Accumulation-Dynamic Decumulation Strategy

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Thu Mar 08, 2018 8:48 pm

Hi,

I'm seeking comments on a withdrawal method I've thought up. In the end I hope it inspire other with similar or even better strategies. I've been reading the forums for a while and I've read many Variable Withdrawal Rate (VWR) threads. I've also read the Withdrawal Methods Wiki page. Yet, I was not satisfied with the methods.

Let be clear, my method is what I currently use (the accumulation phase so far.. not much to write home about). I do plan to continue using this method going into retirement in a good while.

I've done backtesting the method and done a Monte Carlo analysis. I'm not seeking a review of my study.

If you want to see the full study please let me know. I'll post a link. Otherwise it can be implemented fairly easily without it.

Here's the goal of the stratregy:
- Remove the risk of ruin >> done by going "percent withdrawal"
- Reduce the variability of the withdrawal. >> done by the main 3 sub-strategy
- Reduce the "retirement date paradox", that is a SWR retiree of 2007 gets a lot more than a 2008 retiree. >> done by looking at the accumulation phase, not just the decumulation phase. also the sub-strategy to smooth and cap increase
- Have a floor, but not let it be in the end an absolute anchor. In the end at 70-80 years old, why not lower it. >> done by the reserve
- remove hyperinflation risk. Honestly I don't believe that either US or Canada going's the face high inflation in the future. If they do, I hope this method deal with it regardless.
- Provide other method than "Asset Allocation" to manage "volatility of return" risk. If you have a single hammer in the tool box, you're gonna always it. However make no mistake, Asset Allocation I believe is a valid technique if you can't stomach the year to year Market Value.

So here's the strategy in short :
Three strategies to reduce the volatility of withdrawal are used in the ADD strategy.
  1. Smoothing
    o This is a well establish method in the DB pension world. It can be naively explained as taking the moving average over 5 years. The advantage is to have a lower the volatility and assess more prudently the fund. It has the effect of delaying the impact of short term gains/loss, hopefully taking advantage of any reversion to the mean.
  2. Increase Cap
    o Since we take a percentage of the fund each year, exceptional returns would also imply exception increase in withdrawal. Instead, we cap the increase to a maximum versus the previous withdrawal.
  3. Capital Reserve
    o This technique set aside money for economic hardship. The reserve will fund the gap between 95% previous withdrawal and current withdrawal. The reserve would kick-in to fund the deficit. In economic good times, the reserve grows. The 5% gap can be seen as a bonus. Bonuses are not guaranteed. They only are often paid.

I admit, this is not a simple run of the mill strategy. It's more complex than Safe Withdrawal Rate (SWR) or VWR. However, in the end I believe a sheet of paper would be sufficient to do most of it with rather ease.

Here's how:
Approximate Smoothing Market Value (SMV) by averaging last 5 years of Market Value * (1+Expected Return)^2 (note Expected return less Variance is better approximation)

Take a withdrawal of SMV * your selected withdrawal rate

Compare with last year withdrawal * (1+6%). This is the increase cap. Do not adjust for inflation.

Step 3 have a reserve of the fund at retirement of 25% of overall fund. Use this to provide a floor withdrawal if you feel Variable withdrawal drop too low.

My study goes into deeper technicality, but in essence you go a long way with these 3 steps. Another Highly recommend step is before retirment where you need to cap currently estimate withdrawal by 115% of previous year estimate withdrawal. This way a 30% increase in one year will not lead to 30% increase at retirment. This lowers the probability of reduction of withdrawals in the future.

I'll await comment I hope: :|

Here's the Google Drive and DropBox links for the spreadsheets with the strategy.
Last edited by icetrap on Mon Apr 16, 2018 11:47 am, edited 1 time in total.

longinvest
Posts: 3109
Joined: Sat Aug 11, 2012 8:44 am

Re: Accumulation-Dynamic Decumulation Strategy

Post by longinvest » Fri Mar 09, 2018 12:39 am

Icetrap,
icetrap wrote:
Thu Mar 08, 2018 8:48 pm
Hi,

I'm seeking comments on a withdrawal method I've thought up. In the end I hope it inspire other with similar or even better strategies. I've been reading the forums for a while and I've read many Variable Withdrawal Rate (VWR) threads. I've also read the Withdrawal Methods Wiki page. Yet, I was not satisfied with the methods.

Let be clear, my method is what I currently use (the accumulation phase so far.. not much to write home about). I do plan to continue using this method going into retirement in a good while.

I've done backtesting the method and done a Monte Carlo analysis. I'm not seeking a review of my study.

If you want to see the full study please let me know. I'll post a link. Otherwise it can be implemented fairly easily without it.

Here's the goal of the stratregy:
- Remove the risk of ruin >> done by going "percent withdrawal"
- Reduce the variability of the withdrawal. >> done by the main 3 sub-strategy
- Reduce the "retirement date paradox", that is a SWR retiree of 2007 gets a lot more than a 2008 retiree. >> done by looking at the accumulation phase, not just the decumulation phase. also the sub-strategy to smooth and cap increase
- Have a floor, but not let it be in the end an absolute anchor. In the end at 70-80 years old, why not lower it. >> done by the reserve
- remove hyperinflation risk. Honestly I don't believe that either US or Canada going's the face high inflation in the future. If they do, I hope this method deal with it regardless.
- Provide other method than "Asset Allocation" to manage "volatility of return" risk. If you have a single hammer in the tool box, you're gonna always it. However make no mistake, Asset Allocation I believe is a valid technique if you can't stomach the year to year Market Value.

So here's the strategy in short :
Three strategies to reduce the volatility of withdrawal are used in the ADD strategy.
  1. Smoothing
    o This is a well establish method in the DB pension world. It can be naively explained as taking the moving average over 5 years. The advantage is to have a lower the volatility and assess more prudently the fund. It has the effect of delaying the impact of short term gains/loss, hopefully taking advantage of any reversion to the mean.
  2. Increase Cap
    o Since we take a percentage of the fund each year, exceptional returns would also imply exception increase in withdrawal. Instead, we cap the increase to a maximum versus the previous withdrawal.
  3. Capital Reserve
    o This technique set aside money for economic hardship. The reserve will fund the gap between 95% previous withdrawal and current withdrawal. The reserve would kick-in to fund the deficit. In economic good times, the reserve grows. The 5% gap can be seen as a bonus. Bonuses are not guaranteed. They only are often paid.

I admit, this is not a simple run of the mill strategy. It's more complex than Safe Withdrawal Rate (SWR) or VWR. However, in the end I believe a sheet of paper would be sufficient to do most of it with rather ease.

Here's how:
Approximate Smoothing Market Value (SMV) by averaging last 5 years of Market Value * (1+Expected Return)^2 (note Expected return less Variance is better approximation)

Take a withdrawal of SMV * your selected withdrawal rate

Compare with last year withdrawal * (1+6%). This is the increase cap. Do not adjust for inflation.

Step 3 have a reserve of the fund at retirement of 25% of overall fund. Use this to provide a floor withdrawal if you feel Variable withdrawal drop too low.

My study goes into deeper technicality, but in essence you go a long way with these 3 steps. Another Highly recommend step is before retirment where you need to cap currently estimate withdrawal by 115% of previous year estimate withdrawal. This way a 30% increase in one year will not lead to 30% increase at retirment. This lowers the probability of reduction of withdrawals in the future.

I'll await comment I hope: :|
It's interesting.

I like that the goals are clearly stated.

This method seems to be aiming at developing a kind of personal pension fund paying a low-fluctuation pension. It differs from a traditional pension fund in that it won't benefit from continuous contributions of working members while paying pensions to retired members; there's a single member who is retired and receiving a pension.

It isn't clear to me if the stated "floor" goal, for withdrawals, is a firm one. Mathematics and logic tell us that no portfolio withdrawal method can provide a firm inflation-indexed floor for more than 30 years, the longest maturity of Treasury Inflation Protected Securities (TIPS). We don't know the future yields of TIPS in the next 30 years. We don't know if the government will continue to issue TIPS in the future. And, finally, we don't know how long the retiree will live.

