Which platform for separate crystallised and uncrystallised pots?

Triumph13

So the net upshot of TheTelltaleChart's excellent analysis is actually the opposite of the original idea in the thread! That's because the one circumstance where you would win by having separate pots is if you are in danger of hitting the lifetime limit for tax free cash.  This would increase the effective tax rate on the uncrystallised pot.  So you should actually put the growth assets in the crystallised portion.  LOL. 

Pat38493

Triumph13 said:

So the net upshot of TheTelltaleChart's excellent analysis is actually the opposite of the original idea in the thread! That's because the one circumstance where you would win by having separate pots is if you are in danger of hitting the lifetime limit for tax free cash.  This would increase the effective tax rate on the uncrystallised pot.  So you should actually put the growth assets in the crystallised portion.  LOL. 

My original statement
All other things equal, the priority order for keeping your highest growth assets
1) ISA
2) Uncrystallized pension
3) Crystallized pension
4) GIA

Telltale said this was "plain wrong".

I still stand by my original statement since Telltale did not disprove it in any meaningful way - changing the opening allocations after creating a notional split is moving the goalposts.  My original comment was based on the allocations being the same in either situation.  My comment was about what happens in the future, if you move your pot from a split provider to a notional split provider, or vice versa, without changing your asset allocations.

You win by having separate pots with the same allocation mix, if you get a good sequence of returns.  In worst case scenarios, you are worse off.

Creating a notional split disconnects the taxability status of the pots from the overall investment mix going forward.

I don's disagree with Telltale's analysis proving that the taxability status of different items gives a different net value, but I never claimed anything different.

I also accept that there may be some cases for individuals where the order needs to be changed for various reasons.

I even don't disagree that this is largely meaningless in the long run because you will be (hopefully) regularly rebalancing anyway, but that doesn't mean my theoretcial statement was wrong.

TheTelltaleChart

You want proof?

For any asset class which forms part of a portfolio, let's call the initial gross value of that asset class in each account (i.e. in ISA, crystallised pension, etc):

g1, g2, g3, ...

Let's call the tax application ratio, i.e. what each account needs to be multiplied by to give its net value (e.g. 1 for an ISA, 0.8 for a crystallised pot if your tax rate is 20%, etc):

t1, t2, t3, ...

Therefore the initial net value held in this asset class is:

N0 = (g1 * t1) + (g2 * t2) + (g3 * t3) + ...

At any later date (x), let's call the return on this asset class since the initial date, expressed as a ratio (e.g. 1 for no change, 2 for doubling, 0.58 for a 42% loss, etc):

R

If there is no change in the portfolio's allocations to this asset class, between the initial date and date x, then the gross values held in it in each account are therefore now:

(g1 * R), ( g2 * R), (g3 * R), ...

[Note: using the same value of R in all accounts works for all ISA and pension acounts, but not for GIAs, where R may be reduced due to taxes on income/gains.]

Therefore the net value held in this asset class at date x is:

Nx = (g1 * R * t1) + (g2 * R * t2) + (g3 * R * t3) + ...

[Note: that the tax application ratios for each account is unchanged, despite the account growing or shrinking in size, is not always true. There are important exceptions, such as getting part of your taxable pension withdrawals taxed at different rates, or hitting the lifetime TFLS limit.]

That equation can be rearranged as:

Nx = R * ((g1 * t1) + (g2 * t2) + (g3 * t3) ...)

And then as:

Nx = R * N0

That is to say: the net value of your holdings in this asset class, at any future date, depends only on the net value of your holdings in it at the start date and on the returns in the asset class between the 2 dates.

But, since the net value of the whole portfolio is, at any date, simply the sum of the net values of all the assets classes held on that date, we can also say:

The net value of the portfolio at any future date depends solely on the amount of net value assigned to each asset class at the start date and the performance of each asset class during the subsequent period.

In other words (apart from the important exceptions noted above): if you get a different net result by assigning asset classes differently among accounts 1)-3), it can only be because you also assigned different initial net values to the asset classes. If you had assigned the same initial net values, you would have gotten the same net result. QED.

RockPools

I'm hoping for an idiots guide at the end of all this 

michaels

TheTelltaleChart said:

You want proof?

For any asset class which forms part of a portfolio, let's call the initial gross value of that asset class in each account (i.e. in ISA, crystallised pension, etc):

g1, g2, g3, ...

