MY INITIAL OBSERVATIONS ON THE UK GAAR
The following is an edited version of Unreasonably Reasonable, a speech I delivered at a Tax Chambers seminar in November 2012.
What is the difference, it may be asked, between a maze and a labyrinth? The answer is, little or none. Some writers seem to prefer to apply the word ‘maze’ to hedge-mazes only, using the word ‘labyrinth’ to denote the structures described by the writers of antiquity or as a general term for any confusing arrangement of paths. Others, again, show a tendency to restrict the application of the term ‘maze’ to cases in which the idea of a puzzle is involved…We shall find it convenient to leave questions of the definition of words until we have examined the various examples that exist.
– W. H. Matthews, Mazes and Labyrinths 1932
1.1 HMRC have recently published Graham Aaronson’s letter to HMT in which he and the study group of which he was the leader approve of the draft of the GAAR. The next draft is expected in December – this is likely to resemble what has already been produced. I suspect that most of the changes will pertain to the procedural aspects and provide more elaboration on the temporal scope.
1.2 It’s just worth considering the position in Canada which introduced a GAAR a few years ago in 1988. Since GAAR was introduced in 1988, the CRA GAAR Committee ruled that GAAR applied in 769 of 1,027 cases—or 75% of all cases referred.
1.3 The authorities there do not seem perturbed by the lack of clarity that the GAAR involves. As Justice Brandeis said:
If you are walking along a precipice, no human being can tell you how near you can go to that precipice without falling over, because you may stumble on a loose stone, you may slip, and go over; but anybody can tell you where you can walk perfectly safely within convenient distance of that precipice.
1.4 We may expect a similar stance to be very deliberately adopted by authorities (not only legislative and executive but also judicial) here in the UK. For this reason, I do not dwell on constitutional aspects of the GAAR, such as the disregard, which the GAAR typifies, of certain constitutional cornerstones, such as the need for governance by a rule of law. My considerations here are more practical. In the end, the UK position must be considered on the basis of the draft to be enacted here and it is in relation to that which I would like to make some observations.
First observation – Requiring reasonableness
1.5 It seems to me that through the GAAR, the legislator not only requires reasonableness – he even casts a sheen on what he means by reasonableness. He does the two things in one breath – through the definition of ‘abusive tax arrangement’. I treat the two aspects separately and my first observation pertains to the manner in which the legislator requires reasonableness. On this point, what is of interest is the expression ‘cannot reasonably be regarded as a reasonable course of action’.
(2) Tax arrangements are ‘abusive’ if they are arrangements the entering into or carrying out of which cannot reasonably be regarded as a reasonable course of action, having regard to all the circumstances including…
So, the question, as it appears to me, is whether a person (let us call him or her the questioner for now, without ascertaining who he or she is) can regard something (i.e. the entering into the arrangements) as a reasonable course of action and in doing so he or she is being reasonable.
If the questioner regards something as reasonable and he or she is not being reasonable in doing so, then we have an abusive arrangement.
If the questioner does not – or cannot – regard the arrangement as reasonable in the first place (whether or not reasonably), then is it arguable that there is not an abusive arrangement either? I think not. Because in such circumstances, the questioner simply cannot reasonably regard the arrangements as being reasonable.
1.6 How is this different to what the position would have been had the test simply been:
‘The arrangements…cannot be regarded [by the questioner] as a reasonable course of action.’…?
I suppose the concern there was that the questioner might unreasonably regard the arrangements as being reasonable. (Though one would have thought that in such circumstances the questioner ‘cannot’ regard the arrangements as a reasonable course of action anyway.)
I think that one purpose (probably the primary purpose) of ‘reasonably’ is therefore to make clear that the test is objective and that the questioner is a hypothetical questioner – not the taxpayer.
