Locking In Annuity Rates?

Cus

masonic said:

Cus said:

OldScientist said:

SVaz said:

Is it Always the case that when bond prices fall, yields rise and Annuity rates are higher?

So if you had £100k in a 100% bonds fund that dropped to £75k,  it doesn’t matter because the annuity income is the same / similar either way? 

Are there any circumstances where that wouldn’t happen?  

I’m sorted for income until I’m 74, in 14 years time,  after that then using half my pot for a joint life annuity might be a good idea.
I remember Dunstonh saying that you can have it paid into a Sipp,  which would be handy if you don’t need the income - if, however one of us died then it would be needed.  

Sorry for the rather extensive answer...

 The pricing of annuities depends on projected life expectancy, expected asset returns, and fees (e.g., see https://retirementresearcher.com/income-annuity-101/ for the basic method). Values for each of these may very well vary between insurance companies which leads to the difference in available payout rates (e.g., on 1/10/2025 a single life RPI annuity at 65yo had payout rates between 4.91% and 5.28%). For example,
1) Life expectancies are provided by the Institute and Faculty of Actuaries (see https://www.actuaries.org.uk/learn-and-develop/continuous-mortality-investigation/other-cmi-outputs/unisex-rates-0), although health and other factors (e.g., postcode) are considered.
2) Expected asset returns will depend on what the insurance company is invested in (gilts, corporate bonds, infrastructure, etc.). An approximation is to use gilt yields (e.g., see the calculator at https://lategenxer.streamlit.app/Annuity_Valuator ).
3) Fees will also vary between insurance companies.

Assuming 1) and 3) are constant, then the annuity payout rate will depend on yields - so as yields go up, the payout rate will increase and vice versa. The magnitude of change in payout rate will depend on the duration of underlying assets of the annuity - this can be approximated by assuming gilts are held - increasing yields will decrease the duration of the annuity while increasing life expectancy will increase the duration (this means that the duration gets shorter at later ages). I've mentioned on these boards before that the duration is approximately equal to half the life expectancy but the duration also depends on yield (and given that we don't actually know the underlying investments or fees, this is a very rough approximation).

Interestingly, research (US based, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2021579 ) indicated that "Annuity prices react more rapidly and with greater sensitivity to an increase in the relevant interest rate compared to a decrease" (in other words in favour to customers!).

Finally, the change in NAV of a bond fund is dependent on the weighted modified duration which will in turn depend on the bonds held and their yields.

To take an example, assume the bond fund and the annuity both have a duration of 10 and yields increase by 1 percentage point. The payout rate of the annuity will increase by roughly 10% while the NAV of the bond fund will decrease by 10%, so the overall income from an annuity purchase will stay the same as before the change in yields. Note that the duration of both annuity and fund would change after the change in yields, but not necessarily by the same amount.

However, if the duration of the bond fund was 15, then a 1pp increase in yields would mean that the NAV would fall by 15% and the amount of income after annuity purchase would be reduced (roughly to 1.10/1.15=0.96).

A step change in life expectancies (e.g., a cheap cure for cancer) would temporarily disrupt the link between payout rates and yields (and would have the opposite effect on life insurance premiums).

Do you have any research related to the following?

If one bought an annuity in 2018 when interest rates were low, then assuming it would then cost £100k to obtain the annuity with a rate of x%, which would increase with RPI, and assume today that a similar annuity purchased today for £100k obtains a rate of say 2x%, would the original annuity be paying the same rate now, and if not purchased then, would the £100k be valued today at £50k but would still obtain a x% rate now?

Basically the question is that if one had only had a bond pension investment with the intention of buying an annuity, then the drop in bond prices made no difference, so one could say that there is no such thing as 'good current annuity rates' ?  Assuming all other factors the same.

