Investing for Decumulation - Recommended Reading

BritishInvestor

Thrugelmir said:

Historical data shows that when real interest rates are low. Returns on both equities and bonds tend to be lower rather than higher. 

But don't forget we aren't necessarily interested in whether returns are lower rather than higher, just not that they are outside the boundaries we have made our assumptions on, the lower boundary which has contained some pretty challenging times.  

kangoora

The method I used was to create a year by year financial model with parameterised expenditure, % return and %inflation.  One fiddles with the numbers, stress tests the model etc until one has the confidence to jump.  Of course reality has been very different to original plan, but that has not been a problem for 15 years thanks to very cautious assumptions.  The financial model approach makes it easy to adjust the plan in line with actuals, and so one moves forward.

Linton, in hindsight, could you have retired earlier?

Audaxer

BritishInvestor said:

Audaxer said:

TBC15 said:

I’ve been retired for just over a year and didn’t change my investment strategy ( 100% market no bonds) coming up to retirement apart from building up a cash buffer of about 4 yrs of cash.

The remarks of BritishInvestor gave me cause to revisit the 4yr cash buffer comfort blanket part of my plan. I’m now reasonably convinced my supper king size is a bit over the top and a double (1-2ys expenditure) would be more appropriate.

TBC15, are you currently in receipt of any other retirement income like DB pensions, or State Pension yet?  If not, 100% equities and only 1-2 years cash does sound like high risk to me, but I think it maybe depends on what percentage you need to drawdown each year. If for example you need to drawdown over 3.5% of your pot each year, do you not think a bad sequence of returns over the next decade could be particularly risky?

I think it's important to clarify what "high risk" is. For some, it's not taking enough equity exposure and running out of money. For others, they see high risk as their portfolio falling too far during periods of turbulence.
Regarding a poor series of returns, we already have historical returns and can perform this analysis, unless you are suggesting modelling something worse than we have seen previously/have data for?

Yes, you could run out of money if you don't have enough equity exposure to fund your retirement, but I think you could also run out of money with too much equity exposure and a bad sequence of returns when you are having to sell capital for income. Historically I'm not sure if 100% equities has done any better throughout a long retirement than having a more balanced portfolio including a percentage of bonds and other defensive assets.

If someone has enough other forms of secure income and/or built up a larger portfolio than they will need to fund their retirement, they might decide to go 100% equities, which is fine and not necessarily "high risk" in that they will still not run out of money. However in these circumstances some might take the view that going 100% equities is just not necessary for them, and they would rather have more available cash for spending when young enough to enjoy it.   

Thrugelmir

BritishInvestor said:

Thrugelmir said:

Historical data shows that when real interest rates are low. Returns on both equities and bonds tend to be lower rather than higher. 

But don't forget we aren't necessarily interested in whether returns are lower rather than higher, just not that they are outside the boundaries we have made our assumptions on, the lower boundary which has contained some pretty challenging times.  

There's no historical precedent for the unchartered waters we find ourselves in today. 

westv

Thrugelmir said:

BritishInvestor said:

Thrugelmir said:

Historical data shows that when real interest rates are low. Returns on both equities and bonds tend to be lower rather than higher. 

But don't forget we aren't necessarily interested in whether returns are lower rather than higher, just not that they are outside the boundaries we have made our assumptions on, the lower boundary which has contained some pretty challenging times.  

There's no historical precedent for the unchartered waters we find ourselves in today. 

You could say that about any period of uncertainly as no two are the same.

[Deleted User]

BritishInvestor said:

Linton said:

Deleted_User said:

TBC15 said:

Audaxer said:

TBC15 said:

I’ve been retired for just over a year and didn’t change my investment strategy ( 100% market no bonds) coming up to retirement apart from building up a cash buffer of about 4 yrs of cash.

The remarks of BritishInvestor gave me cause to revisit the 4yr cash buffer comfort blanket part of my plan. I’m now reasonably convinced my supper king size is a bit over the top and a double (1-2ys expenditure) would be more appropriate.

TBC15, are you currently in receipt of any other retirement income like DB pensions, or State Pension yet?  If not, 100% equities and only 1-2 years cash does sound like high risk to me, but I think it maybe depends on what percentage you need to drawdown each year. If for example you need to drawdown over 3.5% of your pot each year, do you not think a bad sequence of returns over the next decade could be particularly risky?

The wife’s SP starts next year and mine in 2025. I converted a DB pension to a SIPP 3 years ago. Based on the various simulations available on line I think I’m happy with the risk. Still thinking about the size of my comfort blanket/ cash bucket though.

Keep in mind that simulations are based on a ~100 year dataset (usually), that the monetary system is very different now from what it has been for most of the dataset and that you might need your pot for the next 40. That is a long time and you can easily see events happening that are different from the preceding 100 years. 