How will the validity of this method be assessed?
Last edited by longinvest on Fri Mar 09, 2018 7:57 am, edited 1 time in total.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / domestic (nominal / inflation-indexed) long-term bonds | VCN/VXC/VLB/ZRR

itstoomuch
Posts: 5343
Joined: Mon Dec 15, 2014 12:17 pm
Location: midValley OR

Re: Accumulation-Dynamic Decumulation Strategy

Post by itstoomuch » Fri Mar 09, 2018 12:57 am

OP are you only limiting your self to Indexes and stocks/bonds?
If so, why?
Rev012718; 4 Incm stream buckets: SS+pension; dfr'd GLWB VA & FI anntys, by time & $$ laddered; Discretionary; Rentals. LTCi. Own, not asset. Tax TBT%. Early SS. FundRatio (FR) >1.1 67/70yo

AlohaJoe
Posts: 3923
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Accumulation-Dynamic Decumulation Strategy

Post by AlohaJoe » Fri Mar 09, 2018 2:38 am

Having read a large number of withdrawal strategies, here's what I think the creators of new strategies could do to help their cause:

- Include a super simple & clear explanation of the strategy. The closer you can get to writing it in code (like, say, RStudio) the better because it reduces ambiguity.
- Showing an example of how it works for 3 or 4 years with actual numbers also helps readers see it as a non-theoretical thing.
- There are a massive variety of metrics one could look at all. Basically, the more the better.
- Don't just compare it to 4% withdrawals. That's the worst possible strategy. It is easy to look good compared to the worst possible strategy. Also benchmark yourself against the best strategy.
- If your strategy has multiple moving parts --as yours does -- try to tease it apart and help readers understand what each moving part contributes.
- Any "magic numbers" need to have explanations for how those numbers were arrived at and how sensitive they are to change.

Here are some concrete examples from your OP that relate to the above points:

- I found it fairly hard to follow algorithm, especially ambiguous statements like "Use this to provide a floor withdrawal if you feel Variable withdrawal drop too low"
- You talk about a "reserve" fund. How much benefit does it actually add? In what situations? Can you quantify the improvement of the strategy with & without it? How do you decide what amount of benefit warrants the conceptual complexity it adds? Does a reserve have the same asset allocation as the non-reserve?
- Why is the increase cap 6%? And not 10% Or 4%?

AlohaJoe
Posts: 3923
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Accumulation-Dynamic Decumulation Strategy

Post by AlohaJoe » Fri Mar 09, 2018 3:39 am

Here's my understanding of how the strategy works and a quick comparison to some alternatives:

1. Calculate a "smoothed" portfolio value by taking a 5-year average
2. Multiply by (1+estimated returns)^2 to....I'm not really sure why we bother with this, actually.
3. Multiply the "smoothed" portfolio value by a constant percentage (what percentage to use?)
4. Apply a ceiling of 106% last year's withdrawal to additionally smooth things
5. Apply a floor of 95% last year's withdrawal to additionally smooth things
Special steps: There's all the stuff about "reserve" funds.

I'm not convinced all those variants of smoothing are necessary, to be honest.

Let me ignore the "reserve" fund for now. Here's how the "ADD minus reserve" (ADD-reserve) compares to some alternative strategies.

This is Blanchett's WER (Withdrawal Efficiency Rate):

Image

And this is McClung's HREFF (Harvesting Rate Efficiency something something):

Image

Caveat: remember I left off the "reserve" part of things. But I am a bit dubious that the addition of the reserve would have a profound enough impact to catapult it to the upper tiers of the league table.

The primary effect is (unsurprisingly) to smooth the income:

Image

We also try to run the algorithm with various parts removed, to see how much they actually contribute.

What we if we remove the "multiply by (1+estimated returns)^2" part?

Image
Image

At first check, they seem identical, so that seems like a complication with little value.

Instead of taking a moving average over 5 years we could just use the current portfolio value. (This will rely on the ceiling and floor guardrails alone to provide smoothing.)

Image
Image

Having the multiple methods of smoothing doesn't immediately offer any clear benefits for the complexity that it adds.

Alternatively, let's get rid of the ceiling and floor guardrails -- and rely just on the 5-year rolling average to smooth things.

Image
Image

Here the difference is more noticeable. The blue line -- using both methods of smoothing -- is definitely smoother than just relying on the 5-year rolling average.

Overall, I guess my feeling is that the strategy is overly complicated with parts that don't appear to offer a significant contribution. The heart of the strategy appears to be "constant percentage withdrawals with a ceiling and floor to limit the maximum annual increase/drop". Maybe the reserve adds something extra but, as I said above, I'm dubious about it.

User avatar
AtlasShrugged?
Posts: 613
Joined: Wed Jul 15, 2015 6:08 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by AtlasShrugged? » Fri Mar 09, 2018 7:32 am

This is Blanchett's WER (Withdrawal Efficiency Rate)....
AlohaJoe....Where would 'Prime Harvesting' fall in terms of efficiency?
“If you don't know, the thing to do is not to get scared, but to learn.”

AlohaJoe
Posts: 3923
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Accumulation-Dynamic Decumulation Strategy

Post by AlohaJoe » Fri Mar 09, 2018 7:46 am

AtlasShrugged? wrote:
Fri Mar 09, 2018 7:32 am
This is Blanchett's WER (Withdrawal Efficiency Rate)....
AlohaJoe....Where would 'Prime Harvesting' fall in terms of efficiency?
Prime Harvesting isn't a withdrawal strategy so you can't really compare it to the other things. Prime Harvesting is about managing your asset allocation; it doesn't tell you anything about how much to withdraw.

There's also a whole huge thread on McClung's book with way more information and arguments :)

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 3:08 pm

longinvest wrote:
Fri Mar 09, 2018 12:39 am
This method seems to be aiming at developing a kind of personal pension fund paying a low-fluctuation pension. It differs from a traditional pension fund in that it won't benefit from continuous contributions of working members while paying pensions to retired members; there's a single member who is retired and receiving a pension.

It isn't clear to me if the stated "floor" goal, for withdrawals, is a firm one. Mathematics and logic tell us that no portfolio withdrawal method can provide a firm inflation-indexed floor for more than 30 years, the longest maturity of Treasury Inflation Protected Securities (TIPS). We don't know the future yields of TIPS in the next 30 years. We don't know if the government will continue to issue TIPS in the future. And, finally, we don't know how long the retiree will live.

How will the validity of this method be assessed?

You are exactly right in that there's only 1 member. So it ressembles more to a Single Pension Fund (never worked with them, but I'm guessing should be similar?)


The "floor" is provided by the reserve. Basically, if the reserve drop to 0, you lose the floor. My strategy to manage it is to let the Cash Flow(CF aka withdrawals) drop say 95% of any previous year CF. The reserve pays the for the excess. Ex: If the variable withdrawal are 85% of a previous year, the remaining 10% (95%-85%) is paid by the reserve. To manage the risk of it dropping to 0, I've limit the reserve withdrawal to say 10%. In other word if reserve is 250,000$, maximium withdrawal from it is 25,000$. You need it a high enough limit since it helps going along the goal. The goal is not to keep the reserve intact but to use it in time of need.

Personnaly, I do not fear inflation. For the future, US & Canada seems to be reliably making near 2%. For the past, I honestly have hard time with the 70's where an inflation-indexed floor required a lot. SWR for exemple suggest withdrawing more than 10% to keep up with inflation,in my mind, I would feel unsafe using the "Safe" withdrwal rate method in this context. :?

For the validity of the method, I've used 2 risk indicators.
  1. Risk I : % of run(either historical or Monte Carlo)withdrawal lower than 95%* Initial CF at retirement(ICF).
  2. Risk 2 : % of run(either historical or Monte Carlo)withdrawal lower than 95%* Any previous year CF at retirement(ICF).
Please let me know if this seems valid... It does go along the objective to limit the volatility of the withdrawal.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 3:15 pm

itstoomuch wrote:
Fri Mar 09, 2018 12:57 am
OP are you only limiting your self to Indexes and stocks/bonds?
If so, why?
Dumb question, what is OP?