If you take HL's best buy annuity rates, say for single life, age 60, RPI, then you can do the then and now comparison:

Now: https://www.hl.co.uk/retirement/annuities/best-buy-rates vs Sep 2018: https://web.archive.org/web/20180925152122/https://www.hl.co.uk/retirement/annuities/best-buy-rates

£4,515 vs £2,445 per £100,000

You can also look at an index linked gilt index fund with maturity around 15 years and compare prices, e.g. 

https://www.vanguard.co.uk/professional/product/fund/bond/9226/uk-inflation-linked-gilt-index-fund-gbp-acc

Now: £13,776 vs Sep 2018: £18,090 (based on the chart of growth of £10,000 since inception)

So £100,000 in Sep 2018 used to purchase an index linked annuity would give you £2,445 in 2018 at RPI = 284.1, which would now be £3,495 at RPI = 406.1

If you instead waited, your £100,000 would now be £76,153, which could be used to purchase £3,438, which is pretty close.

So the hedge protected your rubbish annuity rate. Obviously the smart move was not to take the hedge when annuity rates were at historic lows.

Edit: I realise I've glossed over the lack of income in the bond investment scenario, but I'm not sure how to weigh that against locking in a negative real YTM at the point of purchase!

Thanks, helpful as always.  What specifically are you referring to with 'the hedge'?
Can the above indicate that the annuity market predicted a longer period of lower interest rates than what has actually happened?

masonic

Cus said:

masonic said:

Cus said:

OldScientist said:

SVaz said:

Is it Always the case that when bond prices fall, yields rise and Annuity rates are higher?

So if you had £100k in a 100% bonds fund that dropped to £75k,  it doesn’t matter because the annuity income is the same / similar either way? 

Are there any circumstances where that wouldn’t happen?  

I’m sorted for income until I’m 74, in 14 years time,  after that then using half my pot for a joint life annuity might be a good idea.
I remember Dunstonh saying that you can have it paid into a Sipp,  which would be handy if you don’t need the income - if, however one of us died then it would be needed.  

Sorry for the rather extensive answer...

 The pricing of annuities depends on projected life expectancy, expected asset returns, and fees (e.g., see https://retirementresearcher.com/income-annuity-101/ for the basic method). Values for each of these may very well vary between insurance companies which leads to the difference in available payout rates (e.g., on 1/10/2025 a single life RPI annuity at 65yo had payout rates between 4.91% and 5.28%). For example,
1) Life expectancies are provided by the Institute and Faculty of Actuaries (see https://www.actuaries.org.uk/learn-and-develop/continuous-mortality-investigation/other-cmi-outputs/unisex-rates-0), although health and other factors (e.g., postcode) are considered.
2) Expected asset returns will depend on what the insurance company is invested in (gilts, corporate bonds, infrastructure, etc.). An approximation is to use gilt yields (e.g., see the calculator at https://lategenxer.streamlit.app/Annuity_Valuator ).
3) Fees will also vary between insurance companies.

Assuming 1) and 3) are constant, then the annuity payout rate will depend on yields - so as yields go up, the payout rate will increase and vice versa. The magnitude of change in payout rate will depend on the duration of underlying assets of the annuity - this can be approximated by assuming gilts are held - increasing yields will decrease the duration of the annuity while increasing life expectancy will increase the duration (this means that the duration gets shorter at later ages). I've mentioned on these boards before that the duration is approximately equal to half the life expectancy but the duration also depends on yield (and given that we don't actually know the underlying investments or fees, this is a very rough approximation).

Interestingly, research (US based, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2021579 ) indicated that "Annuity prices react more rapidly and with greater sensitivity to an increase in the relevant interest rate compared to a decrease" (in other words in favour to customers!).

Finally, the change in NAV of a bond fund is dependent on the weighted modified duration which will in turn depend on the bonds held and their yields.

To take an example, assume the bond fund and the annuity both have a duration of 10 and yields increase by 1 percentage point. The payout rate of the annuity will increase by roughly 10% while the NAV of the bond fund will decrease by 10%, so the overall income from an annuity purchase will stay the same as before the change in yields. Note that the duration of both annuity and fund would change after the change in yields, but not necessarily by the same amount.

However, if the duration of the bond fund was 15, then a 1pp increase in yields would mean that the NAV would fall by 15% and the amount of income after annuity purchase would be reduced (roughly to 1.10/1.15=0.96).

A step change in life expectancies (e.g., a cheap cure for cancer) would temporarily disrupt the link between payout rates and yields (and would have the opposite effect on life insurance premiums).

Do you have any research related to the following?

