A report by the UK Cabinet Office’s behavioural insights team published earlier this year, reveals that by applying behavioral insights, the government could reduce fraud, error and debt by billions. It claims that small changes to government processes, forms and language can have a big impact on behaviour and if applied nationally could save “hundreds of millions of pounds”.
The cabinet office team has close links with Professor Richard Thaler, co-author of Nudge: Improving Decisions About Health, Wealth, and Happiness, with the result that many of the ideas espoused in the book are outlined and implemented in the research paper below.
Seven simple steps
The report highlights seven simple steps based on evidence from behavioural science that show how “by going with the grain of how people behave, we can reduce the prevalence of fraud, error and debt”
- Make it easy: Make it as straightforward as possible for people to pay tax or debts, for example by pre-populating a form with information already held
- Highlight key messages: Draw people’s attention to important information or actions required of them, for example by highlighting them upfront in a letter
- Use personal language: Personalise language so that people understand why a message or process is relevant to them
- Prompt honesty at key moments: Ensure that people are prompted to be honest at key moments when filling in a form or answering questions
- Tell people what others are doing: Highlight the positive behaviour of others, for instance that ‘9 out of 10 people pay their tax on time’
- Reward desired behaviour: Actively incentivise or reward behaviour that saves time or money
- Highlight the risk and impact of dishonesty: Emphasise the impact of fraud or late payment on public services, as well as the risk of audit and the consequences for those caught
In order to explore the effectiveness of the insights above the team adopted a ‘test, learn, adapt’ approach. These case studies included randomised controlled trials (RCTs), which divided the study population into two or more groups and randomly assigning individuals to each of these groups. They then gave the intervention (e.g. a modified letter, a changed process, or a new text message) to one of these groups while continuing to treat the other group as business per usual. Through this they determined the difference in effectiveness of each of the interventions.
This document describes eight separate RCTs, each testing the effectiveness of one or more interventions – in behavioural science – which suggest may be effective in reducing fraud, error or debt.
- Using social norms: investigates whether informing people that the vast majority of those in their area have already paid their tax can significantly boost payment rates. Result: 15% increase from the old-style control letter which contained no social norm and the localised social norm letters.
- Highlighting key messages and norms: examines whether we can increase tax compliance among doctors by simplifying the principal messages and actions required, as well as using social levers and norms. Result: 2 new behavioural letters have already resulted in voluntary disclosures (from medics with outstanding tax liabilities) worth over £1 million (with the final total expected to be several million pounds).
- Using salient images: investigates whether using images captured by DVLA can help to reduce unnecessary repeat correspondence and encourage prompt payment of fines. Result: expected Spring 2012, but early results suggest that a simplified letter and image (of an offending untaxed vehicle) is outperforming the original and simplified letters.
- Better presentation of information: explores different ways of presenting information to discover which is most effective at encouraging the payment of debts. Result: Preliminary results show the collective and personalised appeals not to overlook the letter (stating that to do so would be treated as an active choice) are significantly more effective (10%) than other forms of letters.
- Personalising text messages: tests the impact of sending more personalised text messages on people’s propensity to pay court- ordered fines. Result: expected Spring 2012, but preliminary results suggest that texts with the recipient’s name appear to increase, by about 10%, the number of people making a payment.
- Prompting honesty: examines whether simplifying key messages, emphasising the consequences of fraud and getting people to sign forms upfront results in more honest declarations. Result: expected Spring 2012, but preliminary results indicate a new letter (comprising many of 7 insights above) resulting in 6% reduction in responses to renew Single Person Discount compared with original letter. This is likely to represent a reduction in fraudulent applications.
Varying the tone of letters: explores the effectiveness of different types of communication in encouraging plumbers to get their tax affairs up to date. Result: expected Spring 2012.
- Using beliefs about tax: tests the effectiveness of different messages – related to the fact that most people think that paying tax is the right thing to do – on payment of tax debts by companies. Result:preliminary findings suggest that what seems to be effective is to point out – in letters to companies – any gap between a taxpayer’s belief that ‘businesses should pay their taxes’ and the fact that their own company currently owes tax.
The results of many of the trials indicate the changes in behaviour that can result from highlighting key messages and actions required, presenting information more effectively, and exploring different types of communication:
The insights outlined in this document, applied in a range of different contexts and settings, show that not only is it possible to apply behavioural insights to reduce fraud, error and debt, but also that it can be done in a highly cost-effective way.
The report strikes a cautionary note, however, that not all techniques are feasible. For example, adding a post-it note or a handwritten name to an official communication seems to increase responses, but is not feasible, “given the scale of many government communications”. Thus the effectiveness of interventions will depend heavily on the context in which they are applied:
Policymakers should innovate, but should do so with humility about the limits of current knowledge, and with respect for what is acceptable and helpful to the public whom we serve.
This echos some of the sentiment from the UK House of Lords Science and Technology Sub-Committee’s report last year Behaviour Change in which they found:
…“nudges” used in isolation will often not be effective in changing the behaviour of the population. Instead, a whole range of measures – including some regulatory measures – will be needed to change behaviour in a way that will make a real difference to society’s biggest problems.
- Cabinet Office: Behavioural Insights Team publish paper on fraud, error and debt
- Cabinet Office: Behavioural Insights Team annual update
- House of Lords Science and Technology Sub-Committee’s report, Behaviour Change