Anyway, I've done my backtesting with stocks US/bonds US. I see no reason to use anything else. I honestly think that going with a more diverse portfolio would provide better results. In the end this strategy is independent of the portfolio allocation. I would fear that changing mid-ways portfolio strategy may affect badly the method.

I've ask Tyler from Portfolio chart, if he was interested in looking to making it available with is very diversify set of portfolio allocation. If he is willing, this may help you get a better view of what you could be looking for.

22twain
Posts: 1600
Joined: Thu May 10, 2012 5:42 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by 22twain » Fri Mar 09, 2018 3:40 pm

OP = Original Poster, the person who started this thread. That's you! 8-)
My investing princiPLEs do not include absolutely preserving princiPAL.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 5:45 pm

I'm glad that I already most of this :-) However, I've not yet try to check vs other method.
I've display the simple math :( However, I fear it have a lots of twist and turns since it have 3 strategies in it. After the simple math I've explain further each of the sub-strategy in 3 seperate post.
AlohaJoe wrote:
Fri Mar 09, 2018 2:38 am
- Any "magic numbers" need to have explanations for how those numbers were arrived at and how sensitive they are to change.
There's a lot of magic number. Most are mostly non sensitive (ex : increase cap, smoothing expected return) other very senstive (% of withdrawal, reserve level) I'll try my best to have 3 separete post for each of the three sub-strategy. The most obvious one is the % withdrawal over the "regular fund" (that is 75% of the fund). The great outcome of ADD is that the Average Cash Flow for each of the asset allocation seems to converge whatever the withdrawal rate used. Too aggresive withdrawal rate (say 7%) gets you more probability of reduction of future Cash Flow. If you go for conservative rate (say 3%), the future cash flow picks up the slack as the fund increase faster than withdrawal.
AlohaJoe wrote:
Fri Mar 09, 2018 2:38 am
- I found it fairly hard to follow algorithm, especially ambiguous statements like "Use this to provide a floor withdrawal if you feel Variable withdrawal drop too low"
Sure, I can elaborate. The floor I've set is 95% of any previous withdrawal. However, the reserve withdrawal are cap at 10% of the fund to try (honestly it only helps a little bit). You can see my following post on Reserve specifically for more info.
AlohaJoe wrote:
Fri Mar 09, 2018 2:38 am
- You talk about a "reserve" fund. How much benefit does it actually add? In what situations? Can you quantify the improvement of the strategy with & without it? How do you decide what amount of benefit warrants the conceptual complexity it adds? Does a reserve have the same asset allocation as the non-reserve?
it provide no benefit. It's only use for the floor. It only apply when withdrawal are lower than 95% of previous withdrawal. Yes I can... I'll try to get the impact of each specific sub-strategy back to you. The reserve is the method the floor get created. In the end, I disliked the idea of having the floor already included in the main fund. I find it helpfull (from a behavioral idea) to have a separete bucket for it. This way It's easy to get "confidence" that everything is going to be ok if the reserve are high. If reserve gets depleted by the year, than we start to be more nervous. Not sure how this may pan out in real event though. One can easily go back and adjust to lower withdrawal if he sees his trouble economics.

I've try to have the same asset allocation in the reserve, but have not yet finish exploring this option. I'll sure try it later.
AlohaJoe wrote:
Fri Mar 09, 2018 2:38 am
- Why is the increase cap 6%? And not 10% Or 4%?
106% increase is appropriate. At 7% ADD withdrawal rate the average withdrawal over 40 years adjusted for inflation is maximized and the drop in withdrawal is the lowest. Over the historical data, cap between 105% and 110% produces similar results in those metrics. Impact are experience when cap is 115% or more. This drop the II Risk Max to 55% instead of 78%. The average cash flow is somewhat reduce by a few % when looking at the cap below 105%.

Honesty, in the end this magic number is not that helpfull. If you increase it to say 10%, you get to enjoy the high increase of the 80's/90's. However, you get hammered by more risk when things go bad. At lower rate say 4% (Nomimal) you are severly limiting increase in good economic. At the upside, it lowers the risk.



Here's the Super simple way I see it in math format:

Background Data: A few years into retirement.
  • Total fund value : 419,483$
  • - Regular fund value : 261,174$
  • -- Smoothed Regular Fund: 277,363$
  • - Reserve fund value : 158,309$
  • Last year withdrawal : 16,498$
  • This was not due to the increase cap.
  • Greatest previous withdrawal : 19,618$
ADD withdrawal:

Total withdrawal : 18,637$ = 17,323$ + 1,314$
Regular withdrawal : 17,323$ = Min (24,038 ; 17,323$)
  • Smoothed withdrawal prescribed : 24,038$ = 8.67% * 277,363$
    [Parameter] ADD withdrawal rate : 6.50%
    This drives the “prescribed withdrawal rate” to 8.67% = 6.5%/(75%). The
  • Increase Cap: 17,323$ = 16,498$ * (105% + 0%)
    • [Parameter] Increase Cap: 105%.
    • [Parameter] Increase Cap is augmented if prior year was cap: +0%.
  • Reserve withdrawal : 1,314$ = Min(18,637$ - 17,323$; 31,662$)
     Maximum covered withdrawal : 18,637$ = 95% * 19,618$
    • [Parameter] Reserve protected : 95%
    • [Parameter] Reserve maximum withdrawal rate : 20%
    o Maximum withdrawal: 31,662$ = 20%* 158,309$
    • No adjustment required since last years and current year reserve fund =158,309/419,483 = 38% > 20%
    • In fact, the reserve exceeds the 30% threshold and will return back to the regular fund the excess 158,309 - 30%*419,483= 32,464. This can be though as a regular contribution.
Last edited by icetrap on Mon Apr 09, 2018 6:33 am, edited 1 time in total.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 5:48 pm

Here's the model in text form :

ADD Model in aggregate
The ADD model functions by withdrawing a certain % of the fund each year. The % of withdrawal will vary each year according to the economic situation that was previous experience. When in good times, the withdrawal rate will be lower. The extra reserve will be used when economic time worsens. Thus in bad times, the withdrawal rate will increase. While this may look counterintuitive, this is a direct following the ant insight in Aesop’s fable. We need to build reserve when the sun is shining.

The funds are split into two. 75% of the account goes in the regular fund. 25% goes into the reserve fund. The withdrawal of the ADD strategy is in 2 parts, the regular and the reserve. The regular withdrawal consist of taking the minimum between the smoothed fund * prescribed withdrawal rate and an increase cap. The increase cap is based on the last year withdrawal times the maximum increase. The reserve is left to grow unless the regular withdrawal is lower than 95% of any previous year’s withdrawals. The reserve covers up to the 95% mark, but not in excess of 20% withdrawal rate on the reserve. Finally in prolong great economic cycle, the reserve is slowly transfer back to the regular fund if the reserve exceeds 35% of the total fund.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 5:53 pm

Heres the smoothing description use linear smooting. I believe it s advantage over Moving average is when there's lots of Cash flow. Otherwise It feels better because it's complex.... :wink: However, I'm starting to question the added advantage of beeing complex when a simple Moving average 5 years adjusted for 2 years (1+expected Return on Asset)^2 could do.

