It is a truth much delivered to clients – by me and by others at the tax bar – that their prospects of succeeding in tax litigation are influenced by what are politely styled “environmental factors”. One of these, if you happen to be litigating an avoidance transaction, is that tax avoidance is much in the public eye. (And not in a way that is helpful to those badged ‘avoiders’.) Another is that pressures on the public finances are everywhere to be seen such that shoulders are to be put to wheels.

It sometimes comes as a shock to the public that judges are human and their legal assessment of a case is not wholly divorced from their feel for it. That’s neither wholly a good thing nor wholly bad. How strongly you feel it’s a good thing that judges overlay a sense of ‘justice’ onto their analysis of the ‘law’ probably depends on the confidence you have in the judges’ sense of justice. It’s for this reason that our appalling – there’s no putting lipstick on this pig – judicial diversity statistics ought to be a matter of much greater public concern. My own view – that the majority of judges are intellectually honest, decent and driven by entirely appropriate notions of liberality and fairness – is bound to reflect that they see through similar eyes to mine. In tax avoidance cases, on the other hand, many might consider that judges should not be slow to plug their thumb into a dyke.

But it’s also true that the overlay of fairness should not jeopardise legal certainty. The validity of the law is a function of its predictably. And the practical applications of this aphorism can be seen even in a tax context: it is a truth widely held that unpredictable tax consequences in the UK jeopardise the inclination of foreign and UK investors to invest.

It was these factors that led me to welcome the introduction of the General Anti Abuse Rule. It gave judges hearing avoidance cases – who I had seen as occasionally articulating their instinct to fairness against the grain of the legislation – a legal framework within which to articulate it. And in so doing it made the operation of the law more certain: how the GAAR will operate is more predictable than the question what will be an unknown judge’s instinctive response to a set of facts.

[Lengthy preamble ends.]

Yesterday morning, I read yet another (to me) surprising decision of the Upper Tribunal in HMRC’s favour. And on a whim I spent some time counting decisions of the Upper Tribunal in substantive (i.e. not case management) statutory appeals from the Tax Chamber of the First-tier Tribunal.

Depending on which ones you count (and I tried to be consistent), there have, in 2014 so far, been 38 such cases. HMRC have won 31 (a win rate of 82%) and the taxpayer has won 7 (a win rate of 18%). On the face of it, that’s rather odd. It’s worth remembering that appeals to the Upper Tribunal have all passed a ‘permission’ threshold to strip out hopeless cases.

I then examined an obvious candidate explanation. The Appellant typically has the more difficult job: perhaps it is that the taxpayer is more often the Appellant? But if you look at the win ratio of Appellants, the taxpayer fought 23 and won 2 (i.e. a win rate of 9%). HMRC fought 15 and won 9 (i.e. a win rate of 60%). So my candidate explanation is helpful but not sufficient.

Another candidate is that HMRC is just much better at picking which cases to appeal. Certainly HMRC have a published strategy indicating that they should generally only pursue litigation where they consider it has a better than 50/50 prospect of succeeding. All things being equal, the taxpayer, on the other hand, will typically ask the question whether the costs of pursuing the litigation are higher or lower than its expected gain. If it costs 1 to pursue a 10% chance of gaining 100, the (deep-pocketed) taxpayer will generally litigate. On the other hand, it’s worth bearing in mind that ‘environmental factors’ are presently so compelling that sometimes even taxpayers whose behaviour is perfectly proper and who have strong prospects of success choose not to challenge poor HMRC assessments of liability because of the commercial risk that their behaviour will be spun as tax-bad to sell newspapers.

For what it’s worth, my instinct is that this candidate takes us closer to – but not all the way – home.

Of course, my data set must be approached with caution: it is small, and is chosen at random (2014 until today). And there are bound to be some counting discrepancies. But to my mind, the numbers are so stark that this is an area meriting further examination. Will the numbers look different once GAAR cases start coming before the Upper Tribunal? What would a longitudinal study of this data setreveal about the final decision post any further appeals? And are there any other explanations I have missed?