: SWR for basic vs discretionary spending
 02-09-2005, 09:01 PM #19 Thinks s/he gets paid by the post   Join Date: Mar 2004 Location: Dallas Posts: 1,211 Re: SWR for basic vs discretionary spending IMHO, it is safer to use a condom than a SWR. Cheers, Charlie __________________
Re: SWR for basic vs discretionary spending
02-09-2005, 09:04 PM   #20
Thinks s/he gets paid by the post

Join Date: Feb 2003
Location: Mesa
Posts: 3,588
Re: SWR for basic vs discretionary spending

Quote:
 Interesting material. How does the required withdrawls from qualified accounts (401s/403s/457s, etc) play into this equation?
I'm sorry, Richard. I missed your post. The FIRECALC analysis does not address the issue of where the withdrawals come from. Clearly, the withdrawal strategy can be important since it will affect taxes, and if you do it wrong can cost you penalties. There are other programs designed to optimize your account withdrawal strategy.
__________________

__________________

 Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)