Nudging policymakers – an overview of Behavioural Exchange 2014

How often have you intended to do one thing, but then ended up doing something less satisfactory? Wouldn’t it be good if you had help to close the gap between your good intent and your actions? How would you feel if it was your government that helped close that gap?

On 2-3 June there was the first international behavioural insights conference held in Sydney, Behavioural Exchange 2014, jointly held by the NSW Government and the United State Studies Centre. It looked at the practice of ‘nudging’ or attempting to help individuals make decisions that better suit them, and that help society more widely.

I was lucky enough to attend this very stimulating event and hear the many great speakers provide a rich vein of material to consider. The following is my attempt to share an overview of the conference and some of the key points and themes (so please blame me for any errors or misattributions). At the end of the post there’s also links to some of the other posts and articles relating to the conference, including a second post from me reflecting on some of the things discussed.

What are nudges?


Professor Cass Sunstein talked about them as being any feature of the social environment which affects what people do, but is not coercive or involving economic incentive.

They include:

  • warnings

  • reminders

  • default rules

  • drawing attention to social norms

  • the order in which things are presented

  • colour and noises.


Professor Sunstein described them as being similar to GPS – i.e. they are something that is helpful, but that can be ignored and that can be directed by the user.

Nudges do not include things such as jail sentences, fines, subsidies, or tax incentives.

David Halpern used the story of an old fish asking two young fish swimming by ‘how is the water?’ with the young fish responding ‘what’s water?’ to make the point that nudges are already all around us, we just may not (always) notice them. Nudges are an existing feature of the environment in which we live and operate in – nudging is inevitable, and is already happening in the rest of the economy/society.

Nudges are not the only tools available to policy makers, and should not be seen as the answer to all problems. They do have advantages over other approaches though (such as mandates or bans) which can tend to squash heterogeneity and diversity when it may not be needed.

Examples of nudges from the conference were numerous. Some of these included:

Why nudge?


To my mind, Professor David Laibson gave the most concise reason for why nudging is useful to the public sector – because behavioural economics explains why good intentions don’t match actual actions.

He noted that the traditional view was that people understood the world and could act in a way to translate their desires into action. However, the Professor emphasises that multiple studies have shown that is not necessarily the case.

For instance, he gave the example of savings behaviour. One study found that even when people were educated as to the increased benefits that they could receive if they changed their behaviour (in a real world example of accessing employer retirement contributions), very few of the people in the study did. This contrasts with a classical view that incentives and education are effective.

A behavioural view instead recognises that often these do not work (or are not sufficient). Economic incentives matter, but so do psychological factors – such as that people have imperfect self-regulation or have social preferences.

Professor Laibson also noted that when choosing for the moment we tend to pick the immediate benefit; however when we are choosing for the future, we choose the better outcome (except when the future comes round and it becomes the now). So we may intend to exercise tomorrow, but today we might put it off.

Some of the other reasons for why the public sector might choose to nudge included:

  • Because self-defeating behaviour can be changed using inexpensive, scalable nudges which preserve choice while changing behaviour

  • Nudging helps make things actionable – it makes it easier for people to act out their good intentions (rather than being about changing their intentions, goals or aims)

  • Behavioural insights can help build better and more rigorous testing and learning into the policy making process

  • Behavioural insights encourages making small changes and doing things in small test groups first

  • That it reflects that people aren’t rational – people make mistakes, and people make mistakes in persistent and predictable ways (and there was discussion of how ‘rational’ isn’t a helpful word, because saying that something is not rational implies that it is wrong, rather than just how things are)

  • The public sector often needs to try things, and behavioural insights provides some of the tools we need to test them in a scientifically critical way

  • Because nudging is proven, legitimate and credible.


How do you nudge?


There was discussion at the conference about applying nudge. While this was not exhaustive, nor intended as a recipe, some key bits of guidance emerged.

The principal advice was the value of randomised controlled trials (RCTs) as the premier form of evidence. RCTs allow much better testing of what works by being able to compare different treatments against a control group who continues to receive the status quo treatment.

In terms of more general guidance, David Halpern outlined that the nudging process for the Behavioural Insights Unit involves:

  1. Define the outcome

  2. Understand the context (using a lot of ethnography and filed research to really understand the issue)

  3. Build your intervention

  4. Test, learn and adapt.


A common piece of advice was that policymakers need to recognise that we don’t know the answer, and that we should let the evidence help us work it out.

Some of the techniques used in delivering nudges include:

  • Behavioural segmentation – recognise that different messages will be more effective for different groups

  • Shared humanity – recognise that there is power in reciprocity and personalisation in government messages, rather than impersonal and unidentified missives

  • Detail – attention to detail, simplifying and making details discrete can help reduce common errors, something that of particular importance to the medical sector

  • Digital – the digital world allows us to use different versions of websites to simultaneously test out different strategies and measure the results of each

  • Iteration – if the answer is not known, then there is going to be continual learning and trying new things

  • EAST – make the intended action/nudge ‘Easy, Attractive, Social and Timely’.


