DB Pension Revaluations reply from Mercer - can you help me interpret their secret code?

Pat38493

Tommyjw said:

Pat38493 said:

Tommyjw said:

It is always just the complete calender years between Leaving Date and Retirement Date (in this case it appears your Normal Retirement Age classes as your Retirement Date as on a Scheme which projects up to then first). Anything different (which would be odd) would still have to be underpinned (the higher of both calculations are then paid) by the statutory minimum which you cannot pay less than.

It is often quite hard to find places where it really gets down to the details on it but some view sites using search terms to find agreeing points:

• https://www.legislation.gov.uk/uksi/2022/1229/pdfs/uksiem_20221229_en.pdf  - (point 7.7 providing a short example)
• https://www.cpc-learning-materials.com/leavers-fact-finding-final-salary  - (weird one but this is a entry level exam in the DB admin world and see "...the whole of the pension in excess .. is increased .. for each complete year from date of leaving to date of retirement."
  
• https://cms-lawnow.com/en/ealerts/2015/12/pension-revaluation-and-increases-in-deflationary-times  (Revaluation in defined benefit (DB) schemes... for complete years between leaving .. and the scheme's normal pension age.)
  

If i assume your NRD is over 25 complete years away (so after May 2033?) Then as an administrator you calculate the complete years (25), you think 'what years inflation do we not know' = 2025 to 2033 inclusive (9 total), 25-9 = 16 years 'known' revaluation to use now.

From reading TVAS wording i think (emphasis) what they were trying to say is (for yourself for example) you can't change the May 2008 start date or your NRD (65?). So lets say we are now in 2033, your reaching your NRD. if your NRD due to your age is March 2033 you're only entitled to 24 years of revaluation at that point as from March 2008 you havent hit the last complete year... if your NRD happens to be after May (so after your DOL anniversary month) you have 25 years of revaluation, which just points to the luck/unfairness of the way they are calculated. 

Thanks Tommyjw - let me clarify again exactly just to make sure as we might be at cross purposes here.  My birthday is in January.

The normal retirement date under the scheme for me would be at age 65 in January 2034.

(there is a part of the pension which has retirement age 60 but I'm not sure whether that changes anything for this discussion).

I am getting a quote for retiring at June 2024.  There have been 16 full years in deferment as of June 2024.  

If I am understanding you correctly, you agree with me that they should use 1.571 as the (minimum) factor and not 1.496.  1.496 should be used if I was asking for payment before May 31st.

From previous threads like the one linked above, I had understood that it's the number of complete years between the deferment, and the in payment date that has to be used, and hence you should carefully select your retirement date within the target year.

I am just double checking this because a lot of the links talk about the year you reach "normal retirement age" (of which on my pension there is arguably no such thing due to the different tranches).

On the other hand if only the January date is relevant, this would then mean that in my situation with a January birthday, I would never trigger the row relevant for 2008, and the other prior threads are not relevant anymore as it all depends on the luck of when is your birthday.

Also - if the current Mercer calculation is correct, doesn't it mean that if I put my pension into payment on any month after my birthday, the payments up to that date in the starting year are just money thrown away - that can't be right can it?

So playing around a bit, May 2008 to June 2034 = 26 complete years so i would expect the known revalution + their future assumptions to = 26, on the basis thats how they seem to do it and apply the ERF at that point, we don't know what inflation will be 2025 -> 2034 inclusive = 10 future years + 16 known as discussed, then apply whatever ER reduction they give.

You are correct a lot of places talk about a 'normal retirement age', a lot of that is down to just simplifying what they are talking about, and because in most/all DB Schemes Early Retirement isn't automatically allowed in the Rules and is only allowed at trustee/company discretion , you only have the right to your benefits at Normal Retirement age (just that these days most have built in it is not allowed for everyone), so most just talk with reference to the NRA as every person in a DB Scheme can retire then, not all are allowed to retire early.

Right yes.  It's possible that they will come back and say it's a mistake and they just had copy and paste from a previous estimate that I requested only 6 months earlier for January 2024 - in fact I specifically planned to request those 2 quotes because I wanted to see if they correctly (in my opinion) made the adjustment for the anniversary date.

I'm probably over suspicous, but the fact that they are trying to confuse me by using weird index points for CPI calculations half makes me think that they know it's a grey area to do it like that and they don't want questions asked  

If they continue to say this is how they do it and it's allowed, unless someone else on these boards can cite other examples where other DB pension admins are doing it like that, or even better some kind of legal decision one way or the other on this exact point, my view is that they are not complying with their statutory obligations as they are required to revalue for known periods using the legislated method.