When the fund is lower than the average, you would withdraw more %. When the fund is higher than you would withdraw less %. However, your fund is not as simple as a sin function. One way to apply smoothing is the linear recognition N-Year smoothing. First, estimate the fund return to be, say 10%. You then recognize 20% of the real rate of return instead of the estimate for last year, 40% 2 years prior, 60% 3 years prior, 80% 4 years prior and 100% 5 years prior. Thus the result would be somewhere near the imaginary line in the above sin function.
The main disadvantage is that smoothing will hide/delay the recognition of Gain&Loss. Thus, this method can serve you as a quick estimate, but the fund value may be vastly different. Be prepare to adjust ADD withdrawal for past event rather than current event. In other word, 2008 crash affected the smoothed average over 5 years. It helps manage the short term variations.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:07 pm

Here's the math for a few years :

Simple example of the smoothing assuming a less than expected return

Data:
Image

Smoothed Value with corridor at 2015 = Min(CorridorMin; Max(CorridorMax; Smoothed Value)) = 558,359$
  • CorridorMin: 530,000*0.8 = 424,000 $
    [Parameter] Corridor Minimum = 80%
  • CorridorMax: 530,000*1.2 = 636,000 $
    [Parameter] Corridor Maximum = 120%
  • Smoothed Value = 1/5*500,000*(-(-4%)+5.67%)+2/5*480,000*(-(2.0833%)+5.67%)+3/5*490,000*(-(4.0816)%+5.67%)+4/5*510,000*(-(3.9216)%+5.67%)+530,000 = 558,359$
    [Parameter] conservative rate of return = 5.67%
    A simple check is to compare with a 5 year moving average. Multiply the average (502,000) by 1.0567*1.0567 = 560,540$

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:08 pm

Here's the in word, what the sub-strategy Capping is trying to do :

Capping
ADD use an increase cap to manage the inflation risk. In simple term, if we withdraw 40,000$ yesterday, we limit today withdraw to 45,000$. This mitigates the effect of financial bubble and provide a conservative strategy to manage increase. On the down side, you delay the gain of excessive return for later. To manage this, ADD increase the cap by 1% for each prior year that the cap was applied. This slowly allows the retiree to gains advantage of any sustained rise in the fund value.

A complementary strategy is to limit the projected withdraw using the same strategy. If the year before retirement, projected withdrawal was 40,000$, ADD limits the retirement withdrawal to 48,000$. The capping is greater to adjust for the 0% withdrawal prior retirement.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:18 pm

Here's in a simple example how the capping applies:

Image

The initial Cap is 105%. The rate of return during these years was exceptional. Without a cap, withdrawals increase by 24% in a manner of 5 years. To manage this high increase, the Cap is increase each year it was previously cap. This increase in cap insures that conservatively increase your withdrawal according to the market condition. It also have the added benefit of removing the decision of increasing your withdrawal. One of the major flaw of the classic 4% rule is that it may become too conservative (withdrawing 1%) after 10-20 years. The strategy does not have the flexibility to decide how to increase it. If you do it too often, then you are at risk of always increasing it in good time, without any buffer for adverse deviation.

That one is the simple one I guess :D

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:20 pm

The Reserve in word:
Banks and insurance company alike have capital reserve. This same principle can be applied to a personal fund without any outside force. This technique is similar to the bucket approach. Say that you have 1 million at retirement. You take 250k and put it in reserve. You then withdraw 5.33% of 750k instead, that is 40k= 750k*5.33%. If the fund earns more than 5.33%, the reserve builds for future year. If it earns less, you use the reserve to reduce the impact of the loss on your pension. Notice that a 5.33% withdrawal on 75% of the fund really implies a 4.33% withdrawal on the total fund. For clarity, ADD withdrawal rate will means the withdrawal rate on the non-reserve account and the effective withdrawal rate will means the withdrawal rate on the whole account.
While we could cover 100% of the past withdrawal doing a floor approach, this strategy let the withdrawal go down if market fails. This way, we capture the most of the benefit of having a floating withdrawal rate, while managing the negative consequence.
Faced with the latest great recession, you would have been able to pull from these reserves during these harsh times. In the end, it’s an accounting trick. The retiree do not create any new money. It applies a “filter” over the portfolio. It helps better manage the volatility and unpredictability of the fund.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:23 pm

The reserve in example:
25% reserve covering 95% drop using an ADD withdrawal rate of 8%:
Image

legend:
Market Value: MV; Reserve: R; Non Reserve: NR; BoY: Begin of Year; EoY: End of Year

As expected in the financial crisis, the fund substantially shrinks during the 4 years. However, we are able to maintain a 95% of the initial withdrawal with relative ease. We can see that the fund value increase substantially after the fabulous year of 2009 & 2010. The 2011 return is 2%, less than the effective withdrawal rate (8%*75% = 6%). This lead to the Non-Reserve fund shrunk in 2012. The reserve would be able to withstand about 10 years at the end of 2012 regardless of future recovery.
[edit : adjusted the date since the crisis was in 2008 not 2011 :oops: ]
Last edited by icetrap on Wed Mar 14, 2018 12:37 am, edited 1 time in total.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:25 pm

For the whole parameter Sensitivity, I would refer to my web site (sorry) where a link can be found of the whole study.

Thanks. Look under Model&Strategy section; the sub-section parameter under each 3 sub-strategy.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Mar 09, 2018 9:49 pm

AlohaJoe wrote:
Fri Mar 09, 2018 3:39 am
Let me ignore the "reserve" fund for now. Here's how the "ADD minus reserve" (ADD-reserve) compares to some alternative strategies.

I believe that the reserve allows for more agressive withdrawal. The "Magic Number" :o I've got is to use a 25% reserve. It's quite high. So out of 1,000,000 you take 250,000 away in that "reserve fund" .

However, with that reserve in place you can push on the aggresive side of the withdrawal on the rest. This is compensated by taking instead out of 75% of the remaining fund. so a 8,67% withdrawal on the "regular reserve" which is super high in a VPW, is only like taking out 6,5%.

So if we compare at various withdrawal rate, we can see a nice thing happening in the strategy. The lower the withdrawal rate, the lower the Initial Cash Flow(ICF) :shock: . However, the Average Cash Flow (ACF) in the following 40 years are about the same. I hope this helps in the efficiency category :)

Image

User avatar
AtlasShrugged?
Posts: 613
Joined: Wed Jul 15, 2015 6:08 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by AtlasShrugged? » Sat Mar 10, 2018 8:39 am

There's also a whole huge thread on McClung's book with way more information and arguments :)
AlohaJoe....Oh yeah. That thread is a real doozy. A lot of good stuff in that thread. It rivals the famous 'Sheepdog' thread. It should be required reading for any new Boglehead. :happy

One other thing....VPW is such a standout in terms of efficiency. It really is an elegant solution. The only drawbacks I see are that you need to pick when you'll die (so as not to run out of money), and not have a bequest motive. I can mitigate both by picking 110 as the age [not likely to make it to the age of Joseph, heh heh].
“If you don't know, the thing to do is not to get scared, but to learn.”

User avatar
Johnnie
Posts: 512
Joined: Sat May 28, 2016 3:18 pm
Location: Michigan

Re: Accumulation-Dynamic Decumulation Strategy

Post by Johnnie » Sat Mar 10, 2018 9:39 am

longinvest wrote:
Fri Mar 09, 2018 12:39 am

This method seems to be aiming at developing a kind of personal pension fund paying a low-fluctuation pension. It differs from a traditional pension fund in that it won't benefit from continuous contributions of working members while paying pensions to retired members...
Minor threadjack but I feel compelled to point out that's not how pensions are supposed to work (although it is in fact the way all too many do work).

Specifically, the annual pension contribution amount for each active worker (the "normal cost)" is supposed to fully finance the pension credits the worker earned during that year. If the contribution is getting low-balled because of rose-colored assumptions in the normal cost calculation, that's mismanagement that leaves the system underfunded.

There's a term for systems that plan to back-fill their accumulated unfunded liabilities by enrolling new members: Ponzi scheme.

</end threadjack>
"I know nothing."

longinvest
Posts: 3109
Joined: Sat Aug 11, 2012 8:44 am

Re: Accumulation-Dynamic Decumulation Strategy

Post by longinvest » Sat Mar 10, 2018 9:53 am

Off topic:

AtlasShrugged?,
AtlasShrugged? wrote:
Sat Mar 10, 2018 8:39 am
One other thing....VPW is such a standout in terms of efficiency. It really is an elegant solution. The only drawbacks I see are that you need to pick when you'll die (so as not to run out of money), and not have a bequest motive. I can mitigate both by picking 110 as the age [not likely to make it to the age of Joseph, heh heh].
Variable percentage witdrawal (VPW) is only a tool to use within a larger retirement plan. It is best combined with stable lifelong non-portfolio income, like Social Security, a pension (if any), and (if necessary) an inflation-indexed Single Premium Immediate Annuity (SPIA).

At age 80, the payout rate on inflation-indexed SPIAs is competitive with VPW table percentages. So, if the retiree is still alive, it is recommended that he liquidates part (not all!) of his residual portfolio to buy enough inflation-indexed SPIA so that he can live comfortably on his total non-portfolio income, without taking any portfolio withdrawals. This completely eliminates what is usually called longevity risk (the risk of having more life than money).