1.7 The effect of the phrase ‘reasonable course of action’ is to lower the standard. The test is not whether the hypothetical questioner considers something to be proper – a specific test – but rather whether he considers something to be unreasonable. It seems to be accepted that ‘reasonable’ refers not just to a particular action or result but rather to the intermediate spectrum which, whilst not considered perfect, is nonetheless not objectionable. In everyday language, the word is often used in the sense of ‘tolerable’. The legal sense is not that different. Consider the words of Lord Greene MR in Wednesbury Associated Provincial Picture Houses Ltd v Wednesbury Corpn  1 KB 223 at 229:
It is true the discretion must be exercised reasonably. Now what does that mean? Lawyers familiar with the phraseology commonly used in relation to exercise of statutory discretions often use the word “unreasonable” in a rather comprehensive sense. It has frequently been used and is frequently used as a general description of the things that must not be done. For instance, a person entrusted with a discretion must, so to speak, direct himself properly in law. He must call his own attention to the matters which he is bound to consider. He must exclude from his consideration matters which are irrelevant to what he has to consider. If he does not obey those rules, he may truly be said, and often is said, to be acting “unreasonably.” Similarly, there may be something so absurd that no sensible person could ever dream that it lay within the powers of the authority….
It would appear from the above that one has to be approaching absurdity before one can be regarded as being unreasonable. That reasonableness is a broad spectrum is in fact implicit in all the judicial review cases in which the application for review is made on grounds of unreasonableness. The test for the court in these cases is not (1) whether the court would have arrived at the very same decision which was reached by the public body concerned – but (2) whether the public body acted reasonably. The only reason why the two things are not the same is because ‘reasonableness’ refers to a broad spectrum and can encompass something which the court itself might not consider to be proper:
But it is important to remember in every case that the purpose of the remedies is to ensure that the individual is given fair treatment by the authority to which he has been subjected and that it is no part of that purpose to substitute the opinion of the judiciary or of individual judges for that of the authority constituted by law to decide the matters in question.
I am not suggesting that it is this particular definition of ‘reasonableness’ which shall apply here – I shall consider this below – however, what is common to all known uses of the word ‘reasonableness’ in the law is that it always refers to a spectrum. Perhaps this is heaping Pelion on Ossa but the point is crucial: The position is not binary – as in proper or improper, the position is ternary – as in (1) proper, (2) not proper but nonetheless reasonable and (3) outright unreasonable.
1.8 This loosening of the word ‘reasonable’ should apply just as much to the other reasonable (‘reasonably’) which I have already discussed.
1.9 Putting all these together, the position which emerges is: Take a hypothetical person. Ask, is this person, if not correct, is he or she at least reasonable in regarding the arrangements as proper or reasonable?
Second observation – Deciding reasonableness
1.10 I now turn to the more counter-intuitive point, which is to do with the manner in which the GAAR defines ‘reasonableness.’ The definition of ‘tax abusive arrangements’ provides:
(2) Tax arrangements are ‘abusive’ if they are arrangements the entering into or carrying out of which cannot reasonably be regarded as a reasonable course of action, having regard to all the circumstances including:
(a) the relevant tax provisions
(b) the substantive results of the arrangements, and
(c) any other arrangements of which the arrangements form part.
In other words, (a) the provisions, (b) the results and (c) the wider arrangements. I would like to make some observations pertaining to these sub-paragraphs.
I turn first to sub-paragraph (2)(b) – the question of results. I think that the inclusion of (2)(b) is helpful. First, as the taxpayer in each case should have some control over the result and the extent to which a tax advantage is to be gained, this provides one way in which the applicability of the GAAR can be precluded through restraint. (As to what degree of restraint is needed – case law will be required to provide guidance). But there is another point. The wording here refers to ‘the substantive result of the arrangement’. It does not refer to the ‘wider implications of the arrangement being successful’. So, the scope of the words is limited to the results of that particular implementation of the arrangement. This makes a difference – because it means that just because a particular implementation of an arrangement falls to be counteracted by GAAR, this does not mean that the arrangement itself can be said to fail. Rather, the arrangement fails on the basis of the results of that particular implementation – a question of fact in each case. If (2)(b) had not been included then the position would have been that a ruling on a particular provision would most likely have wrinkled out any perceived anomaly in the provision in an absolute way – thus forever more occluding any future implementation of the arrangement which was based on that provision.
