Governments have introduced over a hundred anti-avoidance and operational measures since the OBR’s formation in 2010, and as we noted in our 2017 Fiscal risks report, we consider the prevalence and increasing reliance on the relatively uncertain revenue from these to be a potential source of fiscal risk when compared to the relatively certain costs of tax cuts.

Anti-avoidance and operational measures typically attract our highest uncertainty ratings as they target a subset of taxpayers who are already actively changing their behaviour to lower their tax liability. As a result, there is usually relatively high behavioural uncertainty, often referred to as attrition and reflecting the expected decline in the yield from a measure that results from businesses or individuals discovering new avoidance routes after the measure closes one that has been in use.

Similarly, since the measures are directed at uncollected tax, less reliable data are available to inform each costing, for example on the number of individuals or businesses affected and the value of tax under consideration. This is known as the tax base.

In this paper we consider the costings for 19 HMRC anti-avoidance and operational measures announced between 2012 and 2016, evaluating them within six relevant categories:

  • Offshore evasion: these are numerous measures, such as the UK Government’s agreements with Liechtenstein, Switzerland and the Crown Dependencies, where HMRC has entered into arrangements to bring in revenue from the overseas assets of UK taxpayers;
  • Accelerated payments: these notices require those involved in avoidance schemes within the disclosure of tax avoidance schemes (DOTAS) rules, whose arrangements have been counteracted under the general anti-abuse rule (GAAR), or who have received a follower notice, to pay any disputed tax upfront;
  • Employment intermediaries: two measures aimed at countering avoidance through the use of an employment intermediary, which is a business or individual that is placed between an employee and their place of work;
  • Partnerships: two measures to counter commonly used structures involving partnerships that gave rise to a tax advantage for the individuals involved;
  • Base erosion and profit shifting: is an OECD-led initiative that seeks to counter tax avoidance by multinational enterprises through governmental co-operation; and
  • Schemes of arrangement: this sought to tackle a stamp tax avoidance scheme involving the use of share cancellations by a targeted company in a takeover.

The chart below shows that in five of these categories the costings have fallen short of the original estimates – by amounts that vary between 15 and 65 per cent. The exception is the stamp duty ‘schemes of arrangement’ measure, which, in the first two years for which we have outturn data, has yielded significantly more than expected.

Since most of these costings were chosen for evaluation because we knew they had proved inaccurate, the results are not necessarily representative of all such measures. But we draw broader conclusions by considering these new evaluations alongside the results from more than 80 measures that we have now evaluated:

  • The number of measures that have yielded more than expected is broadly similar to the number that have raised less.
  • However, for those measures with the largest original expected yield, more have under- than over-performed.

The question for us is whether the revealed optimism bias in these larger costings is explained by something that we need to account for in future measures. Is there one or more aspect in these costings that consistently explains the shortfall? The answer appears to be ‘no’, with a variety of explanations specific to each costing. For example:

  • while it is highly likely the tax base for the Switzerland costing was overestimated, the tax base for the measures on employment intermediaries, partnerships and diverted profits tax were probably underestimated;
  •  in terms of attrition, while the measures on offshore evasion, employment intermediaries and partnerships seem likely to have underestimated the yield that would be lost, the reverse is probably true for ‘schemes of arrangement’ and previously evaluated measures like the annual tax on enveloped dwellings; and
  • it seems that both the accelerated payments and diverted profits tax measures have probably had a stronger deterrent effect than assumed in the original costings.

So while there are many lessons that can be learnt, there is no simple adjustment factor that could be applied to overcome the apparent bias. We will therefore need to continue to assess the assumptions underpinning each step of any new costing on a case-by-case basis.

For operational measures the main conclusion is that it has often taken longer than expected for measures to become fully effective. We have highlighted many of these delays in Annex A of our Economic and fiscal outlooks.

The inherent uncertainty around anti-avoidance and operational measures coupled with the increasing reliance that Governments have placed on them to meet the relatively certain cost of tax cuts and additional spending make this a continuing risk to our central forecast – one that we aim to ensure lies in both directions, in terms of overall yield and the timing of that yield across specific years.