At age 90, VPW theoretically aims for the residual portfolio oscillate (very widely) around 40% of what it was at age 65. This can represent a significant amount of money, depending on the luck of retirement year. Assuming that the retiree has liquidated 50% of his residual portfolio at age 80 to buy an inflation-indexed SPIA, and that the retiree continued to make VPW withdrawals until age 90, the portfolio's residual value would oscillate (very widely) around 20% of what it was at age 65. This can still be a significant amount of money, especially if the goal it to bequeath it to children who are now in their 60s and might have no more need for it.

I think that it's a good idea for a younger retiree, in his 60s and 70s, to take advantage of larger than necessary VPW withdrawals, after good market years, to give away the excess withdrawal money to his children in their 30s and 40s, helping them with their mortgage and with grand-children related expenses.

Anyway, if one wants to leave an even bigger bequest, when dying in old age, one can simply use constant-percentage withdrawal with part of the portfolio, during retirement. There's simply no need to change VPW; using a retirement plan is the key.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic / international) stocks / domestic (nominal / inflation-indexed) long-term bonds | VCN/VXC/VLB/ZRR

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Sat Mar 10, 2018 3:24 pm

Johnnie wrote:
Sat Mar 10, 2018 9:39 am
longinvest wrote:
Fri Mar 09, 2018 12:39 am

This method seems to be aiming at developing a kind of personal pension fund paying a low-fluctuation pension. It differs from a traditional pension fund in that it won't benefit from continuous contributions of working members while paying pensions to retired members...
Minor threadjack but I feel compelled to point out that's not how pensions are supposed to work (although it is in fact the way all too many do work).

Specifically, the annual pension contribution amount for each active worker (the "normal cost)" is supposed to fully finance the pension credits the worker earned during that year. If the contribution is getting low-balled because of rose-colored assumptions in the normal cost calculation, that's mismanagement that leaves the system underfunded.

There's a term for systems that plan to back-fill their accumulated unfunded liabilities by enrolling new members: Ponzi scheme.

</end threadjack>
I'll bite :) I would say that a major advantage of "DB" is that it spread the longevity risk accross all the group. So yes, the annual pension contribution should in the end pay for most of it. If the member dies early, he funds the other. If the member dies late he gets more out of it. However, ADD do not have the other members to help him out, same as any other Managed decumulation scheme. If he dies late, he have no "group" to fund the excess.

2015
Posts: 2163
Joined: Mon Feb 10, 2014 2:32 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by 2015 » Sat Mar 10, 2018 8:52 pm

AtlasShrugged? wrote:
Sat Mar 10, 2018 8:39 am
There's also a whole huge thread on McClung's book with way more information and arguments :)
AlohaJoe....Oh yeah. That thread is a real doozy. A lot of good stuff in that thread. It rivals the famous 'Sheepdog' thread. It should be required reading for any new Boglehead. :happy

One other thing....VPW is such a standout in terms of efficiency. It really is an elegant solution. The only drawbacks I see are that you need to pick when you'll die (so as not to run out of money), and not have a bequest motive. I can mitigate both by picking 110 as the age [not likely to make it to the age of Joseph, heh heh].
I would agree. I am repeatedly floored at the almost obsession in many threads with adding complexity to investing when none is warranted, and when such complexity is dangerous more often than not. A wide variety of fields outside economics, investing, and personal finance warn of the dangers of complexity in complex, adaptive, systems (such as economics, investing, and personal finance). The propensity for behavioral finance violations multiplies exponentially under such conditions, particularly in open, chaotic systems. I've never understood how people could conflate investing, which is an open system whose outcomes are intimately tied to luck, with closed systems (such as chess), whose outcomes are related to skill. In fact, if you're doing well in most areas of life, it has more to do with luck than anything else, the underlying causes, consequences and unintended consequences of which are always opaque at best. A huge dose of humility is caused for.

As to VPW, I've personally found it to be one of the most simple, robust, efficient, and elegant tools on this board. I am grateful to longinvest for his generosity, almost saintly patience, and commitment to answering some of the same questions over and over in various threads. Such (admittedly trying) repetition has added greatly to my understanding of VPW and to my commitment to continue to incorporate it into my own strategies.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Mon Mar 12, 2018 2:06 am

2015 wrote:
Sat Mar 10, 2018 8:52 pm
I would agree. I am repeatedly floored at the almost obsession in many threads with adding complexity to investing when none is warranted, and when such complexity is dangerous more often than not.
ADD is not an investing strategy. It's a decumulation strategy. For investing, I agree that it's more luck than anything else to find the best stock or what not.

I admit that ADD is more complex than VPW. Perhaps ADD can be simplified also.

The ADD strategy tries to manage those risk and reduce them.

Does anyone have an idea on how much they are willing to lose from a previous withdrawal? SWR keeps the withdrawals constants (increase by inflation). So it never reduce withdrawal.

From what I gather from my test, you can expect the following hardship from the VPW withdrawal :
Image

I first thought that most pensionner would rather see minimal withdrawals drops (95%). Do you think that they are willing to face 90% or more like the VPW implies?

2015
Posts: 2163
Joined: Mon Feb 10, 2014 2:32 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by 2015 » Mon Mar 12, 2018 9:15 pm

icetrap wrote:
Mon Mar 12, 2018 2:06 am
2015 wrote:
Sat Mar 10, 2018 8:52 pm
I would agree. I am repeatedly floored at the almost obsession in many threads with adding complexity to investing when none is warranted, and when such complexity is dangerous more often than not.
ADD is not an investing strategy. It's a decumulation strategy. For investing, I agree that it's more luck than anything else to find the best stock or what not.

I admit that ADD is more complex than VPW. Perhaps ADD can be simplified also.

The ADD strategy tries to manage those risk and reduce them.

Does anyone have an idea on how much they are willing to lose from a previous withdrawal? SWR keeps the withdrawals constants (increase by inflation). So it never reduce withdrawal.

From what I gather from my test, you can expect the following hardship from the VPW withdrawal :
Image

I first thought that most pensionner would rather see minimal withdrawals drops (95%). Do you think that they are willing to face 90% or more like the VPW implies?
Your take on VPW is perplexing. Have you read all of longinvest's posts in many threads here explaining it? I recommend reading this thread from start to finish. Longinvest was quite generous with his time in answering many questions regarding VPW.
Fill the gap in Social Security payments between retirement and age 70 using a non-rolling TIPS ladder. It is important to exclude this non-rolling ladder from the portfolio used for variable withdrawals; this non-rolling ladder is part of the lifelong non-portfolio stable inflation-indexed income.
viewtopic.php?p=3693510#p3695541

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Mon Mar 12, 2018 10:14 pm

To better understand Accumulation-Dynamic Decumulation here's some graphics and comparison with VPW.

At the "risky" fully equity portfolio, invested in S&P500 (100% equity), here's what ADD produces with an withdrawal rate of 6.5%*. As you can see there’s still some bump in all the lines. Most of them go up. Since 6,5% is rather aggressive, most run encounter a trouble period and stagnate for an extensive period of time. Given inflation, the effective available money goes down slowly. Some of the more noticeable line are mostly flat for the whole time like the run 1928, 1932 and 2004.

Image

VPW in the other end expect great swings at 100% equity.
Image

Here's what the 5,5% ADD withdrawal rate with a more conservative portfolio 50% equity/ 50% bonds.
Image

VPW at 50% equity/ 50% bonds the swings are less wide, but still noticeable.
Image

*: Note that due to smoothing, capping the "effective withdrawal rate is more like in the 5,3% range on average (min 3.1% - max 7.8%).

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 14, 2018 12:14 am

AlohaJoe wrote:
Fri Mar 09, 2018 3:39 am
Let me ignore the "reserve" fund for now. Here's how the "ADD minus reserve" (ADD-reserve) compares to some alternative strategies.

