Growth assumptions in your models/spreadsheets

Bostonerimus1

I started using average US stock market returns (~7%) and then did projections with lower returns to see the effects.
But this did not account for annual variations and so I took the average and StDev of the returns and generated a Gaussian distribution and randomly chose from the weighted probability of returns and did this many times. I used data from Wade Pfau. I also used the various "Monte Carlo" based online retirement estimators.

michaels

Bostonerimus1 said:

I started using average US stock market returns (~7%) and then did projections with lower returns to see the effects.
But this did not account for annual variations and so I took the average and StDev of the returns and generated a Gaussian distribution and randomly chose from the weighted probability of returns and did this many times. I used data from Wade Pfau. I also used the various "Monte Carlo" based online retirement estimators.

How do your lower bounds compare to an SWR (historic data) approach?

OldScientist

michaels said:

I think the OP is asking entirely the wrong question. 

I use the SWR tools and a zero historic failure rate over a 40 year time horizon. 

Basically we want to avoid 'failed' outcomes so there is no point modelling any sort of guess on what the central projection might be, only the lower bound.

The only other approach worth taking would be monte carlo but that takes a lot more data on distribution of likely returns.

I agree that taking the lower bound ('plan for the worst, hope for the best') is a good starting point.

There are currently very few free SWR tools for UK retirees (using historical data, https://www.2020financial.co.uk/pension-drawdown-calculator/ is one example), but the historical range of SWR appears to lie somewhere between 3.0% and 3.5% for a 30 year retirement (I note that the one published study of the effect of international diversification on SWR, https://www.advisorperspectives.com/articles/2014/03/04/does-international-diversification-improve-safe-withdrawal-rates , suggests that the UK SWR goes from 3.05% to 3.26% when holding global equities and bonds instead of domestic).

Of course, SWR includes the effect of sequence of returns, but the values can be approximated as an implied constant real growth rate using the excel rate function

For example (for a 30 year period),

SWR(%)  Rate(%)
2.5           -1.87
3.0           -0.71
3.333        0.0
3.5            0.34
4.0           1.31

In other words, assuming a constant 0% real growth rate would imply a not unreasonable SWR of 3.333%.

The range of historical values for growth rates is critically dependent on the period and asset considered. For example, for UK equities real annualised returns over a period of a decade have ranged from about -7% to 16%, over 20 year periods from -1% to 14%, and over 30 year periods from 1% to 11%.

michaels

Really useful, I think from that it suggests that taking modelling a 0% 'lower bound' real growth rate of assets is equivalent to addressing SORR risk using a SWR of 3.3% (UK, 30 year horizon)

Do the numbers change much in relation to 'longevity risk' - ie that many on here are looking to retire in their 50s and probably therefore have a 10% plus chance of a 40 year retirement?

Finally there are of course the unknown unknowns, anything that relies on historic data ranges rather than volatility measures is based on only a few hundred data points compared to an unlimited possible set of future scenarios.  Personally I suspect the 'average growth' of the last 150 years is partially supported via demographics were almost all countries are at or past the positive impact of demographics phase and entering or in the demographics as a drag period (see long term interest rates).

LeafGreen

I thought I was quite unusual because I just do my calcs in today's money, but seems I am not alone.  I am naturally quite risk averse though, and have more in cash than I probably should. 

Cus

Linton said:

The purpose of assumptions in retirement planning is not to accurately predict the future which is impossible, but rather to give you the confidence to jump.  So there is no right answer  as to what those should be, it's very much up to you as to what rassurance you need.. 

You have to accept that there will always be global circumstances when your retirement plan will fail. The best you can hope for is that you will be in a no worse position than everybody else.  If there is a global crash that lasts 20-30 years the absence of real profits in global industry would imply that vast numbers of people are unemployed across the world. 

20-30 years of failure to beat inflation could well arise in individual countries, but to assume that it could occur globally is in my view more a plot for a disaster movie than a realistic scenario that needs to be, or can be, planned for.  These considerations do illustrate that diversification in countries across the world is essential.

Unfortunately, diversifying across all countries using an all market index still leaves you exposed to one country -US, with 64% of your investment. And also exposed/more invested into a couple of single companies based there, more than any individual country investments.

