Introduction
1. There is, of course, already a tax-geared penalty regime which applies to the GAAR, namely, that in Schedule 24 of the Finance Act 2007 for under-declared liabilities to tax. However, the trigger for that regime is the existence of a degree of culpability on the part of the taxpayer for that under-declaration. There will, of course, be instances where the fact that the taxpayer’s return (showing, say, tax of 20) culpably fails to include a tax saving (of, say, 80) arising in consequence of abusive behaviour that is later counteracted by the operation of the GAAR. However, the circumstances where the taxpayer will be culpable for not showing the right tax (100) on his return may be limited given that (a) he will have been advised that the behaviour generates a tax saving (of 80) and is not abusive and (b) there is some scope for policy decisions by HMRC to affect whether the GAAR is in fact operated.
2. Where the GAAR operates it restores the taxpayer to the position he would have been in had he not engaged in the abusive tax conduct (in the example above, his tax liability is restored to 100). However, professional fees, any economic costs, and the remote possibility of Finance Act 2007 penalties aside, the effect of the GAAR will simply be to restore him to the position he would have been in had he not engaged in the abusive behaviour at all.
3. Discouraging abusive behaviour, therefore, requires that it carry a risk over and above the taxpayer being restored to the status quo ante (i.e. a tax liability of 100). The answer is a special penalties regime. An appropriately drawn regime will encourage taxpayers to interrogate their advisers as to whether there is a risk that proposed planning will be abusive. It will also encourage advisers to be more cautious about suggesting abusive planning. I am aware of no reason of principle why one should not discourage abusive planning through the use of penalties.
Policy Choices
4. A penalties regime will also likely have what one might describe as a modest penumbral effect. In other words, it will not merely discourage practices caught by the GAAR but it will discourage practices that might be caught. And the higher the penalty, the stronger the discouragement.
5. Whilst this is bound to have short term positive fiscal impacts, it must also be recognised that the penumbral effect could act as a discouragement to taxpayers to engage in acceptable or pro-purposive tax planning. (I assume this to be an undesirable consequence although others might disagree).
6. One might reasonably assess the penumbral effect to be modest given that the GAAR is tightly drawn. Moreover, one can certainly modulate the extent of the penumbral effect through the rate at which one sets the penalty. The other mechanism that one might adopt is arrangements whereby one mitigates the penalty according to whether the taxpayer had good reason to consider the arrangements not to be abusive. A similar mechanism has been adopted to mitigate, for example, the effects of Follower Notices.
7. Addressing these points in turn, I consider the GAAR (importantly, as presently drawn) to be tightly focused. An experienced adviser approaching the question ‘Is this planning abusive?’ with an appropriately open mind is, I consider, likely to be able to reach a reasonably accurate assessment of the answer. Putting the matter another way, I consider the quality of the GAAR in its present form is likely to have as its consequence that the introduction of penalties is likely to have a limited, and in some measure desirable, penumbral effect.
8. Where there is a penumbral effect, that effect will be strongest on arrangements which are abusive but are nevertheless not caught by the GAAR. As others have observed, the GAAR operates only when arrangements are (putting the matter somewhat crudely) clearly abusive. This is the consequence of a policy call having been taken by Parliament about where to strike the balance between maximising legal certainty and maximising the reach of the GAAR, a policy call articulated in the language used in, for example, section 207(2) Finance Act 2013. The penumbral effect will be strong where the arrangements are (again crudely) probably abusive but there is nevertheless some doubt about whether they are. This particular penumbral effect may be regarded as desirable.
9. As to the size of the penalty, plainly it must be tax-geared i.e. a function of the amount of tax the abusive practice purports to ‘save’ (80 in the example). Plainly also it must appropriately risk the abusive behaviour. Different people will reasonably arrive at differing assessments of what the right level of risk is. My personal view is that 100% is not unreasonable and it also has a symbolic attractiveness. The effect will be that the taxpayer with an initial liability of 100 who engages in abusive behaviour reducing his tax bill to 20 will, when counteracted and penalised, face a final liability of 180.
9. As to mitigation, there will always be cases where an application of a tax geared penalty will lead to genuine injustice. One will also want to encourage taxpayers engaging in behaviour that might be abusive to tell HMRC. And one will wish to incentivise taxpayers to come clean if they have engaged in abusive behaviour.
