The government provides tax reliefs worth billions of pounds each year to encourage economic growth, but despite some significant improvements, HM Treasury (HMT) and HM Revenue and Customs (HMRC) still do not monitor or evaluate reliefs closely enough to understand if they cost too much or achieve their intended economic impacts, according to a new National Audit Office (NAO) report.

As at December 2023, the UK had 341 ‘non-structural’ tax reliefs intended to achieve social or economic objectives.1 HMT and HMRC have made a number of important improvements to how they administer, evaluate and report on tax reliefs since previous NAO reports in 2014 and 2020, including publishing objectives for non-structural reliefs in 2021, increasing the number of reliefs with published cost estimates and starting to publish high-level commentary on movements in the cost of significant reliefs.

In 2022-23, the total cost of the 104 non-structural tax reliefs for which HMRC has multi-year data was around £204 billion, equivalent to over 25% of all tax revenue in that year. The largest non-structural reliefs – making up around £129 billion of the 2022-23 cost – are not those targeted explicitly at economic growth. These include tax reliefs for pensions, capital gains tax relief on people’s main home and zero-rated VAT on food.

Government has not identified the number and cost of reliefs with purely economic objectives. There is no agreed number of reliefs targeted at economic growth, and they are neither monitored as a group nor compared for overall effectiveness. The NAO identified 39 reliefs with an objective aimed at increasing UK business investment, 29 of which were costed at £16.6 billion. However, this does not represent a complete list of reliefs with wider economic objectives.

Some tax reliefs with economic objectives have cost far more than forecast. For example, the Research and Development (R&D) relief for small and medium-sized enterprises (SMEs) cost around £15 billion more than HMRC expected between 2015-16 and 2020-21. The relief for High-end Television introduced in 2013-14 cost £762 million, £557 million (272%) more than HMRC expected it to cost during its first six years. There is no meaningful public reporting by government where tax reliefs greatly exceed the expected cost.

Tax reliefs do not have a statutory timetable for review. Responsibility rests on HMT and HMRC to undertake comprehensive, robust and proportionate evaluations of policy interventions in a timely manner. HMRC has increased the coverage of evaluations and produced an evaluation plan in 2021. HMRC has completed evaluations for 25 of the 341 reliefs since 2015. Together, these 25 reliefs were estimated to cost £22.4 billion in 2022-23.

The government has reduced some reliefs following evaluation, increasing tax revenue by billions of pounds in some cases. In the case of Entrepreneurs’ Relief, evaluation findings helped identify that the relief was costly, ineffective and not value for money. This led to a decision to reduce the threshold, saving the exchequer £1.6 billion in 2021-22 alone. An evaluation of the R&D SME scheme led to a policy decision to reduce it from April 2023, a move forecast to lead to a net reduction in costs of £4.5 billion between 2023-24 and 2027-28.

It has taken a number of years for government to amend reliefs where evaluations found that reliefs were of limited effectiveness. For Entrepreneurs’ Relief and Employment Allowance2 there was a three to four-year gap between the date of fieldwork and amendment of the relief. A range of factors may influence the speed of response, including the fact that tax reliefs can have a number of objectives and the timeframes for tax policymaking.

Some reliefs are also subject to large-scale abuse: HMRC estimates that the most likely level of error and fraud on the R&D SME relief was 24.4% (£1.04 billion) of expenditure in 2020-21.

HMRC does not know how far back large-scale error and fraud has occurred on the R&D SME relief, but the cost of the R&D SME relief increased by nearly 575% in cash terms between 2013-14 and 2021-22, when it cost £4.76 billion.

HMRC underestimated the level of risk and compliance work necessary for R&D SME relief for some years. The pattern of cost increases indicate that HMRC has underestimated the level of error and fraud, and compliance work on older claims may be worthwhile despite the costs and difficulties. HMRC needs to ensure that this mistake is not replicated in other reliefs.