As for sub-paragraph (2)(a), sub-paragraph (3) in the definition of ‘tax abusive arrangements’ provides:
(3) In subsection (2)(a) the reference to the relevant tax provisions includes
(a) any principles on which they are based (whether express or implied)
(b) their policy objectives, and
(c) any shortcomings in them that the arrangements are intended to
This sub-paragraph seems to require a purposive construction. However, the introduction of a purposive approach is not in itself an innovation. Consider the following citation from Barclays Mercantile (their Lordships are here quoting with approval from the decision in Arrowtown):
The ultimate question is whether the relevant statutory provisions, construed purposively, were intended to apply to the transaction, viewed realistically.
Could it be that the sub-paragraph (3)(a) immediately above purports to make an advance greater than that blessed by their Lordships in Barclays Mercantile? In other words, could it be that the legislator, through sub-paragraph (3)(a) is allowing the courts to look directly to Parliament for meaning – even though this meaning might not be supported by the actual words in the statute book? This does not appear to me to be the case. The parenthesis in (3)(a) seems to suggest that even though the courts can try to infer and take into account the principles on which provisions are based, these principles need to be express or at least capable of being implied. If meaning is to be ‘implied’ then it must be by reference to something and it would appear that it has to be implied from the words in the statute. The strange epiphany I seem to be having, therefore, is that through the inclusion of (2)(a) (the provisions) in the definition of ‘tax abusive arrangements’ the legislator seems to be doing little more than codifying the dicta of their Lordships in Barclays Mercantile.
Perhaps (3)(b) and (3)(c) go further? In the context of (3)(c) (and similar points arise in relation to (3)(b)), the question arises as to how the courts determine what a ‘shortcoming’ in a provision is. This must refer to the difference between the objective of a provision and its impact. The key point to make here is that when one considers the objective of a provision, one is restricted to only such objectives as can be gleaned from the wording of the statute – that this fundamental principle of interpretation is not being tinkered with in any way by the GAAR is, as has been discussed above, supported by (3)(a). In other words, it will not be possible for HMRC to argue that the objective of a provision is something which has only a very gossamer connection to the words in the statute and that, on that basis, there is somehow a shortcoming in the provision. Of course, there may be cases where the impact – immediate or remote – of a provision is removed from the objective of a provision (even as determined by adopting a purposive construction). In such cases, the battle will ultimately have to be fought and won on the question of what the objective of the provisions is – and the most I can say at present is that this cannot be too far removed from the fine print of the statute books. It cannot be argued by HMRC (and there is no suggestion in the GAAR) that there is some underlying objective, common to all provisions, which operates to counter a mitigation of tax. And in many cases, the policy objective will simply not be decipherable from the provisions. I cite from my book, Individual Tax Residence:
The Levene and Lysaght cases provide little practical guidance and the tribunals have relied heavily on the dicta of Lord Scarman from the Regina v Barnet LBC, Ex p Shah case. This was a non-tax case and the question of residence was necessary to determine whether the five applicants in that case were eligible for an educational grant. The decision is relevant to tax because it was held by Lord Scarman that one had to apply the natural meaning of ‘ordinary resident’ in that case…The argument propounded by the local authority that Parliament could not have intended so wide a definition so as to include foreign national was rejected. Whilst a policy-based interpretation of statute was acceptable, this was only where one could infer from the statute (or other acceptable material) a policy or purpose. In the present case, the relevant provisions did not hint at any such restriction as was being proposed by the local authority  2 AC 309 at 348.
For another instance where the courts have had difficulties in inferring the objective of a provision, consider Jerome v Kelly  STC 887 where the House of Lords found in favour of the taxpayer. Lord Hoffman stated:
 I accept that this conclusion leaves certain puzzles about what exactly s 27(1) does do in a case like this. It is tempting to say that it simply cannot apply to a case in which the person who enters into the contract is different (or deemed to be different) from the person who completes it…. But the ontological problem is more difficult and can probably be solved only by saying that the disposal must be taken to have happened when the company or trust which completed the contract first came into existence. That is, I would accept, a rather makeshift answer. But I see no elegant solution to the problem posed by s 27(1). Among the inelegant solutions, that offered by the Revenue is in my opinion the least acceptable. I would therefore allow the appeal.
There is another point to make about sub-paragraph (2)(a). Had the position been that this sub-paragraph had not been included, then the question as to whether or not there was a ‘tax abusive arrangement’ would have been determined mostly by the facts and, in particular, the results. The problem then would have been that whensoever there was a significant degree of tax mitigation, it would become arguable that the GAAR was in point. However, the inclusion of (2)(a) leaves it open to argue that the beneficial results, when considered in all the circumstances of the case and, in particular, the provisions, are in fact reasonable.