I am not saying that simulations are completely useless but they give people false confidence. Perhaps the future will be exactly like the past but the small chance that it won’t is exactly what a retiree needs to plan for. 

I agree with this, and would go further.  Data on the past 100+ years is the best we have but to treat it as any more than a useful random set of potential test problems of the type investors may have to face is assigning it more relevence than is justified.  The chance that the future is not like the past isnt small, it's a virtual certainty. We hopefully are not going to have 2 world wars again and if we are there may be little point in planning to survive them with our wealth intact.  Does the rise of automation and AI really correspond in some useful way to the rise of oil? To use historic data to derive meaningful values of % chance of future success is just playing with numbers. 

We have to accept we dont know what the future will bring.  The best we can do is to mitigate against some mainly shorter term eventualities and trust in the long term rise of the global markets.  If we cant do that one can ask whether it is sensible to invest in the first place.

I'm not sure anyone is suggesting the future is like the past. What is suggested as a reasonable starting point is that someone planning for retirement now doesn't have a worse outcome than the worst period over the last 100+ years. Do you think that is unreasonable?

In economic terms  past 100 years since ww1 have been dominated by the US and driven by expansion in trade and free enterprise.   In economic terms, mankind has had the best time, the fastest expansion in human history. 

I can easily see things being different over the next 40.  Not saying it’s unreasonable to assume that expansion will continue, major banks and insurance companies will stay solvent or be saved. Probably. But it’s plausible that the era of US domination, and the nature of their government, the rule of law will change. 

kinger101

BritishInvestor said:

gm0 said:

Not sure this link is "sufficient" or should be relied upon to support 100% equities in deaccumulation. (I agree 100% is fine as a *choice* of place to stand on the risk/return expected volatility curve if that's what you want and accept the volatility and time horizons AND constraints on withdrawls that may imply.  And it's generally totally fine in accumulation

I have done 100% equity for many years (all the way in accumulation) and only recently moved to 75% to support flexibility in what happens next - short term cashflow, wrapper and reinvestment changes.  But then my time horizon for those draws is "this tax year".  As I now enter deaccumulation.   I don't want to muck around with these mechanics without a "buffer" or be constrained on choice of timing by a correction of several years.  So for me, now, 1/4 portfolio "out of the market" and loss of that return for 6 months is worth it for ease of implementation.  But where will I go back to ?

It is all about what question you choose to ask.  Which gets complex fast in SWR debate if you start weaving in buffers, variable income etc. etc.  So the writer may not be "wrong" on their own specific assumption set and test design but that doesn't make it "sufficient".  I would draw attention to another pov.  I followed the same "appropriate" deaccumulation asset allocation argument through the lengthy McClung book (living off your money) and he draws a different conclusion again from backtesting WR scenarios for different retirement start dates, methods and portfolios and market data series.

He seems to find that above 70-80% the "worst of the worst" starting conditions and SORR start throwing in retirement failures.  For sure (as he notes) many or even most people in most cohorts would still died richer with 100% than 70% but *some* of them for some start dates start to get depleted along the way in retirement or have a dodgy 2nd half of a 40 year early retirement run.  Essentially he is saying that volatility driven bad equity sales from "(nearly) all equity" portfolios starts to hurt those portfolios too much on a rare set of occasions - even for sensible extraction approach and choices of WR.    It's a good if challenging read.

Now a buffer can help with that not selling the equities problem or a "fuse portfolio" not correlated and lower risk for a one off rebalance in a correction to buy equities - but then you don't REALLY have 100% any more.   All these ideas get done to death - monevator, ern etc. And nothing is *magic* but several things are helpful - some of the time. 

Backtesting on a single market series beyond a certain point you ARE looking at an over tuned data mining bias.
Clearly if you can push down the WR (large pot vs income need) all these problems largely disappear and the asset allocation decision is how much risk beyond what I need to take do I choose to.  It's only the combination of pushing the retirement length, the WR (up) to the maximum and the desired inheritance and risk asset % that this issue becomes more pressing and the failure levels creep from one offs to 1% or 5% or 10%.     For people getting into this territory I would say it is very helpful to do back testing (or read several trusted sources) about it. And to play with MonteCarlo Sim (using Flexible Retirement Planner) or otherwise to do some simulations for statistical distributions of returns journeys, inflation etc. to see how the selected plan seems to cope - how sensitive or "close to the edge" it actually is.   You won't be Taleb proof (Fat tails) but you will understand where you are on Normal.   Your maths needs to be up to it. My statistics was and arguably is *very* rusty

Monte Carlo can overstate the tails.
https://www.kitces.com/blog/monte-carlo-analysis-risk-fat-tails-vs-safe-withdrawal-rates-rolling-historical-returns/

My interpretation of that article is "I don't like the results so I'll ignore the model".  The whole point of the MC is to base the standard deviation on an equally weighted dataset.  Hardly surprising it comes up with more failures given the SD is higher.

