Optimal Audit Portfolio Design for a Tax Authority

Tax evasion is an important economic and social phenomenon. The most recent HMRC estimate of the tax gap places it in the region of £40 billion, which is about 8% of total tax liability. Good tax administration is essential for reducing this gap, but in some important respects the existing academic literature is not particularly useful in this context. It is my objective to construct a scientifically sound analysis that will be validated, at least in part, by the publication of high quality papers in good journals. This research will be timely (£900 million of extra resource having been given to HMRC to tackle non compliance in the 2010 Spending Review) and will contribute to government objectives.

As a part of their enforcement operations, tax authorities implement a portfolio of taxpayer audits. The British tax authority, HMRC, is currently considering introducing a new category of 'light-touch' audit to complement the more rigorous audit types it has traditionally implemented.

Tax authorities also implement targeted campaigns that entail an initial acquisition of third-party information by the tax authority, followed by written communication with those taxpayers who might potentially wish to make to make a disclosure in the light of the information acquired. Taxpayers are offered an incentive to make a disclosure under the campaign in the form of a lower penalty rate on the tax owed. Recent years have seen a number of such campaigns, including three aimed at taxpayers with offshore accounts, and campaigns aimed at particular professions.

However, the academic literature is yet to properly understand how a portfolio of audit types should be constructed, and the theoretical underpinnings behind the optimal design of campaigns. In the case of audits, this principally stems from a simplifying assumption in the standard model of tax evasion that, when the tax authority performs an audit, all non-compliance is automatically detected. In practice, there is generally a negative relationship between the effectiveness of an audit at detecting non-compliance and the cost of performing it to the tax authority.

The proposed research will identify the most effective way for a tax authority to design a portfolio of interventions so as to enhance compliance. To achieve this, a mixture of insights from game theory and behavioural economics will be applied. Behavioural economics will be used to understand how individual taxpayers will respond in their compliance behaviour as a result of an intervention by the tax authority. Game theory will then be applied to these insights to derive the optimal portfolio design.

The analysis will centre on the strategic interaction between taxpayers and the tax authority - how taxpayers will respond to a given intervention, and, given that, how the tax authority should optimally design a portfolio of intervention types so as to maximise expected tax revenue. It will accommodate taxpayers who are heterogeneous in income, risk preferences, and attitudes towards compliance.

The components outlined will provide a more compelling characterisation of optimal enforcement policy with respect to a portfolio of audits and campaigns. It will provide a rich framework in which to pursue the question of how best a tax administration can design its compliance operations. The wider insights into optimal auditing can be applied to a wider set of enforcement domains than tax compliance. Powerful agent-based computer simulation methods will be utilised to explore the predictions of more complex versions of the analysis involving social networks of taxpayers, which cannot be analysed using conventional methods. The agent-based techniques I will develop can, in future, be applied to a large range of questions in the field of behavioural economics. Ultimately, the research will produce high quality academic papers, and simulation programs that can be used by tax authorities to improve tax administration.

Tax evasion is a worldwide phenomenon, and a significant issue in most developed economies, including the UK. The design of an effective policy to reduce the tax gap requires an understanding of the factors that affect compliance and the sophisticated design of enforcement policies. The proposed research advances knowledge in this area by contributing to the economic theory of tax compliance and thereby enhancing the quality of tax administration.

The economic literature has developed a range of equilibrium concepts for understanding the strategic interactions between actors. To date, however, some unrealistic assumptions of the standard model of tax evasion have prevented these techniques from being applied to the question of how a portfolio of audits and campaigns should be designed. Moreover, behavioural economics is providing a range of new insights to supplement traditional game theoretic approaches. The proposed research will provide a systematic study how taxpayers are expected to respond to a given intervention and, given that, how a tax authority should optimally design a portfolio of audits and campaigns. The research will explore these issues under the standard economic assumption of expected utility analysis as a benchmark, but will then seek to understand the implications of alternative approaches from behavioural economics.

The project will be beneficial to a range of academics. At its broadest, the beneficiaries will be the whole community of scholars who will benefit from our improved understanding of human behaviour. Also, in future, the agent-based techniques that form an element of this project will be used to study a wide range of problems in behavioural economics problems, including saving and pension decisions. Within the set of scholars interested in human behaviour, the project will be of especial benefit to current and future researchers working on the economics and psychology of tax compliance, who are identified in more detail in the section on Academic Beneficiaries. A central element of the proposed research is to allow for the insights of behavioural economics in the compliance decision. Therefore, the research will also benefit researchers working in the field of applied behavioural economics more generally, by illustrating the predictions of this approach within a new context.

Outside of academia the research will be used by tax authorities internationally, and especially in the UK. For instance, HMRC are currently appraising an option to introduce a new 'light-touch' audit type. They are also seeking expert academic input to optimise the design of a relatively new area of enforcement: targeted campaigns. Having worked closely with HMRC in designing this proposal, it has been carefully tailored to address both of these current issues. As tax authorities are increasing benchmarking against each other internationally, other international tax authorities are carefully monitoring these issues in the UK.

The computer simulation element of the proposal will produce programs available for users in tax authorities that will enable them to analyse a rich array of different policy scenarios. They will be able to utilise this tool to shed new light on questions relating to the strategic design of enforcement measures, and thereby to issues of internal resource allocation. New insight will be gained into current questions, such as to whether to introduce a new 'light-touch' audit type, and how best to design campaigns.

Beyond the life of the project, utilising strong links with the Government Economic Service from my time as a civil servant, the research will benefit a wider group of users responsible for enforcement. These include the TV Licensing Authority, the Driver and Vehicle Licensing Agency, and the Department for Work and Pensions. New insights into the operations of these departments will help to develop models of internal resourcing and improve returns to enforcement activities.