10. However, the scope of mitigation should be narrowly drawn.
11. I consider a discount of 50% should be given where the taxpayer discloses on his tax return that he has engaged in conduct that may be considered abusive (so the disclosing abusive taxpayer’s liability is 140).
12. I consider a discount of 25% should be given where the taxpayer who does not initially disclose later discloses (160). This discount is not high. However, it must be smaller than that for the person who discloses on his tax return otherwise one fails to incentivise the taxpayer not to make an initial disclosure – and to wait and see whether HMRC refers his transactions to the GAAR Panel.
13. As to injustice, one must recognise (a) that in certain circumstances the taxpayer may quite reasonably rely on his adviser (b) there are powerful commercial incentives for advisers to give over-optimistic advice about tax risk (professional fees are often a function of whether a taxpayer enters into arrangements rather than whether he seeks advice on them). One has regard to these competing considerations, it seems to me, by mitigating for injustice only where the taxpayer can demonstrate that he took a genuine interest, appropriate to the tax saving, in whether the arrangements in question were or were not abusive. Although this may be a matter of detail which is downstream of these brief thoughts, one may wish to stipulate particular factors that a tax tribunal will accept as demonstrating the presence of a genuine interest including whether the taxpayer took advice from a specialist unconnected to the promoter of the arrangements.
Postscript: I’d suggest readers might have a look at this Schematic which sums up what the GAAR does – and how a GAAR specific penalties regime might affect its operation.
Waiting for Godot 
Would it not be simpler to use the existing regime rather than setting up a new one?
The problem you have identified is that a taxpayer who has been advised that the transaction is reasonable cannot have been careless (as they took care) or have deliberately mis-stated their liability (as they stated the liability they thought was due) and so cannot be penalised.
A simple solution would therefore be to treat any understatement of tax which is corrected by the GAAR as if it were a deliberate understatement, ignoring the question of whether or not it actually was careless or deliberate. That would give you a penalty of 20-70% depending on whether the disclosure was prompted or not.
This is a little less than the figures you suggest. One might therefore change it to the “deliberate and concealed” rate, which would give 30-100%; I have some difficulty with the “concealed” aspect, although I suppose that if you consider abuse to be as abhominable as concealment then that would be fair enough.
Alternatively (or as well) one could treat all adjustments as unprompted, which would give you 35-70% for deliberate understatement or 50-100% if you use the higher rate for “deliberate and abusive”. The latter is pretty much what you were suggesting: 100%, with a potential 50% discount for co-operation. I think that “unprompted” is a fair enough view, given that a prompted disclosure is normally of the form “I have erred, please take my money” rather than simply “You may disagree with me here, but I maintain my position” – the latter is disclosing information, rather than disclosing an understatement.
All this would take is adding “or where the understatement is corrected by application of the GAAR” to the existing category of your choice, and so I would commend it to the OTS.
Surely 100 (your proposed GAAR super-penalty) is somewhat more than 100% of 80 (the GAAR adjustment, the “potential lost revenue”)? Why is the taxpayer paying a penalty related to the 20 of tax that was correctly returned and paid?
As the other Andrew suggests, if you want to turbo-charge penalties for the evil tax avoiders who are caught by the GAAR, why not just deem GAAR adjustments to be either “careless” or “deliberate”?
Such heavy penalties are no doubt “criminal” in ECHR terms, so for example taxpayers would be entitled to legal aid to defend themselves.
A few reflections, offered in a personal, not ICAEW, capacity.
In paragraph 2, you seem to be making light of the hassle factor of being ‘found out’ through the GAAR. The associated costs are unlikely to be cheap, and the hassle of being hauled through the GAAR process and forced, ultimately, to pay would seem to outweigh any timing benefit of using abusive avoidance deliberately. It’s a sort of fiscal ‘stop and search’.
Turning to paragraph 8, the ‘stop and search’ comparison highlights the problem with addressing ‘probable abuse’ using the penumbra. We (unless we’re Daily Mail readers) don’t think the benefit of catching the guilty outweighs the social cost of hassling the innocent, even if they’re doing something that makes them probably guilty. I’m also unclear whether ‘clearly abusive’ is capable of subdivision beyond the binary ‘clearly abusive’/’not clearly abusive’. If it’s not clearly abusive, then I find it hard to cheer its being hassled out of existence.