While the monthly returns for SP500 don't quite fit a normal distribution, it's pretty close.
https://towardsdatascience.com/are-stock-returns-normally-distributed-e0388d71267e

And the reference to reversion to the mean in that article is baloney.  It means if one draws an extreme value, it's likely the next value with be closer to the mean.  I doesn't mean the stock market will be better the next day.  Or you'll roll a 1 next because you just rolled a six.  Ask Japan.
https://tradingeconomics.com/japan/stock-market

[Deleted User]

And remember, S&P500 is 60 years old. Dow Jones is not representative as an index. US fixed its currency to an ounce of gold until 1971. Any data going back more than 50 years isn’t directly comparable

Thrugelmir

westv said:

Thrugelmir said:

BritishInvestor said:

Thrugelmir said:

Historical data shows that when real interest rates are low. Returns on both equities and bonds tend to be lower rather than higher. 

But don't forget we aren't necessarily interested in whether returns are lower rather than higher, just not that they are outside the boundaries we have made our assumptions on, the lower boundary which has contained some pretty challenging times.  

There's no historical precedent for the unchartered waters we find ourselves in today. 

You could say that about any period of uncertainly as no two are the same.

I wouldn't regard now as a period of uncertainty. As they are certainies. Given the high levels of indebtedness that exist now not pleasant ones. 

BritishInvestor

Linton said:

BritishInvestor said:

Linton said:

Deleted_User said:

TBC15 said:

Audaxer said:

TBC15 said:

I’ve been retired for just over a year and didn’t change my investment strategy ( 100% market no bonds) coming up to retirement apart from building up a cash buffer of about 4 yrs of cash.

The remarks of BritishInvestor gave me cause to revisit the 4yr cash buffer comfort blanket part of my plan. I’m now reasonably convinced my supper king size is a bit over the top and a double (1-2ys expenditure) would be more appropriate.

TBC15, are you currently in receipt of any other retirement income like DB pensions, or State Pension yet?  If not, 100% equities and only 1-2 years cash does sound like high risk to me, but I think it maybe depends on what percentage you need to drawdown each year. If for example you need to drawdown over 3.5% of your pot each year, do you not think a bad sequence of returns over the next decade could be particularly risky?

The wife’s SP starts next year and mine in 2025. I converted a DB pension to a SIPP 3 years ago. Based on the various simulations available on line I think I’m happy with the risk. Still thinking about the size of my comfort blanket/ cash bucket though.

Keep in mind that simulations are based on a ~100 year dataset (usually), that the monetary system is very different now from what it has been for most of the dataset and that you might need your pot for the next 40. That is a long time and you can easily see events happening that are different from the preceding 100 years. 

I am not saying that simulations are completely useless but they give people false confidence. Perhaps the future will be exactly like the past but the small chance that it won’t is exactly what a retiree needs to plan for. 

I agree with this, and would go further.  Data on the past 100+ years is the best we have but to treat it as any more than a useful random set of potential test problems of the type investors may have to face is assigning it more relevence than is justified.  The chance that the future is not like the past isnt small, it's a virtual certainty. We hopefully are not going to have 2 world wars again and if we are there may be little point in planning to survive them with our wealth intact.  Does the rise of automation and AI really correspond in some useful way to the rise of oil? To use historic data to derive meaningful values of % chance of future success is just playing with numbers. 

We have to accept we dont know what the future will bring.  The best we can do is to mitigate against some mainly shorter term eventualities and trust in the long term rise of the global markets.  If we cant do that one can ask whether it is sensible to invest in the first place.

I'm not sure anyone is suggesting the future is like the past. What is suggested as a reasonable starting point is that someone planning for retirement now doesn't have a worse outcome than the worst period over the last 100+ years. Do you think that is unreasonable?

There are many criteria one could use for assessing whether a retirement strategy is sensible, none of which are definitive.  Passing the past history test is just one of these.  If it gives you the confidence to retire then fine, that is the only real purpose of checking. Perhaps, only when it is too late to do much about it will you discover whether that confidence is justified.  To take the results of historical simulations any more seriously than this does not seem justified to me.  In particular worrying about precise failure rates is pointless especially when the failure appears very unlikely.   People are too impressed with numbers and should really think through where they come from before over-relying on them.

The method I used was to create a year by year financial model with parameterised expenditure, % return and %inflation.  One fiddles with the numbers, stress tests the model etc until one has the confidence to jump.  Of course reality has been very different to original plan, but that has not been a problem for 15 years thanks to very cautious assumptions.  The financial model approach makes it easy to adjust the plan in line with actuals, and so one moves forward.

I think we are 90% in agreement. Future market outcomes are by no means certain and that should be emphasised. That said, I think we must retain a sense of optimism/balance and there is always a chance that someone retiring today may have a great outcome