The ‘double reasonableness’ test in the draft GAAR was designed to set a high hurdle. It would do so by ensuring that arrangements would fall outside the GAAR not only if the judge himself regarded the entry into the arrangements as being a reasonable course of action, but also where, though he did not himself take that view, he nonetheless considered that such a view might reasonably be held. However, when the current wording of the test is examined against the backdrop of existing legislation and case law concerning the concept of reasonableness, it seems doubtful whether the test achieves its stated objective. The implication of this is that, unless the wording is strengthened, we may find ourselves drifting towards a more intrusive GAAR than what was originally proposed.
The draft general anti-abuse rule set out in HMRC's consultation document of 12 June 2012 captures ‘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’ (clause 2(2)).
The original test proposed by Mr Aaronson QC in his report of 11 November 2011 (the report) provided that an arrangement would not be caught by the GAAR ‘if it can reasonably be regarded as a reasonable exercise of choices of conduct afforded by the provisions of the [tax] Acts’.
As the consultation document notes, this test has come to be known as the ‘double reasonableness test’.
In his report, Mr Aaronson said that he aimed to give the taxpayer ‘the benefit of the doubt […] by placing on HMRC the burden of demonstrating that the arrangement can not reasonably be regarded as a reasonable exercise of choice’ (para 5.23) (emphasis in the original) and he considered that ‘[d]oing this should substantially reduce the scope for doubt as to whether an arrangement falls within the intended target area of the GAAR’ (para 5.24). He described the double reasonableness test as a pivotal safeguard.
The burden of proof (to the civil standard) is indeed on HMRC: clause 5. However, as the consultation document notes, the effect of this is likely to be limited to circumstances where HMRC’s and the taxpayer’s cases are evenly balanced. What seems more important is the proposition made at draft guidance note 32 in Appendix II of the report:
‘Applying this in the context of an appeal to the Tax Tribunal, it means that [this] safeguard […] would apply not only if the judge himself regards the arrangement as a reasonable exercise of choices of conduct but also, where he does not himself take that view, he nonetheless considers that such a view may reasonably be held.’
The report states that this is a high hurdle which the GAAR needs to clear and that this is intentional (para 6.4).
More recently, Chris Davidson, Head of the Anti-Avoidance Group at HMRC, commented that ‘We can have a debate about [whether or not something is reasonable], but [the double reasonableness test is] not asking whether something is or isn’t reasonable, but whether it can be regarded as reasonable’ (Tax Analysts Worldwide Tax Daily; 2012 WTD 133-1).
The question is whether, as currently drafted, clause 2(2) achieves the effect described by draft guidance note 32.
A double reasonableness test is not something that tax practitioners encounter in the ordinary course and Aaronson’s report does not give any examples of its use in any other area of UK law.
Tax practitioners are, however, familiar with the principle in Edwards v Bairstow (1955) 36 TC 207, that conclusions on questions of fact can only be set aside on appeal if the conclusion has been reached ‘upon a view of the facts which could not reasonably be entertained’ (1955) 36 TC 207 at 224, per Viscount Simonds) or (to put in a different way) if the case is ‘one in which the true and only reasonable conclusion contradicts the determination [of the lower court]’ ((1955) 36 TC 207 at 229, per Lord Radcliffe).
Recent civil cases draw a distinction between conclusions of primary fact and those which involve an assessment of a number of different factors which have to be weighed against each other (an evaluation of the facts).
In the latter case, the appellate court ‘ought not to interfere unless it is satisfied that the judge’s conclusion lay outside the bounds within which reasonable disagreement is possible’ (Todd's case [2002] 2 Lloyd’s Rep 293, at pp 319–320, para 129, approved in Datec Electronics Holdings Ltd and others v UPS Ltd [2007] UKHL 23, para 46).
Similarly, an administrative decision will be quashed as being manifestly unreasonable if it is ‘so outrageous in its defiance of logic or of accepted moral standards that no sensible person who had applied his mind to the question to be decided could have arrived at it’ (Council of Civil Service Unions v Minister for the Civil Service [1985] AC 374, at 410, per Lord Diplock)).
In these contexts, the court is not allowed to substitute its view for that of the body which is charged with exercising a discretion (ie, the fact-finding tribunal, minister etc).
This appears to be similar to what draft guidance note 32 is suggesting regarding the role of the Tribunal judge in the context of the GAAR.
Unfortunately, the wording of clause 2(2) is materially different from the judicial dicta above. Can it nevertheless be interpreted in the same or similar way?
Let us ignore the second instance of the word reasonable in clause 2(2), and focus instead on the words ‘cannot reasonably be regarded’.
An electronic search reveals a small number of uses of that formulation in UK legislation. For example:
It must be the case that something which fails a ‘cannot reasonably be regarded’ test would satisfy a ‘can reasonably be regarded’ test (this is illustrated by the ITEPA 2003 example referred to below).
Expanding our search for the latter formulation we find:
What all of these examples seem to involve is an evaluation of the facts (is it commercial/comparable/likely to give rise to a conflict/ready to be copied/representative of an investment of money at interest?) which is to be made on a reasonable basis.