There’s a pretty important conceptual difference. The existing penalty regime punishes poor behaviour in relation to your tax liabilities – did you misdeclare them carelessly or deliberately. The proposed penalty regime doesn’t do that. It just asks whether you have transacted in a proscribed space – and if you have it (in effect) increases the tax burden for so transacting. There could (as I have said) be mitigation of that increased liability on the grounds of good behaviour but the fundamental conceptual difference remains.
Have addressed Andrew 1’s point. As to yours, where do I say that there’s a penalty liability of > 80 (or 100%)? (If I do say that, that’s my error).
As to para 2, I don’t agree (or necessarily agree) with that assessment. I have seen (indeed, I have litigated) transactions that would clearly be abusive under the GAAR for many hundreds of millions of pounds. Those inclined to transact at those sorts of levels won’t much be put off by a little hassle.
As to para 8, I don’t think your analogy stands up. We don’t think the benefit of catching the guilty outweighs the social costs of imprisoning the guilty. We certainly do think the benefit of catching the guilty is worth imposing a burden on the innocent – such as the obligation to submit (in appropriate cases) to stop and search or (to step out of the analogy for a moment) HMRC’s right (long established) to select random returns for enquiry.
Jolyon
In your paragraph 9 you discuss the need to appropriately risk the abusive behaviour. However, hasn’t the very invoking of the GAAR done that for us? So does mitigation of the penalty really need to play a part?
However, I have a fundamental objection: I think Ed Balls is being incredibly unimaginative here. How much impact will this have in comparison with, say, an Accelerated Payment Notice? How many taxpayers and schemes will be impacted and how much dosh will the Treasury actually get? The tighter you draw legislation, the more secure it is. Also I would posit though, the tighter it is drawn, the fewer the examples. This may make core Labour voters feel all warm and snuggly, but else will really be achieved?
Perhaps even more fundamental, what about ‘The Boys’, those who supply the ‘professional comfort’ to the , often, unsuspecting client? What about the promoters? At present the financial risk is borne almost exclusively by the client, why should that be?
The GAAR doesn’t ‘risk’ the abusive behaviour if the effect of the GAAR applying is merely to return the taxpayer to the position he would have been in had he not engaged in that behaviour (less the ‘friction’ of professional costs). And that’s the problem with the status quo.
As to what this will yield, and whether there is more to be done, I don’t want to get involved in that debate other than saying what I’ve said elsewhere.
As to your point about risk falling on the client, I’ve addressed that (at length) in my Hardman lecture
Pingback: Waiting for Godot | Avoidance Transactions, the GAAR, Penalties and Penumbras: a Schematic
Is transacting in a proscribed space not an example of “poor behavior”?
it would seem particularly so if the impact of doing so is mitigated by good behavior.
My suggestion would simply increase the number of behaviours which are regarded as poor because they lead to your declared liability being an understatement:
– Being careless
– Deliberately understating it
– Deliberately concealing it
– Abusing tax law to justify your understatement
I do see one conceptual difference, which is that the first three are focused more on the behaviour of the taxpayer, and the last is potentially more related to the behaviour of the advisors on whom the taxpayer relies – one might argue that the appropriate course is to levy penalties on the promoters rather than the taxpayers. But if you’re wanting to penalise the taxpayer, then that conceptual difference is irrelevant
On the schematic in your other post, I think that (6) doesn’t really give clear water between (5) and (7).
The problem is that you can only distinguish between (1), (2), and (3) after the fact, whereas the penumbra effect is felt beforehand.
Picking that apart: the distinction between (5) and (6) is that (6) potentially comes under the GAAR because it might be abusive, and (5) doesn’t because it’s not.
However, if you know that you’re in (5) because you’re not being abusive then you’re actually in (4).
If you’re not sure whether you’re being abusive or not, you’re in either (5) or (6) but have no way of telling until HMRC conclude any enquiry.
So if you think you’re in (5) you have to acknowledge a risk that you’re in (6) – which means you might be in (7) – and so (5) and (6) are functionally identical.
The penumbra spans both (5) and (6) equally, and any discouragement applying to (6) will also apply to (5). So your potential con does exist.
Sorry Jolyon, I think I misunderstood your meaning by “risk”. You have used it as a verb, to provide a risk. I thought you were espressing ‘risk’ from the HMRC persepctive, the risk of tax loss. By attempting to introduce a notion of gradation of offence and setting penalty loadings according ly. My response to this stands, however: there is no need to consider penalty loadings for this once the GAAR has been invoked. It is “clearly” an avoidance scheme, within very tight parameters and so a set penalty can legitimately be applied. Why apply mitigatation at all?