As for sub-paragraph (2)(c), which requires one to consider the wider arrangements, the inclusion of this sub-paragraph is helpful. Whereas the trend for self-cancelling transactions has abated, much of contemporary planning involves the adoption of additional transactions which require the taxpayer or other third parties to accept very real and often detrimental economic consequences. The requirement in (3)(c) that the wider arrangements be taken into account would appear to strengthen the case for a more reasonable approach being taken in such cases. Once again, the words in (3)(c) are reminiscent of the exhortation of the judge in Arrowtown that ‘facts must be viewed realistically’.
To end this discussion, the GAAR, apart from requiring reasonableness, also lays a sheen on what is ‘reasonable’ – but I do not think it really widens the scope as much as might first appear. This is supported by the expression ‘tax abusive arrangements.’ As was held in Dextra, even though expressions may be expressly defined by the legislator, due consideration must also be given to the words used in the expression itself. (This principle is not to be confused with the principle preventing the factoring in of titles of provisions).
Third observation – Intent
1.11 The removal of the no tax intent qualification is to be lamented. Graham Aaronson concedes in his letter that if there is a tax advantage then there is unlikely to be no intent.
1.12 However, I am not sure that this would be the case. The difference is primarily procedural. The Canadian GAAR does have a ‘no intent’ clause and this has been relied upon by the taxpayer in a recent case: McClarty Family Trust et al v. The Queen. The fact that the taxpayer fell within the ‘no tax intent’ exemption in that case meant that the judge in that case did not need to address the more mechanistic questions thrown up by their GAAR. Furthermore, the absence of a ‘tax intent’ has been relevant to many UK provisions – and most notably this was emphasized in the recent 2009 re-write of the Transaction in Securities legislation. Even though there was no substantive change (as far as the question of intent was concerned) in the course of the re-write, this aspect was given more prominence. The guidance on the re-write explains that the purpose underlying this was to provide a clear get-out for those taxpayers who did not have the requisite intent, thus freeing them from the rigmarole and administrative burden that came with having to consider whether they were caught by any of the other provisions.
Fourth observation – Burden of Proof
1.13 The burden of proof is on HMRC:
In proceedings before a court or tribunal in connection with the general anti- abuse rule, HMRC must show:
(a) that there are tax arrangements that are abusive, and
(b) that the counteraction of the tax advantages arising from the arrangements is just and reasonable.
1.14 In Canada, the burden of proof is split. Another jurisdiction in which the GAAR is in the throes of parturition at the time of my writing is India. India was considering placing the burden of proof on the taxpayer. (By way of comparison with India, India is indicating that there will be no treaty override in their GAAR – unlike in the UK).
Fifth observation – Counteraction
1.15 As for counteraction, if:
(a) there are tax arrangements that are abusive, and
(b) the procedural requirements of [the Schedule] have been complied
the tax advantages arising from the arrangements are to be counteracted on a just and reasonable basis.
Counteraction must be done on a just and reasonable basis. When the legislator uses the expression ‘just and reasonable’ he does not say that one has to have regard to the substantive tax result or the objectives of the provisions. He says simply ‘just and reasonable’. ”Reasonableness’ is not defined in the way it is in the definition of ‘tax abusive arrangement’ and the divorce is made further clear through the insertion of the word ‘just’. The point that emerges from this is that there is no co-relation between the tax advantage that emerges from the abusive tax arrangement and the counteraction. This in itself appears to me to represent a very reasonable and pragmatic stance on the part of the legislator.
On the other hand, one cause for concern is that, as ‘reasaonable’ is not defined here (and as I suggest above, the expression usually refers to a wide spectrum), HMRC would appear to have a wide discretion as to what counteraction be made. HMRC would simply have to demonstrate that the counteraction, whilst perhaps not proper, is still not objectionable. In other words, even though the burden is on them, the standard itself is low. (Compare and contrast with ‘tax abusive arrangements’, where the burden is also on HMRC but the standard is higher). If a taxpayer wishes to defend against counteraction, he would be charged with the uphill task of demonstrating that the counteraction was not only not proper but that it is downright objectionable. This brings with it uncertainty. But, on balance, perhaps this is still preferable to a position where counteraction involves an automatic atomisation of the tax advantage.
