1 HMRC's measurement of tax compliance |
1. On the basis of a report from the Comptroller
and Auditor General on HM Revenue & Customs' (HMRC's) 2013-14
Annual Report and Accounts, we took evidence from HMRC on the
effectiveness of its compliance work to increase tax revenue and
tackle tax avoidance.
2. HMRC brought in £506 billion tax revenues
in 2013-14, 7% more than 2012-13. The taxes that contributed most
to this increase were income tax and national insurance which
increased by £16.2 billion (6.4%), VAT which increased by
£7.2 billion (7.1%), and stamp taxes which increased by £3.4
3. Since the 2010 spending review, HMRC has been
funded to do more compliance work to secure additional revenues.
Its 2010 spending review settlement with HM Treasury (the Treasury)
included a £917 million reinvestment in compliance work to
generate additional revenues of £7 billion a year by 2014-15.
HMRC agreed to measure the impact of its compliance work by estimating
'compliance yield'the additional revenue it generates through
its activities to deter and prevent tax losses, such as from avoidance,
evasion and criminal attack.
4. HMRC estimated that its compliance work in 2013-14
generated additional revenue of £23.9 billionthe highest
level ever reported. Of this sum, £9.2 billion is additional
cash that HMRC expects to collect as a result of settlements reached.
The remainder is an estimate of losses prevented (£8.0 billion),
an estimate of future revenue benefits over the next 5 years as
a result of changing taxpayers' behaviour (£5.5 billion),
and an estimate of the impact of changes to tax law or HMRC's
processes which reduce opportunities to avoid or evade tax (£1.2
billion). HMRC has reported exceeding the targets for compliance
yield it agreed with the Treasury each year during this spending
5. The C&AG's report identified that HMRC made
a mistake when calculating the baseline against which to measure
increases in compliance yield over the 2010 spending review period.
The error meant that HMRC had set its baseline £1.9 billion
too low, which then made the targets HMRC agreed with the Treasury
easier to achieve. HMRC
apologised to us for the mistake. It admitted its baseline estimates
were wrong and that the error had not been spotted; and told us
that until the error was identified in the course of the NAO's
audit, it had believed it was over-delivering against its targets.
During this time, HMRC's briefing to Ministers and information
it presented to Parliament about its compliance performance, such
as in its Annual Reports or its Fast Facts publication
in May 2014, had inadvertently overstated the improvement in its
6. Taking the error into account, HMRC did not exceed
its anticipated level of performance in 2011-12 or 2012-13 as
it had previously reported. For example, it reported in 2011-12
that it had exceeded its performance targets by £1.9 billion
when in fact it had only just achieved the level of performance
anticipated. HMRC told
us that it had still performed well, achieving 44% more yield
in 2013-14 than its baseline at the start of the spending review
period, even allowing for the error. It reiterated that it did
not lose its focus and that HMRC staff did not stop when they
thought they had achieved the target's requirements, but had "strived
to go beyond what was asked of them".
HMRC consider that staff bonus payments made during the period
before the error was uncovered were deserved and do not need to
be reviewed as staff had continued to deliver improved year-on-year
7. We queried why HMRC's internal processes had not
uncovered such a large error in a key performance measure for
three years. HMRC
explained it missed the error because it had not revisited the
baseline calculation and had instead focused on measuring its
performance each year. It
acknowledged it needs to take more care when producing baselines
in future, but insisted that no improvements to its internal processes
were needed as the error did not show anything wrong with its
management information systems.
HMRC told us it would expose its performance data to greater critical
challenge in future, and has invited the NAO to provide external
scrutiny of its compliance performance data.
8. HMRC's measure of compliance yield is complex.
It includes four different types of yield to cover different types
of impact over different time periods. For example, the measure
includes cash HMRC has collected or expects to collect as a result
of tax settlements in 2013-14, losses it has prevented, and revenue
benefits it projects in future years. HMRC told us that while
not all compliance yield is "literally in the bank at the
end of the year", it believes that including a broad range
of impact from its preventative work is important.