Let's call the tax application ratio, i.e. what each account needs to be multiplied by to give its net value (e.g. 1 for an ISA, 0.8 for a crystallised pot if your tax rate is 20%, etc):

t1, t2, t3, ...

Therefore the initial net value held in this asset class is:

N0 = (g1 * t1) + (g2 * t2) + (g3 * t3) + ...

At any later date (x), let's call the return on this asset class since the initial date, expressed as a ratio (e.g. 1 for no change, 2 for doubling, 0.58 for a 42% loss, etc):

R

If there is no change in the portfolio's allocations to this asset class, between the initial date and date x, then the gross values held in it in each account are therefore now:

(g1 * R), ( g2 * R), (g3 * R), ...

[Note: using the same value of R in all accounts works for all ISA and pension acounts, but not for GIAs, where R may be reduced due to taxes on income/gains.]

Therefore the net value held in this asset class at date x is:

Nx = (g1 * R * t1) + (g2 * R * t2) + (g3 * R * t3) + ...

[Note: that the tax application ratios for each account is unchanged, despite the account growing or shrinking in size, is not always true. There are important exceptions, such as getting part of your taxable pension withdrawals taxed at different rates, or hitting the lifetime TFLS limit.]

That equation can be rearranged as:

Nx = R * ((g1 * t1) + (g2 * t2) + (g3 * t3) ...)

And then as:

Nx = R * N0

That is to say: the net value of your holdings in this asset class, at any future date, depends only on the net value of your holdings in it at the start date and on the returns in the asset class between the 2 dates.

But, since the net value of the whole portfolio is, at any date, simply the sum of the net values of all the assets classes held on that date, we can also say:

The net value of the portfolio at any future date depends solely on the amount of net value assigned to each asset class at the start date and the performance of each asset class during the subsequent period.

In other words (apart from the important exceptions noted above): if you get a different net result by assigning asset classes differently among accounts 1)-3), it can only be because you also assigned different initial net values to the asset classes. If you had assigned the same initial net values, you would have gotten the same net result. QED.

I thought we were discussing what would happen where r1<>r2 rather than r1=r2=R?

Pat38493

We can simplify this down to things that have been stated by others on these boards many times:

1) All other things equal, you should try to keep your highest growth assets in wrappers with the lowest marginal tax rate.  

2) If you have a pot of £100K in drawdown in cash, and pot of £100K in equities in an uncrystallised pension, and you leave it there for x amount of time, you will have a high statistical probabiilty of ending up with more money at the very end than the person who put the equities in the drawdown pot and the cash in the uncrystallised.  This is because more often than not, equities grows more than cash.  The marginal tax rate of the uncrystallised pot on withdrawal is lower.

3) If you move the above pot into a provider like II that has a notional split, you can no longer guarantee at all times on every day, that the amount of equities you have will correspond to the amount uncrystallised, because it's now a fixed % and the equities will grow at a different %.  Yes of course you can continually adjust it to get the outcome you think you should have, but the bottom line is that notional split pots behave differently to separate pots regarding ongoing future growth.

4) Point 3 will probably have negligable impact on your long term wealth if you are regularly rebalancing, but it might be more of a difference if you are lazy and never rebalanced your portfolio regularly.

Telltale is basing everything on trying to achieve the desired net (future) value of the overall pension assets at any given point in time, which is all well and good in theory, but in reality we don't rebalance our pension every second or every day.  I also suspect there is something missing from the equations once the split becomes notional because the assets which form t1, t2, t3 will be changing daily (or even continually if the fund is in ETFs).

TheTelltaleChart

michaels said:

I thought we were discussing what would happen where r1<>r2 rather than r1=r2=R?

Those equal Rs are all for the same asset class (e.g. global equities grow at the same rate, regardless of which pot they're in). Different asset classes will grow at different rates.

TheTelltaleChart

Pat38493 said:

We can simplify this down to things that have been stated by others on these boards many times:

1) All other things equal, you should try to keep your highest growth assets in wrappers with the lowest marginal tax rate.

No.

2) If you have a pot of £100K in drawdown in cash, and pot of £100K in equities in an uncrystallised pension, and you leave it there for x amount of time, you will have a high statistical probabiilty of ending up with more money at the very end than the person who put the equities in the drawdown pot and the cash in the uncrystallised.  This is because more often than not, equities grows more than cash.  The marginal tax rate of the uncrystallised pot on withdrawal is lower.

It's because you start with (the same gross value but) a higher net value in equities in the first case. More equities usually (except when it doesn't!) gives you a higher return. Nothing to do with one asset class being better suited to one kind of pot.

