Nudging: Why, How, What Next?

ETHOS Issue 17 June 2017

Shifting Public Behaviour: Early Policy Measures in Singapore

Policymakers overseeing transport in Singapore were in a conundrum: Two decades after the Area Licensing Scheme (ALS) was first introduced in 1975, traffic congestion in the Central Business District, which had been reduced by over 70% in the initial years, had started to creep up again.

Priced as a fixed fee for the day or month, the ALS had a major drawback: the sunk cost effect. Individual drivers tended to continue and even increased the frequency of driving because they had already bought an ALS license. Car owners, who had come to accept the ALS fee as part of the cost of driving in Singapore, were driving more, rather than less. Was there a better way to make the costs of driving more salient?

In 1998, the Electronic Road Pricing (ERP) system was introduced. Drivers were charged on a pay-as-you-use basis. This reduced the sunk cost effect of the ALS. It also gave policymakers the flexibility to set congestion charges based on prevailing traffic conditions. In 2008, the cost of using a congested road was made even more salient with monitors on gantries displaying real-time ERP charges. Some joked that ERP stood for “Everyday Rob People”. Every time a driver passes under a gantry, they are reminded of the cost of road usage by an audible beep.

While increasing saliency worked in managing traffic congestion, it was not as applicable to policymakers in the Ministry of Health (MOH), who hoped to increase organ donorship in the 1980s. Many countries then operated under an “opt-in” policy, where someone willing to be an organ donor needed to give explicit consent to doing so. Public inertia resulted in there being far fewer organ donations than transplants needed.

Understanding the power of defaults, the MOH enacted the Human Organ Transplant Act based on an “opt-out” model in 1987. Under the Act, individuals are presumed to have given their consent to donate organs upon death unless they have opted out of the default. As a result, the number of deceased donor kidney transplantations increased from five per year before the Act¹ to 442 from 2004 to 2015.²

Electronic road pricing and organ donations are just two examples of the application of behavioural knowledge to public policy in Singapore. We have always adopted a pragmatic approach to policymaking. While it may not have been codified as behavioural insights, our public agencies routinely consider how individuals think and make decisions when designing and implementing policies.

Nudge: A New Name for Applying Behavioural Sciences to Policymaking

Richard Thaler and Cass Sunstein’s 2008 book, Nudge: Improving Decisions about Health, Wealth and Happiness, triggered a wave of global interest in applying behavioural insights to public policy issue. Drawing on the emerging disciplines of behavioural economics and social psychology, “Nudge Theory” explains why people often act in ways that are against their interests, which classical economics define as the maximisation of welfare. Faced with tight fiscal situations, policymakers worldwide were excited by the potential of nudges: small, low-cost changes that could make a big difference to the effectiveness of government interventions. As a result, “nudge units” sprung up in governments across the world.

In 2010, the UK under David Cameron’s administration set up a Nudge Unit (known formally as the Behavioural Insights Team) of fewer than 10 people, in the Cabinet Office.³ Starting off as a two-year experiment, it was tasked to inject a new and more realistic understanding of human behaviour across UK Government and to deliver at least a tenfold return on its cost. Today, the use of behavioural insights has moved out of the laboratory into mainstream UK government policy. The Behavioural Insights Team underwent mutualisation in 2014; now more than 100-strong, the outfit has offices overseas in Australia, the United States (US) and Singapore.⁴

In the US, former President Obama issued an executive order in September 2015 directing Federal Government agencies to apply behavioural science insights to their programs. This came after more than 30 pilots conducted by the Social and Behavioural Sciences Team (SBST) to improve programme outcomes in US agencies.

Today, there are 40 nudge units in more than 15 countries around the world, including Australia, Canada, Germany and the US.⁵Even the World Economic Forum has joined the bandwagon — recognising this as a future trend, a Global Futures Council on Behavioural Sciences was recently established to increase awareness of behavioural insights approaches across governments. Nudging is here to stay.

Why Behavioural Insights are Relevant to Public Policy in Singapore

In Singapore, heightened interest in applying behavioural insights (BI) to policymaking can be attributed to three factors.

First, Rising and More Divergent Expectations for Better Public Services

Citizens increasingly expect public services to be easy to use. They benchmark government services against their experiences with commercial firms. They expect public services to be personalised and to be delivered with empathy. They expect consistency and timeliness. Unfortunately, many government services do not meet the mark. Our communications are often laden with officious, bureaucratic jargon. Some of our government websites are ill designed and not user-friendly. There is often a poor appreciation of how people might think about and react to the way services are presented and delivered. The use of BI, which begins from understanding users’ needs and motivations, has the potential to help us improve public services and regain public trust and confidence in the government.