If one bought an annuity in 2018 when interest rates were low, then assuming it would then cost £100k to obtain the annuity with a rate of x%, which would increase with RPI, and assume today that a similar annuity purchased today for £100k obtains a rate of say 2x%, would the original annuity be paying the same rate now, and if not purchased then, would the £100k be valued today at £50k but would still obtain a x% rate now?

Basically the question is that if one had only had a bond pension investment with the intention of buying an annuity, then the drop in bond prices made no difference, so one could say that there is no such thing as 'good current annuity rates' ?  Assuming all other factors the same.

If you take HL's best buy annuity rates, say for single life, age 60, RPI, then you can do the then and now comparison:

Now: https://www.hl.co.uk/retirement/annuities/best-buy-rates vs Sep 2018: https://web.archive.org/web/20180925152122/https://www.hl.co.uk/retirement/annuities/best-buy-rates

£4,515 vs £2,445 per £100,000

You can also look at an index linked gilt index fund with maturity around 15 years and compare prices, e.g. 

https://www.vanguard.co.uk/professional/product/fund/bond/9226/uk-inflation-linked-gilt-index-fund-gbp-acc

Now: £13,776 vs Sep 2018: £18,090 (based on the chart of growth of £10,000 since inception)

So £100,000 in Sep 2018 used to purchase an index linked annuity would give you £2,445 in 2018 at RPI = 284.1, which would now be £3,495 at RPI = 406.1

If you instead waited, your £100,000 would now be £76,153, which could be used to purchase £3,438, which is pretty close.

So the hedge protected your rubbish annuity rate. Obviously the smart move was not to take the hedge when annuity rates were at historic lows.

Edit: I realise I've glossed over the lack of income in the bond investment scenario, but I'm not sure how to weigh that against locking in a negative real YTM at the point of purchase!

Thanks, helpful as always.  What specifically are you referring to with 'the hedge'?
Can the above indicate that the annuity market predicted a longer period of lower interest rates than what has actually happened?

The hedge is trying to liability match the annuity purchase using index linked gilts, rather than selecting a portfolio aimed at delivering a good outcome over the period (which at that point would have been some mix of cash, short dated bonds, and perhaps a small helping of equities). Someone investing for an optimal outcome could have achieved an annuity now that is more than £4,515, because their original £100,000 could have been grown over the 7 years. But unfortunately when hedging, you miss out on upside as well as further downside.

The annuity market just prices its products based on the fixed interest options available to work with at the time the annuity is sold. It was a decade of Quantitative Easing that suppressed long-duration gilt yields and led to annuity rates reaching historic lows. Then there was a rather abrupt switch to QT which removed the BoE as reliable buyer of last resort and increased supply at the same time Russia's invasion of Ukraine was causing supply chain shocks and a surge in inflation that had to be matched with rapid interest rate hikes.

None of that was predicted by the bond markets. But to discerning investors who did not need to buy government bonds, they were looking pretty unattractive from a risk-reward profile in the years leading up to that. Someone hoping to buy an annuity some years in the future would hopefully have elected to take their chances that rates would improve by the time they got there, which indeed they did once this all kicked off. The icing on the cake seems to be a risk premium added due to more recent fiscal concerns, which has sent yields to levels that look attractive by historical standards - and worth locking in, lest they fall back in the coming years.

LHW99

DRS1 said:

OldScientist said:

Secret2ndAccount said:

You can't lock in a price for anything to purchase in 10 yers' time. Not a car; not a house; not a chicken. An annuity is one product where you could realistically price it for provision in 10 years, since they are already commiting to continue paying you for potentially decades. It appears there is not sufficient demand. For a decade, nobody has wanted to buy an annuity, now everybody wants one. If it stays like this for a number of years perhaps a disruptor will come along and start offering all sorts of useful products (a single annuity that pays out 12k extra until SPA then drops?  A product that matches wage inflation instead of CPI?). Until then, all you can do is build a nearest equivalent solution for yourself.
A single bond, TR35 would pay inflation plus 1.6% over 10 years. Or buy TR40 which pays CPI+2%. After 10 years (or at any other time) it is likely you could sell it for an amount roughly commensurate with the expected gain. If the price wasn't what you expected, there's a fair chance that annuity prices would have canceled out the error. The bad news is that you might not get exactly the annuity you hoped for. The good news is that you can get your money back at any time if live intervenes. Or you might get more...