It seems difficult to say that in the above examples the draftsman wished to signal to the tribunal/court that doubts as to the application of the relevant test should be resolved in favour of the taxpayer, HMRC or any other party.
As a policy matter, there is no obvious centre ground to be protected in those cases. Why then is clause 2(2) any different?
Clause 2(2) contains, of course, two instances of the concept of reasonableness – hence ‘double’ reasonableness. Perhaps this is what makes it special.
Double reasonableness is also a feature of the ‘reasonable apprehension of bias’ test which certain Commonwealth jurisdictions have adopted (albeit in a different context). The Constitutional Court of South Africa expressed it as follows:
‘The question is whether a reasonable, objective and informed person would on the correct facts reasonably apprehend that the judge has not or will not bring an impartial mind to bear on the adjudication of the case, that is a mind open to persuasion by the evidence and the submissions of counsel. The reasonableness of the apprehension must be assessed in the light of the oath of office taken by the judges to administer justice without fear or favour; and their ability to carry out that oath by reason of their training and experience. It must be assumed that they can disabuse their minds of any irrelevant personal beliefs or pre-dispositions. They must take into account the fact that they have a duty to sit in any case in which they are not obliged to recuse themselves. At the same time, it must never be forgotten that an impartial judge is a fundamental prerequisite for a fair trial and a judicial officer should not hesitate to recuse herself or himself if there are reasonable grounds on the part of a litigant for apprehending that the judicial officer, for whatever reasons, was not or will not be impartial’ (President of the Republic of South Africa & Others v South African Rugby Football Union & Others 1999 (7) BCLR (CC) 725 at 753, quoted in Locabail (UK) Ltd v Bayfield Properties Ltd & Anor [1999] EWCA Civ 3004).
It is clear from the above that the two instances of reasonableness operate independently of each other.
One has to assume that the litigant is a reasonable person and also that the litigant will take into account the fact that judges have taken an oath and that their training enables them to carry out their oath.
One could therefore conceptualise a reasonable person who fails to take into account those additional factors. Such failure does not mean that the person is unreasonable or that it lacks objectivity. It just means that its apprehension was not reasonable.
Clearly, any test that requires the satisfaction of two conditions or sub-tests will be more difficult to satisfy than one which involves just one condition.
But this doesn't tell us that much. Each condition or sub-test must be assessed to determine just how difficult the overall test is to satisfy.
In the case of the reasonable apprehension of bias test described above, the hurdle is high not because ‘reasonableness’ appears twice but because the reasonableness of the apprehension must be assessed in the light of the judicial oath and other factors which have the effect of weighing the scales against a finding of bias.
Turning now to clause 2(2), one would expect a Tribunal to begin its analysis of the GAAR by considering the meaning of the key concept of ‘reasonable course of action’. The Tribunal will quickly come to appreciate that reasonableness in this context must be assessed in the light of certain novel and progressive principles such as sensitivity towards the policy objectives of tax legislation and an aversion to the exploitation of gaps in the legislation (clauses 2(2)(a) and 2(3)). In the case of UK companies, these factors go beyond the ones mentioned in CA 2006 s 172 (duty to promote the success of the company). Therefore ‘reasonableness’ in clause 2(2) has a special meaning in the context of the GAAR. Without it, ‘course of action’ does not tell us anything. (Arguably, the original Aaronson formulation of ‘exercise of choices of conduct afforded by the provisions of the [tax] Acts’ did have some independent meaning.)
A Tribunal might soon realise that, in the context of the GAAR, ‘reasonable course of action’ really means ‘appropriate course of action’ and that ‘double reasonableness’ is therefore an illusion. If this is correct, then would taxpayers feel adequately protected by a test which captures ‘arrangements the entering into or carrying out of which cannot reasonably be regarded as an appropriate course of action, having regard to all the circumstances’?
A Tribunal which tackles the issue of what ‘reasonable/appropriate course of action’ means can be said to be performing a similar function to a Tribunal which tries to determine whether, in a case involving discounts, the excess represents a return on an investment of money at interest as required by CTA 2009 s 480(5) (referred to above).
In both cases the Tribunal would be dealing with tests that involve the exercise of discretion. If in the case of s 480(5) nobody believes that there is any centre ground that is protected by adding the words ‘can reasonably be regarded’ in front of the main test, why would anyone think that those words have that effect in the case of clause 2(2)?
Once it is accepted that the second instance of ‘reasonable’ in clause 2(2) really means ‘appropriate’, it becomes doubtful whether there is anything special about clause 2(2) in terms of the threshold for its application.
There is therefore a risk that a Tribunal will conclude that nothing in the GAAR legislation prohibits it from substituting its own view (of what is a reasonable/appropriate course of action) for that of the taxpayer even if the latter view is within the bounds within which reasonable disagreement is possible.
As others have noted, the question of the scope of clause 2(2) is too important to be left to HMRC guidance (though, as a minimum, the final guidance should include draft guidance note 32 from the report).
If clause 2(2) is to function as a key taxpayer safeguard as envisaged in the report then it needs to be recast using a formulation that has been proven to set a high hurdle. Looking at cases such as Edwards v Bairstow would be a good starting point.