1.16 Could it be that the GAAR could result in a mitigation of tax?
Clearly, the statutory sub-paragraph cited above cannot result in a reduction of the tax charge? After all, it requires for the ‘tax advantage’ to be ‘counteracted’. However, the position is not as clear as one might first think and the position depends on what the starting position is. To the extent that the treatment under the avoided provision was not just and reasonable, it might be possible to continue to avoid it through falling within GAAR (and thereby securing a just and reasonable outcome), so that in the round the position is actually bettered. Indeed, it would appear to me, given that counteraction is mandatory where there is a tax abusive arrangement and the procedural conditions have been satisfied, that the taxpayer is within his right to require it of HMRC. Of course, the procedural conditions in (b) would have to have been satisfied before counteraction became mandatory, so HMRC would have to think carefully before invoking the GAAR. Once they have, the matter may well be outside their control – as the situation may well be hijacked by a shrewd barrister on the opposing side.
In addition, the draft provides:
An officer of Revenue and Customs must make, on a just and reasonable basis, such consequential adjustments in respect of any tax to which the general anti-abuse rule applies as are appropriate.
These consequential adjustments:
(a) may be made in respect of any period, and
(b) may affect any person (whether or not a party to the arrangements).
The consequential adjustments do not appear to require more tax to become payable – ‘adjustments’. They are mandatory. Since they are ‘consequential’, they can only be made where there is a counteraction. However, whilst sometimes the trade-off between counteraction and consequential amendment will simply be alleviatory (which is itself an improvement), in certain cases the consequential amendment may well eclipse the counteraction and prove to be beneficial when the matter is considered in the round.
The introduction of the GAAR is to be lamented as it brings with it uncertainty and an increased administrative burden (not only on the taxpayer but also on the courts and HMRC – as each case will need to be adjudged on its own facts). When section 75A FA 03 was introduced, there was much criticism – as the provision was mechanistic. This time the legislator has adopted a broader approach (necessarily so, given the width of the scope of the GAAR). Such an approach will inevitably bring with it uncertainty.
Another defect with the GAAR is the removal of the ‘no tax intent’ exemption. For the reasons given above, the legislator should re-consider his rejection of this clause and pressure must be brought to bear on the government by the powers that be.
Setting these points aside, however, if one works on the assumption that there is to be a GAAR, then the proposed draft, when one actually turns to consider it, appears to me to represent a balanced approach on the part of the legislator. To a great extent, it appears to do no more than codify the dicta of their Lordships in Barclays Mercantile. In doing so, the legislator is not acting in vain – as, as the tax avoidance cases evince, the courts have often changed stance as to which principles they employ in countering aggressive tax avoidance – some might suggest not without a certain degree of perplexing mercuriality. What the proposed codification of the Arrowton/Barclays Mercantile principle (if I may call it that) does is to ensure that the interpretative principles which are to guide the judges in their hermeneutics of the tax code are ossified for future reference, if only very broadly.
Another feature which I commend is that the GAAR, rather than providing for an automatic abnegation of the tax advantage that flows from an abusive tax arrangement, simply requires that the taxpayer be placed in a ‘just and reasonable’ position. As far as I know, this is that first time that the notions of justness and reasonableness are to be incorporated on such a broad scale to our regime of taxation here in the UK. And about time too. Despite the satirical title with which I prefaced this speech when I first wrote it (Unreasonably Reasonable and I still abide by the gentle criticism, implicit within it, of the tortuosity of certain parts of the draft), I have come round to the view that one cannot be any more reasonable than to require reasonableness and to proceed on the basis of it.
 Chief Constable of North Wales Police v Evans  3 All ER
 Barclays Mercantile v Mawson  UKHL 51 para 36 citing Ribeiro PJ in Collector of Stamp Revenue v Arrowtown Assets Ltd  HKCFA 46, para 35.
 See Consultation Paper:
Question 7 – The Government would welcome views on these commencement options, how transitional arrangements should be dealt with, and whether there should be different rules for different taxes where appropriate.