Retirement (Decumulation) Specific Portfolio

JohnWinder

On average you are correct and all equity portfolios will support the highest withdrawal levels

What that means to me is that the all equity portfolio is likely to allow the most money to be withdrawn. Well, it should because if the past is any guide it has meant the portfolio will have made more money during retirement than its cousin with bonds. But with all equities you have to be more fussy about how much you can withdraw, and when you can, because as we've seen from firecalc if you withdraw 4%/year the all equity and its bonds cousin give the same withdrawal levels (4%), but the all equity will run out first if they both hit a bad 30 year period. So that wouldn't be the highest withdrawal level.

Pat38493

JohnWinder said:

Again, that's not what historic patterns show us.

I think we’d all benefit from seeing your data on this because it will enhance our understanding. The firecalc data indicates something different, with 70/30 having a lower failure rate (by 1.7%), fewer 30 year periods failing (7 rather than 9 out of 122), and a later time for first failure (by 1 year). So not really much difference by those measures between 100/0 and 70/30. But a lot more 100/0 portfolios finished up with >£3M. I’ll guess your data showing 100/0 is safer shows the difference in failing to be quite small, but we need see it.

I've seen similar claims for example in the various ERN blogs but in the end it will depend what withdrawal rate you are using and what you mean by 100% equities (which equities?).  With 100% equities you might have a slightly lower historically safe withdrawal rate but you have a much higher chance of increasing your withdrawals during retirement.  However 80% I think doesn't make much difference.  It's also better the longer you are planning - if you are an early retiree, you need to stay pretty high in equities otherwise you will need a massive pot.

billy2shots

JohnWinder said:

Again, that's not what historic patterns show us.

I think we’d all benefit from seeing your data on this because it will enhance our understanding. The firecalc data indicates something different, with 70/30 having a lower failure rate (by 1.7%), 

Using FI Calc. Matching equity and bond fees at 0.5% it shows the following. 

40 year retirement.
£900k starting pot

£29,375 drawdown each year

100% equities= 100% chance of success (lasting the full 40 years) 

3 out of 112 'nearly fails' (final pot size less than 35% of the original pot size)

58.4% chance of a large pot at the end of 40 years (greater than 300% of the starting pot). 

Vs

70% equities 30% bonds= 99.11% chance of success (lasting the full 40 years) 

1 fail

3 'nearly fails'

31.25% chance of a 'large end pot'

Vs the much talked about 60/40


60% equities, 40% bonds = 98.21% chance of success (lasting the full 40 years) 

2 fails

7 nearly fails

23.21% chance of a 'large end pot'

Reducing the same example to a 30 year retirement with 100% equities Vs 70/30  gives identical success rates (100%) and 'nearly fail' rates (2/122). 

A big difference in pot size left over 46.72% chance of a 'large end pot' for 100% equities Vs 29.51% for the 70/30

The famous 60/40 over 30 years keeps the same 100% success rate but adds an additional ' nearly fail' making that 3/122. 
It also drops the chance of a large pot to 17.21%

Albermarle

Vanguard have a monte carlo simulation tool. I have no idea how good it is. Playing around with it briefly, it indicates that you are a little bit less likely to run out of money in 30 years, with a 60:40 or 50:50 than 100% equities.
However you are much more likely to end up with a pot many times bigger than you started with, with 100% equities.
Nothing new there, but slightly concerning is that with 100% or 60:40, there is still about a 10% chance of running out in 30 years with a 4% drawdown ( in US) 
Vanguard - Retirement Nest Egg calculator

billy2shots

Albermarle said:

Vanguard have a monte carlo simulation tool. I have no idea how good it is. Playing around with it briefly, it indicates that you are a little bit less likely to run out of money in 30 years, with a 60:40 or 50:50 than 100% equities.
However you are much more likely to end up with a pot many times bigger than you started with, with 100% equities.
Nothing new there, but slightly concerning is that with 100% or 60:40, there is still about a 10% chance of running out in 30 years with a 4% drawdown ( in US) 
Vanguard - Retirement Nest Egg calculator

History no indicator and all that but I would rather base things on that rather than the Monte Carlo tool you posted. 