This is Blanchett's WER (Withdrawal Efficiency Rate):
Thanks for sharing it to me :D Knowledge is power :greedy

I've review Blanchett's WER paper. I am uncertain of the calculation behind the "Certainty Equivalent withdrawal" aka CEW.
CEW = ((1/N)*y*sum((ci^-y)/y)^(1/y)

Shouldn't it be rather CEW = ((1/N)*y*sum((ci^-y)/y)^(1/-y) ?
At first I tried with the actual cash flow(ci) I got using my standard accumulation scenarios. Using Cash flows of 70k doesn't seems to work. I understand that we should rather use c="Cash Flowi/Market Value at DoR" is that right?

Using these changes I get a respectfull 82% WER at 100% equity (6,5% ADD) and 75% at 50% equity to bonds(5,5% ADD). I'm guessing I did an error in the formula above since most of the other method only got 70% or less. Note that I didn't do the whole Monte Carlo on N. I'm unsure if it will have much effect to move it around.

Personnaly, I disagree with the " (we purposely disregard here the bequest motive, which in any case is secondary for most retirees)" as part of the WER objective. It will take me a while to built the wealth, why not make it benefits to others. Most people live like if there will never been a generation afterward. I like the idea behind the indie game "the talos principle". It is based on the "learning machine" theory. After an almost infinite version, the robot learn on the failure of the previous robot version. Some robot earlier version have agreed to stay in the simulation to help others solve the puzzle. I believe that we should make the best to help other future human. Too many are focus on the present, but I bet future generation will face just as dire problems as us. Anyway enough rambling off topic, I agree that it's likely a secondary objective for most retirees.

AlohaJoe
Posts: 3923
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Accumulation-Dynamic Decumulation Strategy

Post by AlohaJoe » Wed Mar 14, 2018 12:22 am

icetrap wrote:
Wed Mar 14, 2018 12:14 am
I am uncertain of the calculation behind the "Certainty Equivalent withdrawal" aka CEW.
CEW = ((1/N)*y*sum((ci^-y)/y)^(1/y)

Shouldn't it be rather CEW = ((1/N)*y*sum((ci^-y)/y)^(1/-y) ?
The formula does show 1/-y, though I guess the minus sign is small and easily missed.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 14, 2018 12:38 am

AlohaJoe wrote:
Wed Mar 14, 2018 12:22 am
icetrap wrote:
Wed Mar 14, 2018 12:14 am
I am uncertain of the calculation behind the "Certainty Equivalent withdrawal" aka CEW.
CEW = ((1/N)*y*sum((ci^-y)/y)^(1/y)

Shouldn't it be rather CEW = ((1/N)*y*sum((ci^-y)/y)^(1/-y) ?
The formula does show 1/-y, though I guess the minus sign is small and easily missed.
Yup Just saw it.. I confuse it for a long division bar :oops:

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Accumulation-Dynamic Decumulation Strategy

Post by getrichslowly » Wed Mar 14, 2018 10:09 am

Interesting strategy. To review any strategy, can you also publish quantifiable metrics contrasting it with other methods?

I suggest starting by defining a utility function, such as

Code: Select all

max{E[U]} = sum{ u(c) * b^t }
 u(c) = log(c)
 b<1
of course. Then find which solves the problem.

For instance, does that arbitrary 25% "reserve" fund raise or lower the objective function? Then you can iterate over different levels and find the optimal "reserve" parameter to maximize.

For simulating the probability distribution, I suggest a combination of historical backtesting and various monte carlos.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 14, 2018 10:57 am

getrichslowly wrote:
Wed Mar 14, 2018 10:09 am
Interesting strategy. To review any strategy, can you also publish quantifiable metrics contrasting it with other methods?

I suggest starting by defining a utility function, such as

Code: Select all

max{E[U]} = sum{ u(c) * b^t }
 u(c) = log(c)
 b<1
of course. Then find which solves the problem.

For instance, does that arbitrary 25% "reserve" fund raise or lower the objective function? Then you can iterate over different levels and find the optimal "reserve" parameter to maximize.

For simulating the probability distribution, I suggest a combination of historical backtesting and various monte carlos.
what is b? 0.5, 0.03? I guess that sum of Log(c) basically tries to get Sum of cash flow (adjust for lower utility for high return). The b^t put more weight on the earlier cash flow. I like that utility function. Currently I've only implemented SWR, VPW, ADD. I'll try to compare the three using this metric once I know what b is :wink:

A lot of work in front of me :happy

Should the c be in nominal or real terms?

getrichslowly
Posts: 73
Joined: Mon Feb 12, 2018 11:48 am

Re: Accumulation-Dynamic Decumulation Strategy

Post by getrichslowly » Wed Mar 14, 2018 12:24 pm

icetrap wrote:
Wed Mar 14, 2018 10:57 am
getrichslowly wrote:
Wed Mar 14, 2018 10:09 am
Interesting strategy. To review any strategy, can you also publish quantifiable metrics contrasting it with other methods?

I suggest starting by defining a utility function, such as

Code: Select all

max{E[U]} = sum{ u(c) * b^t }
 u(c) = log(c)
 b<1
of course. Then find which solves the problem.

For instance, does that arbitrary 25% "reserve" fund raise or lower the objective function? Then you can iterate over different levels and find the optimal "reserve" parameter to maximize.

For simulating the probability distribution, I suggest a combination of historical backtesting and various monte carlos.
what is b? 0.5, 0.03? I guess that sum of Log(c) basically tries to get Sum of cash flow (adjust for lower utility for high return). The b^t put more weight on the earlier cash flow. I like that utility function. Currently I've only implemented SWR, VPW, ADD. I'll try to compare the three using this metric once I know what b is :wink:

A lot of work in front of me :happy

Should the c be in nominal or real terms?
This is basic economics you'll see in academic journals.

c should be in real terms. c is real consumption.

b is an intertemporal discount rate to emphasize earlier cash flows. It's whatever you want. .96 might be an average value but I'm not sure. It probably falls as you age and you can even make it a function of time based on mortality rates and overall quality of life. The value of b will determine how aggressive your withdrawal strategy will be. It's entirely possible the optimal strategy has a chance of bankruptcy or at least severe spending reductions, and that's okay if the pleasure gained while young outweighs the suffering incurred while old.

But this is also an extremely simplistic model and doesn't account for more advanced utility considerations such as hedonic adapation. Under hedonic adapation, utility is more closely tied to the rate at which your consumption is increasing, not the absolute level. Under this model it becomes optimal to "space out" your consumption increases to slow adaptation. Even if you can sustain a high spending level, if you jump to your maximum sustainable spending level immediately, you quickly adapt, and then stagnate and become miserable. So it might be better to start low and gradually ramp up consumption over time. This sort of flies in the face of the intertemporal discount parameter which has the opposite effect of pushing you to frontload consumption. It's also consistent with the experience that people get a rush of joy when they receive a regular raise, more than if they were to have ten years of raises frontloaded and then flat.

The more I think about this, the more I realize so much of this is going to be specific to the individual. People don't always have a well-defined utility function but it usually has a series of discrete jumps and kinks. But the smooth, idealized function may be a reasonable simplification to make the model tractable.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 14, 2018 11:32 pm

getrichslowly wrote:
Wed Mar 14, 2018 10:09 am
Interesting strategy. To review any strategy, can you also publish quantifiable metrics contrasting it with other methods?

I suggest starting by defining a utility function, such as

Code: Select all

max{E[U]} = sum{ u(c) * b^t }
 u(c) = log(c)
 b<1
of course. Then find which solves the problem.

For instance, does that arbitrary 25% "reserve" fund raise or lower the objective function? Then you can iterate over different levels and find the optimal "reserve" parameter to maximize.

For simulating the probability distribution, I suggest a combination of historical backtesting and various monte carlos.
I'll try to maximize some more. Currently all the parameters are based to minimize the Risk II : 80% or more reduction of withdrawals compare to any prior withdrawals. I believe that the goal of a withdrawal strategy is to reduce negative volatility in withdrawals rather then to maximize spendings. Even if less efficient, SWR never reduces the withdrawals. I've therefore also put in this utility table the risk II

Here's some metric in nominal terms. I dislike going with real terms since I find that inflation seems to be less of an issue according to projection and commitment of most developped banks to keep inflation low. Even in the past, I've got a hard time believing that a "SWR" retiree would feel safe withdrawing 10% per year of his fund to "follow the inflation increase" while is fund are shrinking.

