HMRC comment on my tax gap estimates, via the Treasury Select Committee

Posted on

It's not very often HMRC and Parliament dedicate a chunk of a publication to my work, but the Treasury Select Committee have done so this morning, putting out the following, written by HMRC, as part of their response to that committee on the Tax Gap.

In view of the fact I will, inevitably, be writing responses to it over the next day or so I reproduce it in full here:

Annex: External calculations of the tax gap

There is a huge difference between the £35bn tax gap calculated by HMRC and the £120bn figure for the tax gap quoted by Tax Research UK. The scale of the difference has caused confusion and concern. Part of the reason for the disparity is that the two estimates are not directly comparable—partly because the definitions used are quite different. Nevertheless we think the £120bn figure is very misleading. It confuses the tax gap with cash flow and legitimate reliefs in a number of areas.

Table 1: the table attempts to show where our estimates differ. The main points of difference are described below.

Tax Research UK HMRC (2009-10)
Unpaid Tax £28bn Unpaid Tax £4bn
Tax avoidance £25bn Avoidance £5bn
Evasion £70bn Hidden economy, evasion, error, failure to take reasonable care, legal interpretation and criminal attacks £26bn
Total £123bn £35bn


The figure for unpaid tax of £28bn that Tax Research UK use is a snapshot of the total amounts owing to HMRC on a particular day. We think this gives a misleading impression of tax that is lost. Most tax paid late is collected within a few days, and over 90% is eventually collected. Therefore the figure we include in our tax gap estimate shows only the amounts we don't ever collect. 90% of this arises because of insolvency.


For corporation tax avoidance, Tax Research UK calculates the 'expectation gap'. They describe this as a measure of the difference between the contribution society expects business to make by way of tax paid, and what is actually paid. This is defined as:

the difference between the headline or declared tax rate for companies, and the rate of tax they actually pay.

This means that legitimate use of specific exemptions and reliefs such as capital allowances or double taxation relief, which reduce the amount of tax payable, are badged as avoidance. Avoidance is also drawn very widely. For example it includes tax not paid by non-domiciles who choose not to remit their income to the UK. We do not consider this part of the tax gap—the remittance basis is a choice intended by Parliament.


Tax Research UK particularly criticises HMRC's use of bottom-up methodologies to measure the direct tax gap and applies the VAT gap rate to arrive at an evasion figure for all direct taxes. This is highly inappropriate for three reasons:

  • the VAT gap includes all forms of non-compliance such as non-payment, avoidance and criminal attack as well as evasion. So the VAT gap arises from much more than just suppression of turnover that might feed through to evasion of direct taxes;
  • the use of the VAT gap in this way counts debt and avoidance twice for direct taxes—an arithmetical error, and
  • very importantly, tax gaps vary considerably by type of tax.

Tax gaps for taxes using deduction of tax at source, or with significant third party reporting requirements are much lower than for taxes without these features. This is established by very detailed research in the US and Denmark[3] and borne out by UK experience. Using the percentage VAT gap—9.7% for 2010-11 is the latest estimate—to estimate a tax gap for business profits of companies and sole traders may give an answer of the right order of magnitude. But it gives completely the wrong answer for the income tax due from employees where PAYE is operated. International research suggests a tax gap for this of around 1%. This incorrect assumption accounts for £30bn of the £120bn estimate.

Tax Research UK have supported their evasion estimate through comparison with an academic paper produced for the World Bank[4] which contains estimates of the size of the hidden economy for a number of countries including the UK. The estimate for the hidden economy in the UK is 13% of GDP which Tax Research UK then convert to a tax gap estimate of £73bn. Rather than support the Tax Research UK figure we believe that this comparison, if anything, further undermines it. The methodology uses a variant of a 'currency demand' model to estimate the size of the hidden economy. The use of 'currency demand' models for this purpose has been comprehensively and extensively criticised in unusually strong terms by other academics[5] [6] and national statistical bodies.[7] [8] The main theme of the criticism is that the methodology relies upon the application of assumptions which result in estimates that are much too large to be plausible. For example the Australian Bureau of Statistics explore what it would mean for Australia to have a hidden economy of 15% (as predicted in an application of this methodology by the same author). Critically they point out that a hidden economy of this overall size implies much higher levels of non-compliance in the areas of the economy where there is scope for underreporting. For example it implies underreporting of around 50% for every single self-employed taxpayer—which they reject as being implausible. Certainly non-compliance of the scale suggested for the UK is completely incompatible with all of our customer research and operational data.

As a result of the general concern about the use such models a body consisting of a number of international organisations including OECD, IMF, the World Bank, UN and the European Commission have issued a strongly worded statement advising against use.[9] Part of their statement says:

Unofficial estimates are often based on macroeconomic models. For instance, they may assume a fixed relation between the size of the economy and money in circulation. Such methods may yield grossly exaggerated results, attracting the attention of politicians and newspapers and thereby gaining wide publicity.

In a more recent report 'Reducing opportunities for tax non-compliance in the underground economy',[10] OECD comment:

the OECD (and other international organisations) reject these methods as being useful in obtaining exhaustive estimates of GDP or in estimating underground production and have observed that when applied they produce for most countries spectacularly high estimates of NOE [Non Observed Economy] activities which have no sound scientific base but which, nevertheless, attract much attention from the media and other parties.

Gross or net?

HMRC's published estimates are net of compliance yield. The Tax Research UK figures do not take into account direct tax compliance yield—for example £13.9bn for year 2010-11.


Tax gap measurement is not an exact science. But we are confident that £35bn is a much more realistic estimate of the tax gap than £120bn.


2   We periodically review this conclusion to check whether there are top down techniques for the direct taxes that we could use. Our latest review was published in August 2011.See Back

3   Unwilling or Unable to Cheat? Evidence from a Randomized Tax Audit Experiment in Denmark, Henrik J Kleven, Martin B. Knudsen, Claus T. Kriener, Soren Pederson , Emmanuel Saez;Tax Gap for Tax Year 2006, IRS( Back

4   Shadow Economies All over the World New Estimates for 162 Countries from 1999 to 2007; Friedrich Schneider, Andreas Buehn, Claudio E. Montenegro July 2010 Back

5   Estimating the Underground Economy MIMIC models-Trevor Breusch, November 2005 Back

6   The Shadow Economy in OECD Countries : Panel Data Evidence-Konstantin Kholodilin, Ulrich Thiessen, May 2011 Back

7   The Underground Economy and Australia's GDP-Australian Bureau of Statistics, March 2004 Back

8   Estimating the Underground Economy in Canada, 1992-2008-Statistics Canada June 2011 Back

9   Estimates of the unrecorded economy and national accounts, Declaration of the ISWGNA ( Back

10   Reducing opportunities for tax non-compliance in the underground economy, Forum on Tax Administration : SME Compliance sub-group ( Back