To get 100% success on that tool I had to get down to a 1.5% safe withdrawal rate. Stuff dieing with a 1000% size increase in pot whilst living like a homeless person. 

bostonerimus

JohnWinder said:

On average you are correct and all equity portfolios will support the highest withdrawal levels

What that means to me is that the all equity portfolio is likely to allow the most money to be withdrawn. Well, it should because if the past is any guide it has meant the portfolio will have made more money during retirement than its cousin with bonds. But with all equities you have to be more fussy about how much you can withdraw, and when you can, because as we've seen from firecalc if you withdraw 4%/year the all equity and its bonds cousin give the same withdrawal levels (4%), but the all equity will run out first if they both hit a bad 30 year period. So that wouldn't be the highest withdrawal level.

Yes, there are so many caveats and parameters in these models that comparisons become difficult and tedious; longevity, withdrawals, inflation, expenses, investment mixes and even stupidity. The danger for many people is that they use the models as a way to justify higher withdrawal rates on the basis of a simulation using historical data. They might set a failure probability at 5% and say that they are good at an inflation linked 4% initial withdrawal. That misses the potential that future markets won't be like historical ones, that inflation is higher than expected, that they do something silly ie panic and sell at a loss or that they just run into that 5% sequence of returns scenario. Probability distributions are all well and good until you collapse the statistics into the particular scenario of your own retirement and you hope that you aren't in the failure tail.

One major factor in whether a plan is sustainable is the withdrawal rate and a long time ago I decided that I would be frugal and save so much that I could live entirely on dividends and interest...a very old fashioned approach. I then decided to go one step farther and to use rental and DB/annuity type income thus largely decoupling from the stock and bond markets.

Kim1965

bostonerimus said:

JohnWinder said:

On average you are correct and all equity portfolios will support the highest withdrawal levels

What that means to me is that the all equity portfolio is likely to allow the most money to be withdrawn. Well, it should because if the past is any guide it has meant the portfolio will have made more money during retirement than its cousin with bonds. But with all equities you have to be more fussy about how much you can withdraw, and when you can, because as we've seen from firecalc if you withdraw 4%/year the all equity and its bonds cousin give the same withdrawal levels (4%), but the all equity will run out first if they both hit a bad 30 year period. So that wouldn't be the highest withdrawal level.

Yes, there are so many caveats and parameters in these models that comparisons become difficult and tedious; longevity, withdrawals, inflation, expenses, investment mixes and even stupidity. The danger for many people is that they use the models as a way to justify higher withdrawal rates on the basis of a simulation using historical data. They might set a failure probability at 5% and say that they are good at an inflation linked 4% initial withdrawal. That misses the potential that future markets won't be like historical ones, that inflation is higher than expected, that they do something silly ie panic and sell at a loss or that they just run into that 5% sequence of returns scenario. Probability distributions are all well and good until you collapse the statistics into the particular scenario of your own retirement and you hope that you aren't in the failure tail.

One major factor in whether a plan is sustainable is the withdrawal rate and a long time ago I decided that I would be frugal and save so much that I could live entirely on dividends and interest...a very old fashioned approach. I then decided to go one step farther and to use rental and DB/annuity type income thus largely decoupling from the stock and bond markets.

I know its frowned upon but, if you only spend the dividends, how is it possible to run down the pot? Surely it also avoids sequence of returns issues? Your approach makes perfect sense to me. Selling assets and rebalancing is also reduced some what, i think. 

JohnWinder

Firstly, I don’t think it ought to be frowned upon if it’s part of a sensible plan. 

Secondly, what’s providing the dividends? If it’s unfortunately chosen blue chip good dividend payers like Kodak or General Electric, then things can go from golden to brown in a few short years. That's how the pot runs down. If it’s dividends from a widely diversified holding, that's a horse of a different colour.