Image

Even if it have little effect in all the metrics, all like the smoothing since it reduce the "retirement date paradox" where a 2007 can retire with a way higher amount than the 2008 retiree. Current market change should not alter your withdrawal strategy.

Note that both WER and sum{ u(c) * b^t } were made using the average result over 78 retirement periods using end dates 20 years from DoR, 30 and 41.

Thanks again for the help. Note that it was the first time I've used WER or that utility function, I may have made several mistakes.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Variable Percentage Withdrawal (VPW)

Post by icetrap » Wed Mar 21, 2018 11:58 am

[Posts moved to here from: Variable Percentage Withdrawal (VPW), see below. --admin LadyGeek]

I've played around with the concept of adding my Accumulation-Dynamic Decumulation strategy over the VPW instead of the CPW. I've found some interesting results for those who would like to manage their risk of volatile withdrawal while maintaining for the most part the curve and initial cash flow of VPW.

I've adjusted the VPW by 140% of its original withdrawal rate and reduced the reserve protection from the ADD to 90%. It produces the following cash flow over 1928(worst), 1966(not good) and 1980(good) over 100% equity (100% S&P500) and over 50% equity and 50% bonds.

I've used a scnerario where a person invested 150,000$ 20 years prior retirement and 5,000$ per year increased by 3% each year up to retirement.

Here's what happens (000's):
At 100% equity :
Image
At 50% equity / 50% bonds :
Image

Its very probable that other setting may help further your goal. See the full study ofADD here.

On a side note, the 25% reserve can also be seen as a neat emergency fund. if you use it, you simply loose the safety it provides.

Enjoy :D

PhilosophyAndrew
Posts: 456
Joined: Sat Aug 13, 2016 10:06 am

Re: Variable Percentage Withdrawal (VPW)

Post by PhilosophyAndrew » Wed Mar 21, 2018 12:27 pm

If you are multiplying the VPW rate by some factor but dividing the total assets by a different factor, could you accomplish simply by modifying the withdrawal rates included in VPW?

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Variable Percentage Withdrawal (VPW)

Post by icetrap » Wed Mar 21, 2018 1:20 pm

PhilosophyAndrew wrote:
Wed Mar 21, 2018 12:27 pm
If you are multiplying the VPW rate by some factor but dividing the total assets by a different factor, could you accomplish simply by modifying the withdrawal rates included in VPW?
ADD have Dynamic withdrawal rate... they fluctuates quite a bit. For instance using retirement date= 1928 in 100% equity, the VPW(x140%) + ADD = 4,62%, 3,68%, 4,23%, 6,22%, 10,25%, 12,8%, 9,54%... vs VPW(40 years) = 5,5%, 5,6%, 5,6%, 5,7%, 5,7%, 5,8%, 5,8%, 5.9%,...

So in short, no :( , no unique factor could be used. It depends on the return received. ADD increase cap builds a provision for adverse deviation using the "very good return" for future "very bad return". This is why the shape of the curve is mostly increasing smoothly in 1966 and 1980.

1924-1928 had an average of 28,5% Rate of Return. ADD sees that as an opportunity to add provision for the winter that may come after.

PhilosophyAndrew
Posts: 456
Joined: Sat Aug 13, 2016 10:06 am

Re: Variable Percentage Withdrawal (VPW)

Post by PhilosophyAndrew » Wed Mar 21, 2018 1:24 pm

Thanks for the prompt reply. If you had time to post a step-by-step account of how one could use ADD+VPW, that could be extremely useful.

I’ve read the blog posts, thread, and additioal always you linked to, but remain unclear about exactly how one would implement this idea....

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 21, 2018 1:48 pm

PhilosophyAndrew wrote:
Wed Mar 21, 2018 1:24 pm
Thanks for the prompt reply. If you had time to post a step-by-step account of how one could use ADD+VPW, that could be extremely useful.

I’ve read the blog posts, thread, and additioal always you linked to, but remain unclear about exactly how one would implement this idea....
Step 0 : Create a 25% fund reserve; Out of 1 million, put 250,000$ in reserve fund; put 750,000$ in regular fund
Step 1 : Smooth the Market Value
Step 2: Cap regular withdrawal (115% prior retirement, 108% at retirement, 105% after retirement)
Step 3: Withdraw money from regular + withdraw money from reserve (if regular withdrawal below 90% threshold)
Repeat 1-3 each year

The step-by step are pretty much the same as ADD using CPW, but add a twist using the VPW rates.

Ex using 95% protection and 140% VPW :
Background Data: A few years into retirement.
Total fund value : 419,483$
- Regular fund value : 261,174$
-- Smoothed Regular Fund: 277,363$
- Reserve fund value : 158,309$
Last year withdrawal : 16,498$
This was not due to the increase cap.
Greatest previous withdrawal : 19,618$
ADD withdrawal:

Total withdrawal : 18,637$ = 17,323$ + 1,314$
Regular withdrawal : 17,323$ = Min (24,038 ; 17,323$)
Smoothed withdrawal prescribed : 24,038$ = 8.67% * 277,363$
[Parameter] ADD+VPW withdrawal rate : look at VPW recommended withdrawal % * 140% (ex= 6.5%)
This drives the “prescribed withdrawal rate” to 8.67% = 6.5%/(75%). The
Increase Cap: 17,323$ = 16,498$ * (105% + 0%)
• [Parameter] Increase Cap: 105%.
• [Parameter] Increase Cap is augmented if prior year was cap: +0%.
Reserve withdrawal : 1,314$ = Min(18,637$ - 17,323$; 31,662$)
 Maximum covered withdrawal : 18,637$ = 95% * 19,618$
• [Parameter] Reserve protected : 95%
• [Parameter] Reserve maximum withdrawal rate : 20%
o Maximum withdrawal: 31,662$ = 20%* 158,309$
• No adjustment required since last years and current year reserve fund =158,309/419,483 = 38% > 20%
• In fact, the reserve exceeds the 30% threshold and will return back to the regular fund the excess 158,309 - 30%*419,483= 32,464. This can be though as a regular contribution.
Last edited by icetrap on Mon Apr 09, 2018 6:34 am, edited 1 time in total.

User avatar
LadyGeek
Site Admin
Posts: 49171
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Accumulation-Dynamic Decumulation Strategy

Post by LadyGeek » Wed Mar 21, 2018 2:00 pm

I have moved icetrap's discussion from Re: Variable Percentage Withdrawal (VPW) into here (with a link back to this thread).

This combined thread is in the Personal Finance (Not Investing) forum (retirement planning).
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.

PhilosophyAndrew
Posts: 456
Joined: Sat Aug 13, 2016 10:06 am

Re: Accumulation-Dynamic Decumulation Strategy

Post by PhilosophyAndrew » Wed Mar 21, 2018 2:27 pm

Thanks, icetrap -- I'll see if I can understand better as I work through this.

One immediate question: Is the reserve fund held in cash or cash equivalent, or is it just a mental bookkeeping partition of one's regular asset allocation?

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 21, 2018 2:50 pm

PhilosophyAndrew wrote:
Wed Mar 21, 2018 2:27 pm
Thanks, icetrap -- I'll see if I can understand better as I work through this.

One immediate question: Is the reserve fund held in cash or cash equivalent, or is it just a mental bookkeeping partition of one's regular asset allocation?
It is just a "mental" bookkeeping. Honestly, it's easy to set up a seperate account and put the money there for accounting purpose. 25% of the fund is quite a substantial portion of the asset. My testing showed that reserve in a higher equity allocation helps lower the volatility risk, but I don't see why you would do that. If you are unconfortable using a higher equity allocaion for your portfolio, why increase it in the reserve. In brief, you should put it in the same asset allocation.

For example the 1980 retiree would see his fund grow for 20 years in a row. If the reserve is put in a "cash or cash equivalent", the reserve will get lower and lower over time versus the regular fund and lose it "safety' it provides.