Whether I can actually force them to do that if they want to be difficult about it is another matter - I'm sure they have a bigger legal budget than me  

Edit:  I guess my suspicious nature is probably over stated because if the pension happened to have been deferred in 2009, I guess this would have worked in my favour - unless of course they are picking which way to do it to minimise what they have to pay!

GMPEqCommenter

My guess as to what is going on (possibly a bit late) - I don't know the scheme specifics so this is definitely a guess, but I hopefully have enough general pensions knowledge for it to be a vaguely useful one:

The statutory revaluation requirements apply when you retire at NRA. If you retire early, your pension has to be at least equal in value to the pension had you retired at NRA, but the statutory revaluation requirements don't apply in the same way.

The administrator is therefore trying to calculate what pension you would get if you retire at NRA, and then apply an early retirement factor that compensates for the fact that you're going to be paid the pension sooner and for longer. This is a valid alternative approach to just revaluing the pension to retirement date and applying a (different) early retirement factor then - it's a bit more complex, but makes it easier to be confident you're providing sufficient value.

You say you reach your NRA in January 2034. You left the pension scheme in May 2008. You therefore have 25 complete years in deferment to NRA. In theory, the deferred revaluations you get should be the statutory higher rate deferred revaluations for 25 years to 2034. Unfortunately those haven't been published yet, so the administrator has to guess them. They're doing this by looking at how the statutory revaluations would be calculated, giving you as much as they can of them and guessing the rest.

25 years of statutory deferred revaluations would be calculated (broadly) as 

Max [1.05^25, CPI @ September 2033 / CPI @ September 2009 

x RPI @ September 2009 / RPI @ (2033 - 25 = 2008)]

We know CPI up to September 2023 in the more recent quotation. The administrator is doing:

Max [1.05^25, (1+ CPI assumption) ^(2033 - 2023) x CPI @ September 2023 / CPI @ September 2009 

x RPI @ September 2009 / RPI @ (2033 - 25 = 2008)]

The "(1+ CPI assumption) ^(2033 - 2023)" will be the 10 years of Aviva assumptions mentioned in their reply.

CPI @ September 2023 / CPI @ September 2009 x RPI @ September 2009 / RPI @ (2033 - 25 = 2008)

= 132.0 / 87.1 x 215.3 / 218.4 = 1.496. 

This is actually consistent with 15 years to 2024 of higher rate statutory deferred revaluations - note that RPI inflation decreased over the year to September 2009, which is why the total inflation is lower than the figures you were getting when you were looking at the CPI in isolation.

So overall I don't think the administrator is doing anything obviously immediately obviously unreasonable, although I can't vouch for whether it's correct under the Rules of the Scheme or not.

Pat38493

GMPEqCommenter said:

My guess as to what is going on (possibly a bit late) - I don't know the scheme specifics so this is definitely a guess, but I hopefully have enough general pensions knowledge for it to be a vaguely useful one:

The statutory revaluation requirements apply when you retire at NRA. If you retire early, your pension has to be at least equal in value to the pension had you retired at NRA, but the statutory revaluation requirements don't apply in the same way.

The administrator is therefore trying to calculate what pension you would get if you retire at NRA, and then apply an early retirement factor that compensates for the fact that you're going to be paid the pension sooner and for longer. This is a valid alternative approach to just revaluing the pension to retirement date and applying a (different) early retirement factor then - it's a bit more complex, but makes it easier to be confident you're providing sufficient value.

You say you reach your NRA in January 2034. You left the pension scheme in May 2008. You therefore have 25 complete years in deferment to NRA. In theory, the deferred revaluations you get should be the statutory higher rate deferred revaluations for 25 years to 2034. Unfortunately those haven't been published yet, so the administrator has to guess them. They're doing this by looking at how the statutory revaluations would be calculated, giving you as much as they can of them and guessing the rest.

25 years of statutory deferred revaluations would be calculated (broadly) as 

Max [1.05^25, CPI @ September 2033 / CPI @ September 2009 

x RPI @ September 2009 / RPI @ (2033 - 25 = 2008)]

We know CPI up to September 2023 in the more recent quotation. The administrator is doing:

Max [1.05^25, (1+ CPI assumption) ^(2033 - 2023) x CPI @ September 2023 / CPI @ September 2009 

x RPI @ September 2009 / RPI @ (2033 - 25 = 2008)]

The "(1+ CPI assumption) ^(2033 - 2023)" will be the 10 years of Aviva assumptions mentioned in their reply.