Thirdly, I suppose taking only dividends does avoid a SoR risk, but with the usual SoR thinking you worry about being forced to dispose of too much assets because prices are down but then recover; choosing Kodak or Enron, they never recover. That’s a different kettle of fish. You can eliminate a SoR risk by deciding to sell off the same x% or your portfolio each year, however much it’s fallen; the ‘cost’ is you have less to live on in the year of the crash, but SoR is a non-issue. In contrast, taking dividends only during the crash might mean they also are reduced as businesses have poorer earnings.

In de-cumulation, rebalancing can be done by selling which is what de-cumulation implies.

bostonerimus

JohnWinder said:

Firstly, I don’t think it ought to be frowned upon if it’s part of a sensible plan. 

Secondly, what’s providing the dividends? If it’s unfortunately chosen blue chip good dividend payers like Kodak or General Electric, then things can go from golden to brown in a few short years. That how the pot runs down. If it’s dividends from a widely diversified holding, thats’ a horse of a different colour.

Thirdly, I suppose taking only dividends does avoid a SoR risk, but with the usual SoR thinking you worry about being forced to dispose of too much assets because prices are down but then recover; choosing Kodak or Enron, they never recover. That’s a different kettle of fish. You can eliminate a SoR risk by deciding to sell off the same x% or your portfolio each year, however much it’s fallen; the ‘cost’ is you have less to live on in the year of the crash, but SoR is a non-issue. In contrast, taking dividends only during the crash might mean they also are reduced as businesses have poorer earnings.

In de-cumulation, rebalancing can be done by selling which is what de-cumulation implies.

My US equity index funds produce between 2% and 3% in dividend distributions each year. Right now I'm just reinvesting those dividends and using the DB pension and rental income to live on. If I was using the equity funds for income I'd be taking the dividends and maybe using a Guyton-Klinger type algorithm if I needed to sell something to top up the income. All this would be controlled by actively modulating my budget.

Audaxer

Kim1965 said:

bostonerimus said:

JohnWinder said:

On average you are correct and all equity portfolios will support the highest withdrawal levels

What that means to me is that the all equity portfolio is likely to allow the most money to be withdrawn. Well, it should because if the past is any guide it has meant the portfolio will have made more money during retirement than its cousin with bonds. But with all equities you have to be more fussy about how much you can withdraw, and when you can, because as we've seen from firecalc if you withdraw 4%/year the all equity and its bonds cousin give the same withdrawal levels (4%), but the all equity will run out first if they both hit a bad 30 year period. So that wouldn't be the highest withdrawal level.

Yes, there are so many caveats and parameters in these models that comparisons become difficult and tedious; longevity, withdrawals, inflation, expenses, investment mixes and even stupidity. The danger for many people is that they use the models as a way to justify higher withdrawal rates on the basis of a simulation using historical data. They might set a failure probability at 5% and say that they are good at an inflation linked 4% initial withdrawal. That misses the potential that future markets won't be like historical ones, that inflation is higher than expected, that they do something silly ie panic and sell at a loss or that they just run into that 5% sequence of returns scenario. Probability distributions are all well and good until you collapse the statistics into the particular scenario of your own retirement and you hope that you aren't in the failure tail.

One major factor in whether a plan is sustainable is the withdrawal rate and a long time ago I decided that I would be frugal and save so much that I could live entirely on dividends and interest...a very old fashioned approach. I then decided to go one step farther and to use rental and DB/annuity type income thus largely decoupling from the stock and bond markets.

I know its frowned upon but, if you only spend the dividends, how is it possible to run down the pot? Surely it also avoids sequence of returns issues? Your approach makes perfect sense to me. Selling assets and rebalancing is also reduced some what, i think. 

I agree it should avoid the Sequence of Returns risk in most cases. I'm more relaxed about the volatility of established equity income ITs that have been paying increasing dividends every year for many decades like for example City of London IT and Merchants IT. They may not have the best long term total returns, but if they are paying 5% in dividends, increasing every year, that seems preferable to deciding when to sell capital from growth funds.