I also stand by my first objection, the more imaginative measures have done more to change the climate than any notionally togh measures might. The Accelrated Payment Notice, or even just the suggestion of it, is changing persepctives ‘on the ground’.
As regards ‘The Boys’, I will clearly need to read your Hardman leacture; happy bedtime reading for me!
Andrew, If you’re right that you can only distinguish between (1) (2) and (3) after the fact then your conclusion follows. But I don’t think you’re right. I think you can distinguish before the fact. That’s what I do every time I give tax advice.
But if you can reliably distinguish between (1) and (2) then you can reliably distinguish between (5) and (6), in which case there is no penumbral effect in (5).
I personally don’t think I can *reliably* distinguish between (1) and (2). I can try to distinguish, and I can say how likely or otherwise I think something is to be in (2), but I’d expect HMRC to have their own view and the final decision is with the courts.
That is, I can identify stuff that I consider abusive, but I’d be slow to be unequivocal as to whether it would fall inside the (probably rather narrower) ambit of the GAAR.
Hence for me the GAAR has a penumbral effect that would help discourage planning that’s at the (6)-ward end of (5).
Personally however I quite welcome ammunition for persuading people to stick in (4), so your potential “con” is actually something of a “pro” for me 🙂
Ah, I was misreading this paragraph:
“The effect will be that the taxpayer with an initial liability of 100 who engages in abusive behaviour reducing his tax bill to 20 will, when counteracted and penalised, face a final liability of 180.”
Re-reading, I think you are adding 20 (disclosed tax) +80 (tax avoided) +80 (penalty)? (Which makes sense, when you apply a 50% reduction, that it falls to 140=20+80+40) I read that as an additional further liability of 80 (tax avoided) + 100 (penalty), on top of the 20 already paid. My mistake.
Would your putative GAAR-offending taxpayer also face further penalties of up to 200% under the Schedule 24 regime?
Good thoughts – thanks.
Plainly each of the vertical lines is a bit fuzzy – without that fuzziness there’d be no penumbral effect at all. So I agree with much of what you say. But the interposition of (6) between (5) and (7) limits (I think) the impact of the penumbral effect of (7) to (6). If you had a Scottish GAAR – by contrast – then (6) and (7) would be caught by the GAAR and the penumbral effect would certainly extend to (5). The (key) starting point is recognising that (as I put it early) (7) is reasonably – tolerably – closely defined.
No prob and no idea respectively. I’m prepared to imagine a penalties regime and talk about what it might look like – it’s a step further to look at its relationship with completely separate bits of the tax code… although it would be odd if one was better off being GAAR abusive than GAAR abusive and dishonest.
They are only completely separate bits of the tax code if you choose to deal with them separately. If you are proposing a new layer of GAAR penalties, on top of the existing penalty regime, then a taxpayer could face a GAAR penalty of up to 100% and then a regular penalty of up to 200% too. Is that the intention?
The two could be rolled together, so behaviour found to be subject to GAAR counteraction could be an aggravating factor, in the same way that disclosure and cooperation are currently mitigating factors. The aggregate level of penalties would reflect the perceived culpability in a taxpayer’s particular circumstances.
“The problem [Jolyon has] identified is that a taxpayer who has been advised that the transaction is reasonable cannot have been careless (as they took care) or have deliberately mis-stated their liability (as they stated the liability they thought was due) and so cannot be penalised.”
If that is indeed the ‘problem’ then I don’t see it as a problem. See below for a case where penalties were levied after a purchased ‘scheme’ failed. The taxpayer was presumably advised that the transaction was reasonable yet the Tribunal agreed a penalty was due. Do we really need yet more laws when the current ones seem to be working?
Litman V HMRC [2014] UKFTT 089 (TC).
Another point perhaps worth making here is that, as Graham Aaronson mentioned in para 5.48 of his GAAR report, a special GAAR penalty regime “would be seen as presenting an irresistible temptation to HMRC to wield the GAAR as a weapon rather than to use it, as intended, as a shield. For this reason I do not consider that it would be appropriate to include any provisions for applying special rates of interest or penalties to tax recovered by use of the GAAR.”
HMRC accepted that in the GAAR consultation, saying that the GAAR is self-assessed and the usual penaty regime would apply.