Milltir

LeafGreen said:

I thought I was quite unusual because I just do my calcs in today's money, but seems I am not alone.  I am naturally quite risk averse though, and have more in cash than I probably should. 

You are certainly not alone. I also use today's money and tax rates.

OldScientist

michaels said:

Really useful, I think from that it suggests that taking modelling a 0% 'lower bound' real growth rate of assets is equivalent to addressing SORR risk using a SWR of 3.3% (UK, 30 year horizon)

Do the numbers change much in relation to 'longevity risk' - ie that many on here are looking to retire in their 50s and probably therefore have a 10% plus chance of a 40 year retirement?

Finally there are of course the unknown unknowns, anything that relies on historic data ranges rather than volatility measures is based on only a few hundred data points compared to an unlimited possible set of future scenarios.  Personally I suspect the 'average growth' of the last 150 years is partially supported via demographics were almost all countries are at or past the positive impact of demographics phase and entering or in the demographics as a drag period (see long term interest rates).

A very quick look at 40 years indicates the SWR is somewhere between 2.5% and 3.0% (it depends on asset allocation, https://www.2020financial.co.uk/pension-drawdown-calculator/ ) which then using the rate function infers a constant real return of somewhere between 0% and 1% (i.e., maybe slightly higher than the 30 year case).

I think your last paragraph applies - there are fewer 40 year periods in the historical data than there are 30 year periods and the potential worst cases (assuming that returns do follow some random distribution) just haven't happened yet.

Anyway, it seems to me that 0% real growth is a good starting point, but that investing in some guaranteed income (e.g., inflation linked ladder and/or RPI annuity) to cover core expenses removes some of the randomness/volatility from the retirement plan (but not all risks).

OldScientist

Cus said:

Linton said:

The purpose of assumptions in retirement planning is not to accurately predict the future which is impossible, but rather to give you the confidence to jump.  So there is no right answer  as to what those should be, it's very much up to you as to what rassurance you need.. 

You have to accept that there will always be global circumstances when your retirement plan will fail. The best you can hope for is that you will be in a no worse position than everybody else.  If there is a global crash that lasts 20-30 years the absence of real profits in global industry would imply that vast numbers of people are unemployed across the world. 

20-30 years of failure to beat inflation could well arise in individual countries, but to assume that it could occur globally is in my view more a plot for a disaster movie than a realistic scenario that needs to be, or can be, planned for.  These considerations do illustrate that diversification in countries across the world is essential.

Unfortunately, diversifying across all countries using an all market index still leaves you exposed to one country -US, with 64% of your investment. And also exposed/more invested into a couple of single companies based there, more than any individual country investments.

Research in the 60s suggested that  8-10 stocks was sufficient diversification although more recent research suggests 30 to 50. Concentration risk in FTSE world index (and others) is definitely heading towards the lower of those two ranges.

michaels

OldScientist said:

Cus said:

Linton said:

The purpose of assumptions in retirement planning is not to accurately predict the future which is impossible, but rather to give you the confidence to jump.  So there is no right answer  as to what those should be, it's very much up to you as to what rassurance you need.. 

You have to accept that there will always be global circumstances when your retirement plan will fail. The best you can hope for is that you will be in a no worse position than everybody else.  If there is a global crash that lasts 20-30 years the absence of real profits in global industry would imply that vast numbers of people are unemployed across the world. 

20-30 years of failure to beat inflation could well arise in individual countries, but to assume that it could occur globally is in my view more a plot for a disaster movie than a realistic scenario that needs to be, or can be, planned for.  These considerations do illustrate that diversification in countries across the world is essential.

Unfortunately, diversifying across all countries using an all market index still leaves you exposed to one country -US, with 64% of your investment. And also exposed/more invested into a couple of single companies based there, more than any individual country investments.

Research in the 60s suggested that  8-10 stocks was sufficient diversification although more recent research suggests 30 to 50. Concentration risk in FTSE world index (and others) is definitely heading towards the lower of those two ranges.

But aren't stocks that make up a large proportion of the global tracker likely to be highly correlated (because they are in the same sector) so may not really represent a diversification?