Second, a More Complex Operating Environment

Many ‘wicked problems’, such as urban density, transportation and demography, cannot be adequately addressed by traditional approaches to policy formulation and programme implementation. Take demography, for example. Legislation can be passed to raise the re-employment age, but is that enough to deal with an ageing population? Beyond Baby Bonuses, what more can be done about our declining fertility rate? Our businesses are reliant on immigration but some citizens are cautious, and others outright hostile, to the prospect. How can policymakers better tackle these problems? Our citizenry is getting more diverse. The 80/20 rule of designing policy to cater to the broad majority is constantly being challenged, because the “average” Singaporean exists only in the idealised realm of policymaking.

As society is not homogenous, we cannot assume that people will respond uniformly to an incentive, penalty or new law. In many instances, it is also not possible to foresee how a proposed policy or programme will interact with its intended users. For example, providing a monetary incentive to encourage social graciousness is not only unsustainable, it could also crowd out the intrinsic motivation of people to be thoughtful to each other. The application of BI, along with the use of randomised controlled trials (RCTs), can help tackle such complexities, by facilitating a more empirical and evidence-based approach to policymaking — without overgeneralising how we expect an “average Singaporean” to behave.

Third, Tightening Resource Constraints

Set against rising public expectations, a more complex operating environment and diverse populace, the Singapore Public Service is also faced with an increasingly tight fiscal and manpower situation. We are expected to increase our service footprint and manage an increasing number of public programmes.

Not a month goes by without new policy announcements and programmes that seek to improve societal outcomes. Using BI principles in service delivery design offers the potential to do more with less, and to make better use of limited resources for effective delivery. Simple changes, from tweaking a user-interface and providing information in a simpler format to tapping on defaults and social norms, could produce big payoffs in compliance or adoption at relatively low cost.

How It Got Started in Singapore

Unlike the UK, the application of behavioural sciences in Singapore public policy did not begin with a big bang driven from the centre. Instead, it was a ground-up movement, with various agencies exploring and experimenting with small-scale projects. The first teams started out as “skunk works”, learning the techniques while scrounging for willing partners prepared to give these new approaches a try. Quick wins were needed to gain confidence and win support from senior management.

A team from the Ministry of Manpower (MOM) and the Central Provident Fund Board (CPFB), for example, collaborated to increase the take-up rate of a new pre-retirement planning service. This was aimed at guiding those turning 55 years of age through their retirement options. As a pilot, CPFB sent invitation letters to 1,000 Singaporeans — 15% took up this free service. Believing that more Singaporeans could benefit from this service, the team simplified the letter and increased its relevance by providing personalised information. They also shifted the focus to pre-commit the recipients, getting them to think about “When should I go?” instead of “Should I go?” for the service. A four-arm RCT showed that the new letter doubled take-up rates to 32%. Given this initial success, the teams are now working to apply behavioural insights to help CPF members make better retirement decisions, and to encourage self-employed persons to make timely Medisave contributions.

Lessons Learnt

What can we learn from Singapore’s journey in applying BI in public policy? I suggest four insights:

• BI is one of many policymaking tools. The full potential of BI can only be realised when complemented with other policy tools such as design thinking, RCTs and data analytics. Sound data analytics can be used to examine the existing problem more closely and specify where root causes lie. Design thinking and qualitative research can complement established behavioural theories to design policy options, while RCTs and experiments test what works and what does not. Together, they provide the evidence that policymakers need to design user-centric policies and tackle complex challenges. Without such rigour, the discussion around behavioural interventions will remain at a philosophical level.
• BI is not a silver bullet. While nudges are useful, not every policy outcome is “nudge-able”. If cognitive biases are not one of the key factors holding back the desired behavioural change, then BI would not be applicable or would have very limited use. For example, in cases where negative externalities are generated (e.g. industrial pollution), what’s needed to address the problem is government intervention in the form of a “shove”. Nudging also falls short when it comes to preventing serious crimes such as violence or drug abuse. In such cases, legislation and active enforcement may be more appropriate.
• Specificity is key. There is a tendency to start discussions on a policy problem at a high level. However, applying BI to the problem means drilling down to a very specific issue to be tackled. While the intervention may seem to be a narrow, ‘downstream’ solution, the very nature of BI measures and the testing process requires such specificity. Each intervention then contributes to addressing part of the broader policy problem.

  Applying BI to the problem means drilling down to a very specific issue to be tackled.

  For example, policymakers may want to help vulnerable families to be more financially independent. However, several factors come into play, including education levels, employment and family structure. An assessment based on empirical and qualitative evidence could suggest sustaining employment as a possible area for intervention. Subsequently, BI interventions could be as specific as the design of job consultancy, the appointment process at career centres and follow up programmes to help employable members of the family stay in their jobs — these ‘last mile’ solutions ultimately make or break the multitude of programmes that seek to help vulnerable families.
• The devil is in the details. Knowing that a personalised SMS can be an effective nudge to prompt action is not enough. How it should be written, who should be seen sending it and when it is sent out are just as important. Working through the details of a behavioural intervention is an art, demanding acumen, broad consultation and an open mind. Having the discipline to test the hypothesis using RCTs is what distinguishes an evidence-based BI intervention from gut-based applications.