9. HMRC told us it must use estimates to calculate
the value of some types of compliance yield. These include all
of the future revenue benefits it predicts (£5.5 billion
of the £23.9 billion yield reported in 2013-14) and some
elements of revenue losses it has prevented (a proportion of the
£8.0 billion of revenue loss prevented it claimed). HMRC
explained that it makes estimates based on available evidence
when more certain or accurate calculations are not possible. We
note that an increasing proportion of HMRC's compliance yield
is calculated using estimates and that these types of calculation
are more subjective, more susceptible to errors and less certain.
10. We think it is reasonable that HMRC should base
some of the figures making up total compliance yield on estimates.
But it has not been clear enough about the nature of the estimates
or the inherent uncertainties when reporting its performance publicly.
It improved its reporting in this year's Annual Report, reporting
its performance in each of the four yield types separately for
the first time. However,
it has still not done enough to explain that it has not collected
all the £23.9 billion compliance yield it reported in 2013-14,
or explain the level of uncertainty attached to some of its estimates.
Performance data has therefore been open to misinterpretation:
for example, the media reported that HMRC's compliance work brought
in £23.9 billion extra cash this year, when the real picture
is much more complex. 
11. HMRC has changed how it measures compliance yield
significantly since 2010-11. It made changes both because it had
new evidence about its estimates and to reflect a wider range
of the impacts of its compliance work.
From 2011-12 HMRC has included yield from disrupting organised
crime, and significantly expanded the measurement of "future
revenue benefits"a measure of how HMRC expects companies'
and individuals' behaviour to change following a compliance investigation.
These changes have increased the amount of yield HMRC reports:
for example, the effect of broadening the range of interventions
for which future revenue benefits are estimated was to increase
HMRC's yield in 2011-12 by £1.9 billion.
12. HMRC now acknowledges that its compliance yield
data are not comparable over the longer term because of the changes
it made to measurement in 2011-12. The baseline error showed that
the changes HMRC made to compliance yield in 2011-12 were more
significant than previously thought and that it should not have
made longer term comparisons with older data. It had therefore
inadvertently presented misleading information that compared "apples
For example, HMRC's Fast Facts publication, published in
May 2014 before the error was uncovered, showed compliance yield
data from 2005-06 to 2013-14 on a single graph with no explanation
that the measurement methodology had changed. In July 2014, HMRC
reissued this document with new data which took account of the
13. We accept that HMRC has improved how it publicly
reports improvements in compliance yield. But when it reissued
Fast Facts in July 2014 it again presented in one graph compliance
yield data from 2005-06 to 2013-14 that was not comparable. HMRC
caveated the graph to explain that data were not comparable over
time, but we consider any presentation that shows compliance data
before and after 2011-12 together to be misleading.
14. HMRC explained that it aims for its public reporting
about compliance yield to provide both simple and accessible information
despite the inherent complexity of the underlying data. However,
it accepted that prior to the NAO's audit it had not been sufficiently
clear about the differences between compliance data reported before
and after 2011-12, and that caveats it had provided when it published
this data were insufficient. It recognises the need
to be clear about the impact of making any changes in methodology
to take account of changes in the nature of compliance work or
better evidence of its impact.
1 C&AG's Report, HMRC's Annual Report and Accounts
2013-14, HC 19 Session 2013-14, 3 July 2014 Back
C&AG's Report, paragraph 1.2 Back
C&AG's Report, paragraph 11 Back
C&AG's Report, paragraphs 2.5 to 2.7 Back
C&AG's Report, paragraphs 13, 15 Back
C&AG's Report, paragraphs 2.38 Back
C&AG's Report, paragraphs 2.9 Back
Qq 1-2 Back
Q 1, C&AG's Report, paragraph 2.10 Back
Qq 2, 18 Back
Q 7 Back
Qq 6, 7 Back
Qq 1, 7 Back
Qq 13-16 Back
Qq 18, 20 Back
C&AGs Report paragraph 2.41 Back
Qq 17-19 Back
C&AG's Report, paragraph 14 Back
C&AG's Report, paragraphs 2.13 and 2.36 Back
Qq 11, 14; C&AG's Report paragraph 2.38 Back
Qq 8, 9 Back
Q 10. The revised caveat added in July 2014 is: "Our methodology
for measuring compliance yield is updated to reflect the latest
evidence of the impact of our compliance work - targets are adjusted
in line with this. Previous outturns are therefore not directly
comparable with figures from 2011-12 onwards, where we have been
using an improved methodology - see HMRC Annual Report for full
Qq 14, 16 Back