Other important considerations included:

  • Making sure there is democratic ‘permission’ for using such approaches and that there is acceptance of government intervention in this form

  • Recognise the risk involved in our current approaches

  • Note that when using RCTs that the outcome may not be better than what exists now – and that this can help reduce concern about some groups ‘missing out’

  • Failure is an important source of learning and that trial and error is okay

  • That everyone should have some familiarity with behavioural insights, but that to do it well, public sectors will need access to some specialised skills.


Limits to nudge?


What are the limits to the nudge approach?

Some of the areas identified as being unsuitable for nudging were:

  • If there is harm to third parties (e.g. pollution harm, some types of crime)

  • Areas where government needs to remain neutral or detached (e.g. the political process)

  • Where there are strong values or social preferences and a nudge approach would be deemed unacceptable.


There were some other concerns raised including that:

  • For complex problems it can be hard to tease out causality and therefore what really works

  • It can be difficult to encourage or support testing and evaluation in an environment of fiscal consolidation

  • Behavioural insights can encourage a prizing of choice as the most valuable attribute, when it may not be the best or most suitable option

  • Sometimes interventions may simply shift a problem to somewhere else, or actually lead to worse results. A small trial or localised approach in employment services, for instance, may lead to better outcomes for those seeking employment in that area, but not lead to any overall benefit (or could actually harm the chances of others elsewhere)

  • Sometimes a ‘nudge’ may only lead to a small change/difference. How do you show that it is significant or meaningful?

  • There might conceivably be ethical concerns/risks – though there are always unintended consequences, even when not nudging

  • There might be a risk of confirmation bias, and failures or non-successful nudges will not be talked about or shared

  • The behavioural insights approach might be limiting the possible options to be explored in addressing problems

  • The efficiencies provided by nudging might unintentionally improve or maintain a sub-optimal strategy at the cost of exploring newer and better options.


Overall there was an acknowledgement that experiments aren’t going to be able to solve everything or be appropriate in all situations.

Who is nudging?


Some of the main government actors that are already applying behavioural insights and nudges, and who were represented at the conference, included:

  • The Behavioural Insights Team (the original Nudge Unit) in the UK. It was originally a part of the UK public service (as part of the UK Cabinet Office) but is now a ‘social purpose company’ that is co-owned by the UK Government, Nesta, and a mutual ownership element

  • The US Office of Science and Technology Policy

  • The NSW Behavioural Insights Unit

  • The Singaporean Government

  • The Australian Government (various agencies).


There was also strong representation from academia, particularly from Harvard University.

How to help nudging along?


What needs to be done to help nudging along?

Some of the things discussed included ensuring that there is a good ‘choice architecture’ for the ‘choice architects’. The biggest support for that identified was the need to make data available and transparent so that results could be observed and tracked.

A number of the speakers noted that a lot of cutting-edge private sector companies (particularly in the ICT sector) are already investing heavily in behavioural insights, hiring social scientists, undertaking data analysis and continually running experiments. In some ways the public sector will need to become more sophisticated at this field, simply to keep up with the private sector.

There was some interesting discussion around recognising that as policymakers we are subject to biases and fallibility. So we are also subject to

  • Confirmation bias

  • Over confidence

  • Saliency bias (putting more significance on some developments/crises over others, particularly longer term ones)

  • Sunk cost fallacy

  • Hyperbolic discontinuity (valuing today over the longer-term).


Some of the things that could be done to help counteract these biases revolved around encouraging diversity, bringing in an outside view, and encouraging alternate voices within our organisations. Organisations should try and provide their staff with the training and experiences to develop richer and more complex mental models of the world, which will be more resilient to biases.

There was also discussion of the potential of organisational nudges, recognising that people can’t de-bias themselves. While these weren’t discussed in depth, some of the organisational nudges discussed were around ensuring decision making processes helped counter instinctual and systematic biases.

Where next for nudging?


The biggest area of potential for behavioural insights discussed was around the opportunities offered by ‘big data’ and the ability to be able to track the results of interventions much more closely.

There was also discussion of an open ‘behavioural insights’ approach, with greater sharing of experiments, experiences and tools. This might also include a register of trials, so that even the unsuccessful trials and counter-results are included and tracked.

There was some discussion about whether behavioural insights had been or would remain confined to some of the transactional/service end ‘low hanging fruit’ rather than the bigger, more complex policy-end of the spectrum of problems. I don’t think that a consensus emerged on this question, though the strongest advocates suggested the nudging would be able to be applied to most big ‘wicked’ problems.

Perhaps these questions and other matters will be discussed in more detail at the next Behavioural Exchange Conference, which will be held in London in 2015.

In summary


The conference had a very impressive line-up of speakers and considerable amount of content, and I am sure that I have missed some of the relevant material or examples, but I hope that it provides a useful summary of the event.

For further reading and insights from, or about, the conference you can also refer to the following articles and posts:

You can also read our previous posts about behavioural insights:

A ‘nudge’ in new perspectives.