How to SEED the Use of BI in Your Organisation

Applying BI to policymaking has been a journey of testing, learning, and adapting. Agencies intent on building up BI capabilities in their organisations will do well to keep in mind four key ingredients to successfully SEED this approach.

Strategic Leadership:

Top leadership commitment to incorporate BI into the policymaking process, including at the more upstream stages, is a must. Without high-level support, applying BI could be reduced to ad-hoc efforts instead of being an integral part of policymaking. The support of senior management also signals the importance and impact of taking a more human-centred and evidence-based approach to policy design and implementation.

Experimentation:

Create a fail-safe environment, because not all interventions will be spot-on, no matter how well-designed or how much deliberation was put into the process. BI approaches need an environment that is tolerant of failure so that important lessons can be learnt. For example, the MOM ran an RCT to test messaging that reminded self-employed people to make mandatory Medisave (health insurance) contributions. Infographics were included in the letter as it had been effective in explaining difficult government policies in public communications. However, the result of the RCT showed that the letter with infographics significantly reduced contributions! The team hypothesised that using cartoons may have trivialised the subject matter, which was about encouraging compliance. Instead of being viewed as a ‘failure’, this result provided a valuable lesson, and helped the Ministry better identify what works in different contexts.⁶

Execution:

Execute BI interventions and testing swiftly by working on simpler problems. This will reap “quick wins”, which are critical to establishing credibility and getting the buy-in required to move to more complex challenges. For example, working on letters or simple prompts to improve response rates to a survey or a payment reminder can yield quick results, and create immediate impact on resource savings. In addition, unlike academic institutions that are interested in BI experiments with a view towards publication, it is important to keep in mind that as policymakers, we are not striving for novelty but workability. There is an increasing pool of international studies on similar issues and challenges. Avoid reinventing the wheel. Contextualise interventions for the operating environment, then move quickly to execution.

Diversity:

Assemble a diverse team to be BI “fire-starters” in your organisation. As BI draws from theories in various established disciplines such as economics, psychology, neuroscience and sociology, a diverse team is likely to work more effectively and innovatively. But it is important to look beyond competencies in these disciplines to also identify officers with policy and operations background within the organisation. They will provide the context and connections to identify the right challenges and sponsors for this ‘seed’ to germinate.

What Next?

Today, with over 250 members in a community of practice across 50 public agencies, the use of BI in Singapore public policy has evolved from an initial fascination of how cognitive biases challenge the traditional way of designing policies, to a more sophisticated framework of testing and accumulating insights on behavioural interventions. This shift has also shown that the use of BI is more than improving the last mile experience of citizens — it has the potential to fundamentally challenge the way we think about government policies and programmes.

There are three things we can do for BI to become even more useful and relevant in Singapore and elsewhere.

One is to find new and innovative ways to integrate BI with other disciplines and tools towards a more human-centred approach in public policymaking. For example, data analytics could be used to provide insights to customer segmentation, guiding more customised BI interventions. BI could work with emerging fields in the data sciences and artificial intelligence to tackle complex and cross-cutting issues. This will become increasingly important as citizenry becomes more diverse, policymaking more complex, and technology more pervasive.

Two is to apply BI in a whole-of-government approach to address ‘wicked problems’. In Singapore, and perhaps in other governments where BI efforts are emerging, there is a tendency for interventions to be applied by individual agencies. This has and will continue to work well for quick wins (i.e., addressing problems that are agency-centric and simpler in nature). However, we should also think about how BI can be applied to broader challenges which are more complex and cross-agency in nature. For example, we could look at retirement adequacy from the perspective of citizens’ life stages that correspond with major financial decisions that are also key points of interaction with government services (e.g., getting their first salary, getting married and purchasing a home). This means we could re-design our services not from an “inside-out” agency perspective, but an “outside-in” citizen’s perspective, in ways that nudge citizens towards decisions that improve their retirement adequacy. Taking a whole-of-government, citizen-centred approach means we must consistently apply BI across all government work, from how information is presented to how a regulation or incentive is structured. This means being more coordinated as a public service, with BI capabilities established across the public sector.

Three, exploring how we can also build and tap on BI capabilities outside government to develop and deploy more innovative solutions to problems. The Government has no monopoly on BI capabilities. For instance, the Singapore Management University set up a Behavioural Sciences Institute (BSI) in 2010, while the National University of Singapore has brought together a diverse pool of researchers under the Behavioural Insights Group (BIG). These are platforms we could potentially tap on to co-create policies and programmes with citizens, going beyond the current mode of engagement which is heavily reliant on feedback to government. Apart from more robust policy design, this process will also help to build deeper trust and create more resilient relationships between government and the people.

The use of BI, integrated with other tools and tapping on capabilities within and outside government, holds great promise to help us address many pressing policy issues, including those of increasing complexity.

So go forth and nudge for good!