CPI @ September 2023 / CPI @ September 2009 x RPI @ September 2009 / RPI @ (2033 - 25 = 2008)

= 132.0 / 87.1 x 215.3 / 218.4 = 1.496. 

This is actually consistent with 15 years to 2024 of higher rate statutory deferred revaluations - note that RPI inflation decreased over the year to September 2009, which is why the total inflation is lower than the figures you were getting when you were looking at the CPI in isolation.

So overall I don't think the administrator is doing anything obviously immediately obviously unreasonable, although I can't vouch for whether it's correct under the Rules of the Scheme or not.

I'll have to go through your calculations tomorrow in detail as it's a bit late today to fully understand it

However if it's correct for them to do it this way, they may say they have calculated sufficient value, but in the case of the 2008-2009 changes in the statutory revaluation tables, if they were doing it the other way that you describe (and that was discussed in an earlier thread here as the way it is usually done), in my case the starting point for the pension before ERF would be about £1000 per year higher, assuming they count forward the complete number of years to the retirement date.

Also, the retirement pension estimate for retiring in June, was only about £20 per year higher than retiring in January on a pension estimate of around £16400 per year.  No matter what approach they are using I would have expected it to be about £200-300 higher based on my calculations.  Of  course they may say that they have changed the ERFs due to changing assumptions or suchlike.

It's quite hard to understand what is going on in all this, but it seems important because we had some threads here indicating that you can get a lot more pension by retiring after the anniversary of the deferement date.  However in my case I am seing the opposite - I am seeing that if I retire several months after my birthday, I lost several months of pension payments in exchange for a negligable increase in the actual pension amount calculated.

I recently asked an IFA about this on a Q&A session and he said that even from IFA perspective, DB pensions estimates are a bit of a black box and they sometimes feel like they change randomly from one year to the next without being able to get clear explanations.

GMPEqCommenter

Neither approach to calculating revaluations produces systematically higher deferred revaluations* (i.e. different combinations of dates of leaving service, retirement and Normal Retirement Date will vary as to which approach gives the higher answer), but on an individual level there will certainly be winners and losers, and it looks like in this case the approach Aviva is using is the worse one for you.

(*Actually, people deliberately choosing their retirement date to maximise revaluations under the "revalue to DOR" approach may change this, but I suspect Aviva would consider the lack of ability to do this under the "project to NRA" approach a feature rather than a bug.)

Buy-outs are outside my core area of expertise, but my (not fully reliable) understanding is that insurers update their factors a lot more often than pensions schemes do, so the early retirement factors probably have changed.

Pat38493

So the plot thickens.

I have now received a reply from Mercer:

"We refer to your request regarding your query within the above-mentioned scheme.

Please accept our sincere apologies for the delay and any inconvenience caused in this regard.

Please be advised that all calculations are now being carried out by a third party. According to the response received from the third party, there are 16 years of revaluations from the date of your leaving to your retirement date, which is on xth June 2024. However, it's important to note that your normal retirement age is 65 years, not 55 years and 5 months.

The third party will revalue your benefits to age 65 years, specifically on x June 2034, using the known rates and the projected rates which they had previously sent. After revaluation, they will then reduce the benefit using the early retirement factor..

Meanwhile, if you require any further assistance, please do not hesitate to contact us."

Since my birthday is in January, this reply appears to suggest that I am right and their original reply was wrong.  However I think they might be miscommunicating between them and their 3rd party as if that was the case, I would expect an acknowledgement that the last estimate they sent me was incorrect and they need to recalculate it (unless the use of future tense indicates that this part of the reval was not included in the estimate process and is only done on retirement).

The other thing is that if they are doing it "their" way and they have done it that way for all other members, I doubt they are going to change it without a big fight because that would then imply that they need to go back and check their calculations on every member including already retired members.

They also seem oddly reluctant to tell me what the current ERFs actually are.  Is this like a trade secret or something?

GMPEqCommenter

That could be clearer. My best guess is that they're saying "yes, there are 16 historic revals to DOR, but that's not relevant because the pension is being revalued to 65 instead", and the reference to June 2034 rather than January 2034 is just an error, but I'm having to make some significant assumptions there.