HMRC does not routinely monitor how much compliance activity it is doing for other high-cost reliefs. However, for capital allowances involving Annual Investment Allowance claims, compliance work has reduced significantly since 2018-19, partly as a result of the response to the pandemic.

NAO recommendations include the need for HMT and HMRC to build on the improvements they have made so far in administering tax reliefs and make a step-change in how they assess whether tax reliefs achieve their economic objectives, are not too costly, and are not exploited to avoid or evade tax (see Notes to Editors for full list).

“Tax reliefs are an important policy tool for government, but their number and cost makes administration a significant task.

“The government should carry out sufficient evaluation and compliance work to understand whether claims are legitimate and reliefs are working, and act promptly to get reliefs back on track if problems are identified. There are now encouraging examples of changes where evaluations have found reliefs are not achieving their economic objectives.”

Gareth Davies, head of the NAO

Read the full report

Tax measures to encourage economic growth

Notes for editors

  1. There are two broad categories of tax reliefs: structural tax reliefs and non-structural tax reliefs. Structural tax reliefs are largely integral parts of the tax system. These reliefs have various purposes including defining the scope of taxes and making taxes more progressive (such as the personal tax allowance). Non-structural reliefs are reliefs designed to help or encourage particular types of individuals, activities or products in order to achieve economic or social objectives.
  2. HMRC published an evaluation of the relief in 2015 which concluded that the allowance was having a limited impact on businesses’ spending decisions. Fieldwork was conducted in 2014. At Budget 2018, the government announced that from April 2020 Employment Allowance would be restricted to employers with a total National Insurance Contributions (NICs) liability of less than £100,000 in the previous year. It forecast that this change would generate an additional £1.1 billion of revenue for the Exchequer between 2018-19 and 2023-24.


a) HMT and HMRC identify which non-structural reliefs have economic objectives and how these could be grouped together to provide oversight on how reliefs with similar objectives are working together to deliver government objectives, for example those aimed at business investment. This should include assessing the extent to which reliefs are achieving their objectives and whether the economic gains justify the scale of relief.

b) HMT should support ministers by ensuring that objectives for reliefs are expressed in as specific and measurable a way as possible. For example, including an expectation of acceptable costs and timescales, and a measurable definition of what success looks like in the advice provided.

c) HMT and HMRC should make a clear articulation of resources required to effectively administer and evaluate tax reliefs in tax information and impact notes. HMRC needs to ensure that teams can access the subject matter expertise needed for complex or technical tax reliefs.

d) HMT and HMRC should ensure the annual budget for evaluating tax reliefs is commensurate to the number and complexity of reliefs that need evaluating each year. The departments should ensure evaluations are completed in a timely way and commit to publishing evaluations within three months of completion, subject to ministerial approval.

e) HMRC should design proportionate controls and plan interventions at the beginning of a relief’s lifecycle based on risk assessment and learnings from other reliefs to:

  • establish appropriate up-front checks on generous reliefs
  • set out plans for evaluations to be carried out as soon as practicable and publish the timetable
  • plan for timely review of monitoring information to get interim evidence
  • ensure speed of response where reliefs are identified as not meeting desired objectives or do not function as planned

f) HMRC should improve the transparency of relief costs by committing to provide comparisons of forecast and outturn costs compiled on the same basis. This should also inform internal reporting and analysis:

  • HMRC should report where tax reliefs greatly exceed the initial expected costs
  • HMRC should investigate differences, using robust evidence to form conclusions

g) HMRC should demonstrate that it has applied lessons learned from the problems with the R&D SME relief to other reliefs that could be vulnerable to similar control failures.

h) Where the costs of tax reliefs rise rapidly, and beyond an economically credible scenario, HMRC should, where proportionate, investigate a sample of claims to check whether there is widespread non-compliance. It should put in place a rapid response capability where widespread non-compliance occurs.

i) HMRC should monitor agent activity to assess the risks for tax reliefs and intervene early where it sees that expected standards of behaviour are not being met.

j) HMRC should publicise the areas where it would welcome academic research on tax reliefs.

Latest press releases