The National Audit Office, the tax gap, HMRC and ‘other estimates’

Posted on

The National Audit Office issued its report on the annual report of HM Revenue & Customs on Friday. It’s a good job it did as the HMRC report was in many ways the closest thing to an exercise in pure puffery that has ever had the term annual report and accounts attributed to it. The data was also riddled with inconsistencies: table 5 suggested there were 58,911 full time equivalent employees in the year and yet Lin Homer persisted in claiming throughout the report there were 56,000 (page 3 for example). It's a shame the NAO did not pick up on the issue: if they had they would have needed to have explained hopw thousands of jobs must have been shed to wards the year end, but no mention is made of that.

What they did pick up on is the tax gap. On this they said (and I will be blogging on this issue more than once so the choice here is selective):

The tax gap is inherently difficult to estimate and HMRC acknowledges that no estimate of the tax gap can be definitive and that its estimates carry a degree of uncertainty.  

They add:

Around two-thirds of the tax gap is estimated using established methodologies, with the remainder estimated using developing and experimental methodologies.


HMRC is not able to estimate the scale of uncertainty around its best estimate of the tax gap. For around 60% of the value of the tax gap, it has not made any estimate of the scale of uncertainty.

In other words, they haven't got a clue whether it is anything near right, or not.

After which the NAO note:

Other commentators have published estimates of the tax gap in the UK which rely on macroeconomic data to make ‘top-down’ estimates. Without access to the range of data held by HMRC, it is necessary to infer relationships from aggregated data in order to produce such estimates. They also define the tax gap differently, such as by using a much broader definition of tax avoidance than that employed by HMRC. Such approaches typically lead to much higher estimates. HMRC argues that top-down approaches are not possible for direct taxes. Instead it uses the data it has from sources such as random enquiries, risk registers and data matching, to produce ‘bottom-up’ estimates for direct taxes.

There is, as far as I am aware, only one other commentator on this issue, and that is me. I did go and see the NAO about it earlier this year.

So, what the NAO is saying is that my approach is simply different.

Do they, then, have an opinion on who might be right? They note this recommendation from the IMF so I guess the endorse it:

The IMF recommended that HMRC improve its measurement and reporting of the tax gap, such as by combining ‘bottom-up’ and ‘top-down’ approaches for those taxes where this is possible, including direct taxes.

I have long said the same.

They also add:

It also recommended that HMRC improve its estimates of undetected non-compliance by replacing uplift factors derived from US evidence with those produced from UK data.

In other words, HMRC should try to use reliable data.

All of which being said I note they conclude:

No measure of the tax gap can be definitive, and there will always be uncertainties inherent in such a calculation.

I entirely agree, as I agree with their comment that I had to work with much more limited data than HMRC have available to them. But I can't help thinking that the comments on assessing scales of uncertainty and methodology suggest that the NAO think that HMRC could do a lot better on this issue, and that the place to start would be adopting my approach.