3) If you move the above pot into a provider like II that has a notional split, you can no longer guarantee at all times on every day, that the amount of equities you have will correspond to the amount uncrystallised, because it's now a fixed % and the equities will grow at a different %.  Yes of course you can continually adjust it to get the outcome you think you should have, but the bottom line is that notional split pots behave differently to separate pots regarding ongoing future growth.

The move to a pot with a notional split itself changes the net value in equities. Starting with a different net value in equities gives a different outcome. If the net value in equities it gives you isn't what you wanted, you can change it immediately after moving to the notional split provider. You do not need to make any further continual adjustments after that.

4) Point 3 will probably have negligable impact on your long term wealth if you are regularly rebalancing, but it might be more of a difference if you are lazy and never rebalanced your portfolio regularly.

N/A, because point 3 is wrong.

Telltale is basing everything on trying to achieve the desired net (future) value of the overall pension assets at any given point in time, which is all well and good in theory, but in reality we don't rebalance our pension every second or every day.  I also suspect there is something missing from the equations once the split becomes notional because the assets which form t1, t2, t3 will be changing daily (or even continually if the fund is in ETFs).

You're imagining things in the proof which aren't there. There is no assumption of continual rebalancing, or any rebalancing.

When pots are merged into a notionally split pot, t1 is set to a specific value for that pot at that time, and is not changed at all by subsequent fluctuations in investment values; it is only changed by further mergers of pots, or crystallisations, or drawing from either notional part of the pot, or making new contributions to the (uncrystallised part of the) pot.

Triumph13

Let's try a really simple example to prove TheTelltaleChart's point.

Assume that you are going to be a basic rate taxpayer in retirement, and your asset allocation calls for you to have £85k after tax value of equities and you have a choice of two places to hold it - ISA or uncrystallised pension.  You achieve that by either having £85k of equities in the ISA or by having £85k / 85% = £100k in the pension.  If equities then double in value you would end up with either £170k in the ISA, or £200k gross = £170k after tax in the pension.

Exactly the same logic applies to any asset class and to any number of wrappers.  It only makes a difference where you put what if the amount you hold in a particular wrapper is enough to risk making a change in the tax rate applied to withdrawals from that wrapper.

Pat38493

Regarding point 3 - if I transfer a pot from a real split to notional split whilst keeping the exact same allocation by gross value, surely the net value of the pot hasn't changed on the day of the transfer.  In fact, I could change the theoretical allocation to any asset allocation whatsoever, and the net value would not change (e.g. if it was a Sunday and markets were closed, the net value of the pot is driven by the notional split and nothing else)?

Let's say the notional split if 50/50 and the pot value is £1m.  The net value is fixed on that day at that moment in time - you can theorize any change to asset allocation that you like but the net value is determined only by the gross value and the fixed notional split % value, nothing else?

My point is that I am no longer in moment to moment control of how the assets are allocated to each wrapper as there is no tracking of this anymore.

Therefore I am still struggling to see in Telltale's earlier example, how would I be able to know to make an adjustment of (exactly) £5K on the day that I move the pot into a notional split?  I might know that an adjustment will be  needed, but how do I calculate the 5K number?  The notional split at the time of transfer is calculated exactly based on the real split that was moved in, so the numbers should be identical.  Even if I calculate the target gross allocations based on net results, if the split did not change yet and the target net allocation is the same, surely I will get the same answer?

Regarding the other points, I guess I have to admit I am wrong if your objective is to achieve a fixed net asset allocation after theoretical tax is deducted.  However this is not how I have really looked at it up to now.  In fact, I have paid hundreds of points for an retirement planning course from an IFA, and watched many videos from IFAs on Youtube, and none of them ever pointed out that you have to decide your asset allocation based on the calculated net value of your pots, and then recaluclate the asset allocation back up to the wrapper you want the asset to be in.  In fact the video had a worked example in detail and this adjustment was not included as far as I remember.

As it happens, I am calculating my asset allocations using a cash flow ladder which was populated from data from a cash flow planning software, so probably these factors are already taken into account in another way, since the expected tax is already included in the annual tax flow.

I am getting better results when putting the same gross allocations in lower marginal tax wrappers.

Further, even when accepting your POV, as soon as you start to take into account that having the equities in higher marginal tax wrappers will be statistically more likely to put you into a higher tax bracket at some future point, my point one actually becomes "yes" but not for the reason I was stating.