Insurers tend to view their assumption sets as trade secrets (I doubt Mercer know what they are), but I'm less sure about the resulting ERFs. If they won't tell you the specific ERFs they used for your calculations that seems odd - Mercer should definitely have it.. If they won't provide you with a full ERF set, that may be less odd - the insurer may be providing factors on an individual member basis as needed for quotations rather than producing full sets of tables in the way pension schemes usually do.

FIREDreamer

Usually pensions are revalued up to early retirement date using known historic revaluation rates with any early retirement factor applied to that. Usually around 4% for every year early.

If using future assumptions up to NRD then I would assume the early retirement factors would be much higher to reflect this.

GMPEqCommenter

FIREDreamer said:

Usually pensions are revalued up to early retirement date using known historic revaluation rates with any early retirement factor applied to that. Usually around 4% for every year early.

If using future assumptions up to NRD then I would assume the early retirement factors would be much higher to reflect this.

I'd agree that 4% p.a. simple to pension revalued to DOR is probably the single most common set of factors I've seen, but there are still a lot of schemes that use something different. In particular, it's probably more generous than cost neutral on a lot of bases schemes are likely to be using these days, which means people retiring early on them are expected to cost the scheme money.

If I agree that it's on top of extra increases up to NRD the ERF reduction is going to be bigger to compensate.

Pat38493

GMPEqCommenter said:

That could be clearer. My best guess is that they're saying "yes, there are 16 historic revals to DOR, but that's not relevant because the pension is being revalued to 65 instead", and the reference to June 2034 rather than January 2034 is just an error, but I'm having to make some significant assumptions there.

Insurers tend to view their assumption sets as trade secrets (I doubt Mercer know what they are), but I'm less sure about the resulting ERFs. If they won't tell you the specific ERFs they used for your calculations that seems odd - Mercer should definitely have it.. If they won't provide you with a full ERF set, that may be less odd - the insurer may be providing factors on an individual member basis as needed for quotations rather than producing full sets of tables in the way pension schemes usually do.

This raises a few questions though
1) If the insurance company can treate their assumption set as trade secret, how can anyone be sure that they are correctly fulfilling the trustee duty to be fair to all members?
2) Presumably if the original rules of the insured schemed dictated elements of these assumptions or ERFs, the insurance company cannot just decide to change that on a whim?
3) If they won’t provide you with a full ERF set as a point in time snapshot, how are you supposed to decide whether to take the pension or not, especially considering the estimate says that it is an estimate and will be recalculated again if you decide to go ahead and put it into payment.

GMPEqCommenter

1) Generally by comparing whether the results the insurer get are broadly similar to the results you get when you use your own basis that you consider reasonable. Actuarial companies will have their own sets of "solvency assumptions" which they expect to broadly replicate typical assumptions used by insurers (although these are more typically used to say "we expect it to cost £X to have the insurer take the Scheme off our hands" rather than "we expect an insurer to use factors of Y". Exactly how often checks on the factors are done is getting too far outside the scope of my knowledge, although I have a vague memory of hearing of checks being done on sample sets of factors as part of a buy-in process).

Obviously that requires quite a bit of specialist knowledge, but so would interpreting whether an assumption set was reasonable in the first place - it's fundamentally a complicated area.

2) In practice for a lot of schemes the Rules don't go into that level of detail - they'll just say something along the lines of "reduced for early payment on a basis agreed by the Trustee" (possibly with some requirement for the actuary to certify it as reasonable, and even if that's not explicitly there the Trustee would usually be getting actuarial advice anyway). There may be exceptions - for example, you may get some Rules that specifically say "4% p.a. simple per year" - and in those cases I'd expect it to be agreed with the insurer that they'd keep the same factors.

3) I agree it will be a bit of a gamble. Even if they did provide you with a factor set that said "if you go 5 years early, the reduction will be X, at 4 years early it will be Y", by the time you get to 4 years early the factors may well have changed anyway. Indeed, I wouldn't bet against them having changed by the time you get to 4 years 11 months early. Trustees like having fixed factor sets so members can plan (alongside other advantages like it being easier and cheaper to administer), and when they do change their factor sets, some Trustees may honour previous quotations if they would be less generous under the new factors. My impression (with the caveat again that this is outside my area of expertise, and I might be wrong), is that insurers care more about the factors accurately representing the actual cost of retirement, so they'll change as expectations of future inflation and future investment returns do (and future expectations of mortality, but those tend to be more stable).