In the US, deferred annuities (called deferred income annuities DIA or QLAC) are available - deferred annuities are even mentioned on some UK insurance company sites, but do not (yet?) appear to actually be available. Unfortunately for the Americans, they no longer have access to CPI annuities only nominal.

An annuity that pays to 67 is already available (fixed term annuity).

I sometimes wonder if the UK references to deferred annuities are actually reference to section 32 buyout policies which were also called deferred annuity policies.  I think they were used for buying out deferred pensions when a final salary scheme wound up but were also available on an individual basis.

Not just for DB's winding up, in the days when DB transfers were not limited by regulation a salesman tried to get me to exchange a DB for an S32. This would have been probably early 1980's, and the DB I had was still running (and still is).

Triumph13

Thanks all.  Some really useful information and insights.  I foresee some interesting conversations with said friend as he gets himself up to speed and to a position where he can make an informed decision.

Triumph13

OldScientist said:

Just to add: Bear in mind that effectively matching to a deferred annuity is an approximate process. I've done some preliminary (i.e,., unfinished) historical modelling of the outcomes for nominal annuities (there's more data than for ILG) which suggests that while the expected income can be locked in to within about 5% most of the time (i.e., if you were expecting an income of £1000 you'd end up with somewhere between £950 and £1050), the worst cases had a tracking error of upwards of 15% (i.e., between £850 and £1150). I've not yet finished the analysis (I've got distracted by other things!), but my initial thoughts are that the worst cases occurred when there were very large changes in yields over the course of delay between gilt purchase and annuity purchase.

Two funds (e.g., the under 10 year ILG fund by ishares and 'all stocks' ILG funds offered by various fund houses) could be combined to provide an average duration around the right value (roughly half the unisex life expectancy at annuity purchase plus the delay) - this would be an ongoing process since the weighting towards the shorter fund would gradually increase with time. I've yet to model this, but would expect similar outcomes.

In reality this is a variation on the generalised 'lifestrategy for annuity purchase' offered by many pension platforms.

Thinking about the required income floor in retirement might help decide on precisely how much to commit to this strategy - the rest can be left invested normally.

Can I just check something with you @OldScientist ? Am I right in thinking that the modelling you are doing excludes coupons in the deferral period?  If my googling is correct, ILG 2049 trades at just under par and has a 1.85% coupon.  So, if my understanding is correct, you'd expect the bond value to track annuity rates - subject to the above tracking errors, but you'd also be getting ten years of coupon in the deferral period and could reasonably expect that to more than offset any adverse tracking error?  Assuming you reinvest it of course.

OldScientist

Triumph13 said:

OldScientist said:

Just to add: Bear in mind that effectively matching to a deferred annuity is an approximate process. I've done some preliminary (i.e,., unfinished) historical modelling of the outcomes for nominal annuities (there's more data than for ILG) which suggests that while the expected income can be locked in to within about 5% most of the time (i.e., if you were expecting an income of £1000 you'd end up with somewhere between £950 and £1050), the worst cases had a tracking error of upwards of 15% (i.e., between £850 and £1150). I've not yet finished the analysis (I've got distracted by other things!), but my initial thoughts are that the worst cases occurred when there were very large changes in yields over the course of delay between gilt purchase and annuity purchase.

Two funds (e.g., the under 10 year ILG fund by ishares and 'all stocks' ILG funds offered by various fund houses) could be combined to provide an average duration around the right value (roughly half the unisex life expectancy at annuity purchase plus the delay) - this would be an ongoing process since the weighting towards the shorter fund would gradually increase with time. I've yet to model this, but would expect similar outcomes.

In reality this is a variation on the generalised 'lifestrategy for annuity purchase' offered by many pension platforms.

Thinking about the required income floor in retirement might help decide on precisely how much to commit to this strategy - the rest can be left invested normally.

Can I just check something with you @OldScientist ? Am I right in thinking that the modelling you are doing excludes coupons in the deferral period?  If my googling is correct, ILG 2049 trades at just under par and has a 1.85% coupon.  So, if my understanding is correct, you'd expect the bond value to track annuity rates - subject to the above tracking errors, but you'd also be getting ten years of coupon in the deferral period and could reasonably expect that to more than offset any adverse tracking error?  Assuming you reinvest it of course.