In the end, the whole reserve scheme is just an accounting trick. No money is created or protected by a third party. If you are confortable with a 75% equity to bond portfolio than you should keep it :)

Ms.Fortuna
Posts: 15
Joined: Tue Dec 26, 2017 11:12 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by Ms.Fortuna » Wed Mar 28, 2018 12:49 pm

icetrap wrote:
Mon Mar 12, 2018 10:14 pm

At the "risky" fully equity portfolio, invested in S&P500 (100% equity), here's what ADD produces with an withdrawal rate of 6.5%*. As you can see there’s still some bump in all the lines. Most of them go up. Since 6,5% is rather aggressive, most run encounter a trouble period and stagnate for an extensive period of time. Given inflation, the effective available money goes down slowly. Some of the more noticeable line are mostly flat for the whole time like the run 1928, 1932 and 2004.
Would this work for longer retirements, say 50+years? how does the 6.5% (starting rate?) change for someone who targets to retire at 47-50?
The strategy seems appealing to an aspiring early retiree with a smaller expected Social Security benefits for a "floor" as in the VPW model.

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 28, 2018 2:34 pm

Ms.Fortuna wrote:
Wed Mar 28, 2018 12:49 pm
Would this work for longer retirements, say 50+years? how does the 6.5% (starting rate?) change for someone who targets to retire at 47-50?
The strategy seems appealing to an aspiring early retiree with a smaller expected Social Security benefits for a "floor" as in the VPW model.
Yes, ADD+CPW works for longer retirement period over 50+++ years. Withdrawal would slowly grows (real) in time the lower the ADD withdrawal rate.
  • At 6.5% 100% equity, after 40 years, the backtested fund value would have grown 700% (nominal) and the withdrawal (real) by about 160%.
  • At 5.5% 100% equity, after 40 years, the backtested fund value would have grown 1100% (nominal) and the withdrawal (real) by about 230%.
  • At 6.5% 50% equity, after 40 years, the backtested fund value would have grown by 310%(nominal) and the withdrawal (real) reduced by 70%.
  • At 5.5% 50% equity, after 40 years, the backtested fund value would have grown by 500%(nominal) and the withdrawal (real) by 0%.
  • At 4.5% 50% equity, after 40 years, the backtested fund value would have grown by 650%(nominal) and the withdrawal (real) by 140%.
The ADD withdrawal rate should not change due to the age of retirement in ADD+CPW. Rather, it should be set in accordance to the risk you are willing to take in the volatility of withdrawal. Can the retiree handle 80% drop in withdrawal or only 95% drop in withdrawal. Of course the tradeoff risk vs reward can lead to other outcome. Finally, if the future holds lower Rate of Return or Japan like scenario, lower rate would reduce the risk of future drop in withdrawal.

The "6.5%" ADD withdrawal rate is the "desired withdrawal rate". I've failed to get a better approximation for now. It's not the starting rate. The reserve, smoothing and cap all work to decrease the withdrawal rate. We take 6.5%/(1-25%) out of the regular smoothed fund. This adjustment "should be" therefore close to what you expect to take from the total fund(75% regular fund; 25% reserve fund). However, overall due to strategies, a 6.5% withdrawal rate would result on average at Retirement date to an effective withdrawal rate (on the total fund) of 5.3%, minimum of 3.1%, maximum 7.8%. That's why I say it's a dynamic strategy 8-)

I agree with you that ADD appeal more if you have low social security/high floor limit. It should also appeal to those who would like to reduce the volatility of withdrawal that VPW or CPW provides and dislike to have risk of ruin that a CDW(constant dollar Withdrawal; like the classic Safe Withdrawal Rate). It certainly would be more beneficial for high equity portfolio which have more volatility of withdrawal in VPW or CPW, but the risk reduction also affects low equity portfolio as well. Let's remember that VPW also face 80% reduction on withdrawal on virtually all of its historical run at 50% equity allocation.

Note that my last post concerned with ADD+VPW which by definition depends on the VPW "end date". If you reassert the "end date" it may carry on longer than 50+ years, but it's less certain. ADD+CPW is better if the objective do not contain an "end date"/desire to have 0$ at the end date.

Ms.Fortuna
Posts: 15
Joined: Tue Dec 26, 2017 11:12 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by Ms.Fortuna » Wed Mar 28, 2018 3:09 pm

Thank you for the quick response!
I need to study this more, at the first glance - it is a little over my head :shock:
I am getting the general idea though..definitely very interested in this strategy. It also seems to me that it would especially work well with a 100% equity portfolio which I think is better for longer retirements anyway.
icetrap wrote:
Wed Mar 28, 2018 2:34 pm

Note that my last post concerned with ADD+VPW which by definition depends on the VPW "end date". If you reassert the "end date" it may carry on longer than 50+ years, but it's less certain. ADD+CPW is better if the objective do not contain an "end date"/desire to have 0$ at the end date.
Can you please point to that post about end date? Sorry if I overlooked it in this this thread..

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Wed Mar 28, 2018 3:40 pm

Ms.Fortuna wrote:
Wed Mar 28, 2018 3:09 pm
Can you please point to that post about end date? Sorry if I overlooked it in this this thread..
Sorry, VPW requires to input the "start date" see VPW wiki basically it requires you to input your "Starting Age" in combination with a fixed "end age" of 99 years. At age 39, that means a 50 years away end date. VPW withdraw 100% at the end age and about 20% per year in the last 5 years; about 10% in the 5 years prior. If you wanted to have a 100 years period for a new born baby, then you would have the Age 40 withdrawal rate for 40 years and after you would follow the table.

My point was that you need to be clear on the objective. If the objective is to 0$ after x years ADD+VPW would work best. However, if you are age 90 and taking 10% or more out of the fund, there's a high risk it won't last another 30 years without high volatility and reduction in $ withdrawal. That's why VPW thought is to revise the "end date" if you are starting to be concerned with this.

ADD+CPW(Constant Percentage Withdrawal) however tries to get "constant" percentage withdrawal. It does increase a bit in time, but not as much as ADD+VPW. ADD+CPW most likely would leave more money at death than at retirement date. VPW tries to go for no money at death; less money at death by increasing the percentage withdrawal through age.

Let's our objective decide what tool should be used in the tool box :)

Personally, I don't see the point to taking so much more withdrawal near death (age 90+). It looks like an attempt to destroy the most value before death without consideration of "old age" or bequest motives. I love however the increase of rate in the (70-80 periods) to try to maximize withdrawal while useful. Perhaps a moving "end date" would be best for this purpose to keep the increase. In the end, its just a plan. Probably life will decide otherwise :o

Feel free to ask me any other questions if you want :) I've started doing a tool similar to the VPW tool for the casual user to use and have fun with. I'll update it here once finish.

Ms.Fortuna
Posts: 15
Joined: Tue Dec 26, 2017 11:12 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by Ms.Fortuna » Wed Mar 28, 2018 4:06 pm

icetrap wrote:
Wed Mar 28, 2018 3:40 pm

Personally, I don't see the point to taking so much more withdrawal near death (age 90+). It looks like an attempt to destroy the most value before death without consideration of "old age" or bequest motives. I love however the increase of rate in the (70-80 periods) to try to maximize withdrawal while useful.
Agree!
icetrap wrote:
Wed Mar 28, 2018 3:40 pm

I've started doing a tool similar to the VPW tool for the casual user to use and have fun with. I'll update it here once finish.
Looking forward to it!

icetrap
Posts: 74
Joined: Mon Nov 06, 2017 12:30 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by icetrap » Fri Apr 06, 2018 12:12 pm

Ms.Fortuna wrote:
Wed Mar 28, 2018 4:06 pm
[Looking forward to it!
Here's the Google Drive and DropBox links for the spreadsheets with the strategy.

I hope it's free of any issues :)

Have fun with the advanced parameters if you wish to adjust.
Last edited by icetrap on Mon Apr 16, 2018 11:48 am, edited 2 times in total.

Ms.Fortuna
Posts: 15
Joined: Tue Dec 26, 2017 11:12 pm

Re: Accumulation-Dynamic Decumulation Strategy

Post by Ms.Fortuna » Sat Apr 07, 2018 8:05 am

Thank you! Will check it out!

Post Reply