Translate page into
« Go back to list

Applying behavioural theory to collecting tax debts

Innovation image
An innovation provided by

Michael Hallsworth

Published On: 17 June 2014

If you are a public official, please sign in to see the contact details of the innovator.

Organisation: Cabinet Office Behavioural Insights Team

Country: United Kingdom

Level of government: Central government

Sector: General public services

Type: Communication, Methods

Launched in: 2011

Overall development time:

18 months

Public sector fraud, error and debt cost the UK government billions of pounds each year, costs that are borne by law-abiding taxpayers and firms. Using behavioural theory, HMRC (Her Majesty's Revenue and Customs) sought to understand the impact different contexts, messages and incentives had on debtors’ likelihood to pay overdue taxes.

A randomised control trial approach (RCT) was used to test the impact of variations in wording (such as more personalised language, or emphasising social norms) on hundreds of thousands of letters sent to debtors. Analysing the results of this RCT revealed which wording was more effective in persuading people to repay their debt.

The innovation advanced GBP 160 million of tax debts to the Exchequer over the six-week period of the trial.

Why the innovation was developed

Need to reduce the balance of tax debts owed to the UK by its businesses and private individuals.


Develop staff capacity, Improve effectiveness, Improve efficiency

  • Increase public sector revenues.
  • Reduce administrative costs.
  • Demonstrate the potential value of randomised controlled trials in government.
  • Increase voluntary tax compliance. 

Main beneficiaries

General population, Government bodies, Government staff

The UK population benefits from reduced tax non-payment, through reduced borrowing costs, greater fairness in tax paying, and more money available for public services.

Existing similar practices

UK Behavioural Insight pilotsIn public administration of my countryUK Government

The Cabinet Office Behavioural Insight Team has worked with a wider range of public services to test the opportunity for applying behavioural theory more widely, including in employment support and promoting charitable giving.


More revenue collected by letters, resulting in less manual processing and lower implementation costs. 


  • Multiple instances of payment rates being increased by 5 percentage points. Increase in payment rates of 10 percentage points at one point.
  • Messages bringing date of payment by 3 days within a 23 day test period.
  • GBP 210 million of revenue brought forward in the 2012-13 financial year.

Service quality


Improved clarity of letters.


User satisfaction

Other improvements

Results not available yet


Randomised controlled trial of different letter variants. Logistic regression on payment rates; ordinary least squares regression for improvements in day of payment.


Frontline staff in the tax agency, in collaboration with staff from the central coordinating department.

Design time: 2 months


The impact of the letters was assessed through randomised controlled trials.

A standard letter was created which included basic payment information. This acted as the ‘control’. A trial letter was then created, which was identical to the control – except for the inclusion of a short phrase that encouraged the recipient to pay. These phrases were based on behavioural insights (for example, a “social norm” message might point out that most other people have paid). Taxpayers were randomly allocated to get trial or control letters. We then analysed whether more people paid if they got a trial letter or a control letter. Since the allocation was random, we could infer that the higher payments were because of the specific phrases included.

Testing time: 2 months


Tools used:

The innovation used behavioural insight techniques to improve the efficacy of existing management information services and letter-issuing infrastructure.

Resources used:
  • The technology required had already been developed by HMRC (Her Majesty's Revenue and Customs).
  • Staff resources were required to determine how exactly to implement the trials and to develop the messages.
Implementation time: 2 months


The results have been published in government reports. Academic publication of results is forthcoming. The UK Government’s Behavioural Insight Team has been working with a range of public services to extend the use of behavioural theory into new areas.

Challenges and solutions

Initial difficulties around randomisation – to determine who received which variation of wording – were resolved through innovative use of taxpayer ID numbers. More specifically, ranges of taxpayer numbers (which are randomly generated) were allocated to receive certain letters.


Cabinet Office’s Behavioural Insights Team
Other Public Sector

The Cabinet Office’s Behavioural Insights Team was involved in the conceptualisation / development / testing and implementation phases.

Academics and Research Bodies

Academics were involved in the implementation phase.

Private sector

Informal consultation with the private sector at the conceptualisation phase.

Lessons Learned

  • Develop a way of receiving very quick feedback on the results of each trial (even if these results are not 100% accurate), so that adaptation and revisions can be made quickly.
  • Expert advice on randomisation and sample size is very important.
  • Continually liaise with the staff who are implementing the trials in practice.
  • Successful messages should become the new default approach, but the organisation should always keep trialling new options, to look for possible improvements.
  • The person leading the project needs to be personally persuasive.
  • Results from the trials need to be actively disseminated throughout the organisation, to show the outcome of the effort that has been put in.
  • Trials should not be too complex at the start, but can become more complex once those taking part become more skilled.

Conditions for success

  • Accurate management information.
  • Capacity for automated letter issue.

Other information

This innovation was virtually costless to implement.