The modelling I've done assumes zero coupon bonds since this makes the code easier (for other projects, I have modelled actual gilts going back to 1870, coupons and all, it is a right pain!). This method implicitly assumes coupons (where they exist) would be reinvested which would then introduce some further uncertainty in the matching since the price at which coupons will be reinvested is unknown in advance. For small coupons (1.85% is small in a historical context), the uncertainty is also small.

Triumph13

Thanks @OldScientist .  If I can just check I'm understanding that properly as trying to understand index linked gilt pricing is making my brain ache?  What I think you are saying is that rather than tracking actual bond prices against annuity rates, what you did, for very sensible reasons, is to impute what the price of a zero coupon bond would have been and do your analysis based on that?  So the conclusion is that, during deferment periods, appropriately dated zero coupon bonds track closely with annuity rates, and any actual coupon in the deferment period would have to be reinvested to keep that tracking performance?  With the added issue that the bigger the coupon the less accurate the tracking because of variability of price at reinvestment?

OldScientist

Triumph13 said:

Thanks @OldScientist .  If I can just check I'm understanding that properly as trying to understand index linked gilt pricing is making my brain ache?  What I think you are saying is that rather than tracking actual bond prices against annuity rates, what you did, for very sensible reasons, is to impute what the price of a zero coupon bond would have been and do your analysis based on that?  So the conclusion is that, during deferment periods, appropriately dated zero coupon bonds track closely with annuity rates, and any actual coupon in the deferment period would have to be reinvested to keep that tracking performance?  With the added issue that the bigger the coupon the less accurate the tracking because of variability of price at reinvestment?

Your last two sentences match my understanding. The question then becomes 'how much tracking error?', to which the answer is (as always with finance) 'it depends'.

To take an example, let's assume we have a gilt with a 2% coupon we are holding in the 'deferral period' which is priced at par and we want to calculate the total return over a year. Assuming the price remains at 100 for the whole year the total return for the first coupon period of 6 months is

(100+1)/100=1.01=1% (i.e., the final price plus coupon divided by the initial price)

and the return for the second coupon period is the same, for a total annual return of 1.01*1.01=1.0201=2.01% (i.e., almost the same as the yield of 2%). This would also be the return for an appropriately priced zero-coupon gilt.

Now, let's assume that the price after 6 months is 98 instead of 100, while the starting and ending prices remain at 100

The total return for the first coupon period is then

(98+1)/100=0.99=-1%

While the return for the second coupon period is

(100+1)/98=1.0306=3.06%

The total return is then 0.99*1.0306=1.0203=2.03%

i.e., the total return where the price has decreased at the time when the coupon is reinvested is a very small amount larger (2bp) than where the price has stayed the same. Larger coupons and larger changes in price will lead to larger changes in the return. My view, is that the tracking errors that result from difficulties in matching the duration of the annuity far exceed those resulting from coupon reinvestment, but the latter does mean that the process isn't entirely 'fire and forget'.

Triumph13

Just thought I'd report back on what my cautious friend ended up deciding to do. After lots of thought, research and analysis, he is going to use half of his pension fund to buy an appropriately-dated ILG, ready to buy an annuity when he reaches pension access age. He recognises that this is likely to underperform vs staying in the market for that period, but considers the price well worth paying for the relative peace of mind.

The kicker is that knowing he has this 'deferred annuity' in place, plus the SP later, has given him the confidence to go for a 90%+ equity allocation with the other half of his pension. Overall he will, therefore, end up pretty close to 50% equities, 50% bonds, but feeling a lot more relaxed about it all as the annuity plus SP will cover all his core spending, with the invested portion being jam.

Thanks again to everyone who responded and especially to OldScientist.

tigerspill

https://forums.moneysavingexpert.com/discussion/comment/81858766#Comment_81858766

Sounds like a decent plan.

I personally treat ILGs (held to maturity) as a totally different asset class to what is normally classed "bonds". The reason being is their real value is locked in regardless of what the equity or normal bond markets do.