Evaluating Annuity vs Drawdown

DCF_Evaluator

These forums naturally contain threads which:

• discuss the pros and cons of annuities vs drawdowns
• the optimal timing for taking one of the above
• the annuity options (JL, RPI etc)
• the setup/advice cost re a drawdown.
  

Within an environment where the following will vary:

• the personal taxation of the individual
• the level and type(s) of advice/charges
• assumptions re market(s) growth
• individuals may 'overdraw' a drawdown so as to treat it like a 'quasi annuity'.

I am new to this. I thought that given the number of variables here the retirement implications could only be evaluated via a Discounted Cash Flow Model with input options for taxation, charges, market growth assumptions etc. So I came up with one.
The upshot so far – with assumption that the pension income is the only income and taxed at basic rate and no 'overdraw' as above takes place - using the model - is that:

• Low Cost Index Tracker Drawdown – the relevant market(s) has to grow by an annualised 4.25% to 4.5% in order for the drawdown income to equal a comparable annuity (50% JL, no guarantee, no RPI)
• Discretionary/Actively Managed Drawdown (charges 2%pa) – about 6% growth of the relevant market(s) is required.
• Having got to here the above required growth can be compared to market(s) growth this century on a Total Return Index Basis. Figures are hard to obtain and we are aware that past is no guide to the future and taking no account of an individual's attitude to risk but:
• last 5 years FTSE grew about 1.44% on an annualised TRI basis
• last 10 years, 8.30% same basis
• last 20 years, same (guesstimate) 7% - 9% same basis
• last 100 years (about 7%?) same basis

So, the DCF model gives us a 'sighting shot' which indicates:

• the low cost Drawdown would have been significantly better that an annuity long term over the past 10 to 100 years
• short term comparisons would be less favourable
• the Discretionary approach would have to be justified on the extent, annualised, that the manager was able to exceed the market(s)
• Even if the Discretionary approach does no better than a tracker it is still better than the annuity
• etc

Has anyone else tried producing a similar model?
Incidentally, the model also has some spin off benefits eg RPI annuities – these appear to be assuming currently an RPI of about 4.7% - currently RPI is only about 3.1%. It also shows whether to defer the state pension or not.

atush

First of all, you are assuming you have no running down of the fund over the decades,

and 2 your FTSE figures don't take into acct dividend income as figures i have seen that included this are no where near the figures you mention.

Linton

You may find Firecalc of interest. It is based on historic US data but does give a good indication of what drawdown % is reasonable to ensure you dont run out of money before you die.

I (and others including Firecalc) believe that a 4.5 % drawdown is reasonable to provide a broadly inflation matching income over the long term. This makes drawdown a better option than an annuity particularly if you have some flexibility in the amount drawn down.

DCF_Evaluator

Atush

Thanks, it does assume no run down. There is an option to input an 'extra drawdown' amount which would simulate a run down, For example, to run the fund down to about 10% over 30 years would change the required market yield on the low cost basis upwards to about 5.25%.

I did check that the market growth included income. Note that the growth shown is annualised not cumulative. Does this help?

jamesd

How did you get comparability when drawdown provides a 100% spousal pension? To do that you'd need a mixture of drawdown and single life annuity. Or to accept that drawdown provides a better spousal benefit.

You should model:
1. With deferring the state pensions for several years, because this decreases your failure risk, particularly in the long life variations.
2. With at least a year or two of spending in cash savings accounts that you use to smooth income levels and avoid having to do any selling of investments at bad times. Again, risk reduction.

DCF_Evaluator

jamesd

Thanks for the input. I was not aware of the 100% spousal pension with drawdown but thinking about it its obvious. However, the annuity rate is an input parameter so I changed it and (eg) the 6% required growth figure above is, as expected, lower - now only 5.6%.

1. Good point re deferral - will have a look
2. Also good point, will have a look using drawdown of zero for years 1 and 2. Will let you know!

gadgetmind

jamesd wrote: »

avoid having to do any selling of investments at bad times. Again, risk reduction.

Such a cash/bond buffer is in my plans, but I read a fairly convincing argument the other day that you should reduce all asset classes in proportion as you otherwise risk having to sell equities at the *very* bottom for a long dip.

I guess it depends on length and severity, which is always impossible to predict.

DCF_Evaluator

jamesd

I had a look.

1. State Pension Deferral ... yes, the model estimates the benefit or otherwise of deferral on basis that state pension is 'top slice' income so all is taxed and at 20%, basic single person pens, from age 65, £107.45 + a SERPS value £60 and 2% annual increases going forward in basic pension. Given this the model reckons no deferral scenario is worthwhile unless a recipient longevity at least 88 ... the more longevity the more the gain from deferral (obviously) with a the 3-5 year deferral showing the best result. The drawdown model has no cash flows beyond age 90 so, using this for the state pension also the model indicates a £20k deferral benefit. However, this deferral is valid with both Annuities and Drawdown so don't think it is relevant for comparison here.

2. I discovered that the settings of the model I used had been left to assume zero drawdown in the first two years. If there is a full drawdown - to the extent of the fund growth - in those two years the market growth requirement increases by about 0.25%.

jamesd

gadgetmind wrote: »

Such a cash/bond buffer is in my plans, but I read a fairly convincing argument the other day that you should reduce all asset classes in proportion as you otherwise risk having to sell equities at the *very* bottom for a long dip.

I'd be interested in reading that if you have a reference.

Since it's to buffer income changes more than capital changes the drawing rate shouldn't be very high, assuming that income decreases are less likely and less severe than capital value changes and that in the early years the GAD limit causes much of the income to come from dividends and corporate bond interest, more than capital

gadgetmind wrote: »

I guess it depends on length and severity, which is always impossible to predict.

True, though I'd hope that a couple of years of spending to cover ten years of a twenty percent drop would do a pretty good job.

DCF_Evaluator wrote: »

1. State Pension Deferral ... yes, the model estimates the benefit or otherwise of deferral on basis that state pension is 'top slice' income so all is taxed and at 20%, basic single person pens, from age 65, £107.45 + a SERPS value £60 and 2% annual increases going forward in basic pension. Given this the model reckons no deferral scenario is worthwhile unless a recipient longevity at least 88

88 is around the age that half of men aged 65 and in normal good health are expected to reach. So even if it took that long, it'd be good for long term risk reduction to do it for a year.

That seems too high as a threshold age for deferral. Ignoring both inflation increases and the time value of money the payback is 9.6 years, taking just until 76. That's 52 weeks of after tax £133.96 = 6965.92 to be recovered by the £13.93 increase after a year's wait. 6965.92 / 13.93 = 500 months = 9.6 years. So that makes me wonder how the model is causing the payback time to increase from 10 to 22 years.

The cash fund isn't to avoid taking any income, it's just to smooth the income level, so a 10% drop in income can be covered for ten years by one year of anticipated income kept in savings. The income is likely to be partly interest/dividends and partly capital but in a pension situation the GAD limit will tend to bias is strongly towards interest and dividends and away from capital growth early on, assuming say FTSE tracker dividends are taken.

The early years are also where the long term consequences of a big capital value drop while drawing on some capital is most severe, so the cash pot can be gradually reduced over time.

jaybeetoo

What age difference have you assumed between the first life and second life?

DCF_Evaluator

jaybeetoo

Annuity taken when first life 60, second life 56.