Introduction
The Behavioural Insights Team (BIT) has been developed in the UK Cabinet Office and has evolved into an international social purpose consulting organisation with an ownership structure and ethos that permit elements of capacity-building research alongside provision of direct services. They have unquestionably led in the promotion of the adoption of the behavioural literature in public policy. Sanders, Snijders and Hallsworth (Reference Sanders, Snijders and Hallsworth2018) provide a thought-provoking overview on the development of BIT and the potential future directions it might take. There are three points I would like to add to this discussion; firstly, how will BIT and the wider behavioural policy literature develop from a focus on small, single trials to scaling and large-scale policy analysis? Secondly, how should ethical and professional standards issues be dealt with in a pragmatic way in practice? And thirdly, can a consideration of the intellectual history of behavioural science be informative for future applied directions? The three questions are to some extent interwoven and concerned with the principles of applied policy research, which might be set aside under the time pressure of conducting many applications.
Scaling and mainstreaming
The issue of methodologies for scaling interventions is identified prominently by the authors and is clearly vital for the future development of the area. BIT has become synonymous in the last decade with site-specific randomised controlled trials (RCTs) and some of their early trials attained an exceptional degree of media and public policy profile. This clearly helped to promote the development of the behavioural literature in policy, but has arguably led to an over-focus on small RCTs and too little conceptualisation of the many and varied issues involved in scaling interventions over a population. An excellent recent paper by Banerjee et al. (Reference Banerjee, Banerji, Berry, Duflo, Kannan, Mukherji, Shotland and Walton2016) provides a comprehensive discussion of issues in scaling, including the role of administrative enthusiasm, demographic variation, multiple equilibrium effects and so on. The development of models of practice that incorporate such effects would be a major development for BIT through the next decade. Collaborations between BIT and other organisations with specialist econometric modelling capacities such as the Institute for Fiscal Studies (IFS) is worth considering in this regard.
A key issue for the mainstreaming and scaling of behavioural science in public policy will be the credibility of the ongoing findings. I agree with the authors that the maturation of this field into a normalised area of public policy will require the development of research protocols that take into account the serious potential for publication bias and related methodological issues. There are many pressures in policy environments that can create an atmosphere where reporting null results is discouraged. BIT has made ground on this issue, but there are clearly substantial practical obstacles to achieving a model where trials are focused on importance in terms of human welfare improvements rather than potential salience, where protocols are transparent and available in advance and where null results or negative results are published as readily as trials with positive outcomes. This is an issue not just for BIT, but for all of us involved in developing and applying the area of behavioural public policy.
It will also be important to develop a greater connection between the work being conducted on behavioural trials and the wider cost–benefit literature. Recent papers have shown impressive experimental evidence on the impacts of behavioural trials on behavioural change in policy-relevant contexts. However, there is a danger that behavioural insights trials will accumulate a large amount of local information on projects specially selected for their suitability for treatment and with outcomes determined by local agency pressures. Grüne-Yanoff (Reference Grüne-Yanoff2016) discusses the limitations of examining impact in randomised trials when it comes to uncovering deeper-level mechanisms. Harrison (Reference Harrison2014) argues forcibly against a ‘what works’ approach divorced from theory and advocates the development of trial frameworks that include information on latent constructs such as risk and time preferences that permits calculation of programme benefits and greater comparability across programmes. The potential for inclusion of information on well-being and economic preferences in behavioural trials to aid in this objective is important to discuss further (e.g., Layard, Reference Layard2005).
Ethics and practice
One area I would like to have seen the authors address more comprehensively is the ethical dimension of behavioural public policy. As Sunstein has pointed out in several places (e.g., Sunstein, Reference Sunstein2015a, Reference Sunstein2015b, Reference Sunstein2016), there is a danger that discussions of ethical issues can veer off into abstractions about the nature of government that will have little influence on policy. However, an organisation such as BIT has a strong opportunity to lead in the development of pragmatic models of ethical behavioural public policy. Policy frameworks such as MINDSPACE and EAST provide accessible frameworks for conceptualising behavioural change in real-world settings (Cabinet Office, 2010; BIT, 2014). However, they also potentially frame behavioural research as a set of tools to be used to achieve an outcome, rather than capturing the wider dimension of the responsibility of state and private organisations to the people involved. Much thought is needed on this issue.
One of the advantages of behavioural projects is that they can often be developed and rolled out quickly compared to mandates or other interventions. But such speed may also mean lower levels of scrutiny and ethical review. The use of ethical checklists in policy projects and the development of norms governing practice in this area are some ways of ensuring that the growing enthusiasm for this field by governments does not lead to citizens being exposed to interventions that are potentially harmful. Questions such as the extent to which any interventions involve manipulation or deception, the transparency of the intervention, potential harm, public acceptability and several others are all important for both the effectiveness and ethics of behavioural policies, but to date there is a gap between a large theoretical literature discussing these issues and the applied policy literature and policy applications more generally.
The question of ethics points also to a wider problem in the emerging area of behavioural public policy. As well as PhD programmes, many European programmes now exist training people at graduate level in one- and two-year programmes. However, there is a lot of uncertainty still as to what counts as a professional behavioural scientist. What training do people need to make claims to expertise in this area? What professional standards should people working in this area adhere to? Clearly, most researchers in this area are subject to standards of evidence in ethics set out in their contracts and in the journals and other outlets they publish in. However, matters becomes more unclear for practitioners essentially working in a consultancy capacity in a range of newly emerging and not fully defined roles. BIT has worked with top researchers and advisors throughout its development, but it has not, to date, published explicit statements of its approach on ethical issues and issues of competency and expertise with regard to its projects. As a first-mover and leader in this field, it is in a good position to help develop the state of the art in this regard. However, it is clearly not just a responsibility for them alone, and it is worth serious thought at this stage as to what a professional body would look like in this cross-cutting area, one that at the very least offered guidelines on ethical practice, scientific standards, competencies required for different projects and so forth.
If behavioural research is to be scaled, this also raises the issue of how this is to achieved. A lot has been written and will continue to be written about the insights team model, but we need to think more about the relative merits of different models across different situations. Some countries have moved explicitly towards insights team models. Others have developed capacity within specific policy areas. Training of public servants in the capacity to use behavioural evidence is another promising route being explored by various groups, including BIT. Questions abound about the optimal structure of training, career structure and organisational form in this area of policy. In particular, is a central government team the best model for developing this area in different countries or would a model where individual capacities are developed in different agencies and departments be preferable? Or should a thousand flowers be allowed to bloom, with different countries finding mechanisms that fit their own policy environment best? Once again, these are questions for BIT, but also for all of us involved in training and research in this area.
Intellectual context and field coherence
The wider intellectual context of BIT is also worth dwelling on. Much ink has and will continue to be spilled on what the emerging area of behavioural science actually consists of. Is it a rebranding of psychology? When did behavioural economics really begin and how does the current work relate to many previous efforts to integrate economics and psychology? Such questions can lead to a type of naval gazing that is at odds with the pragmatic considerations of an organisation conducting trials in real time with time- and attention-constrained policy officials. However, I think ultimately the fuel for BIT and the wider enterprise of behavioural public policy comes from considering these intellectual roots.
With that in mind, it is clear that the dominant intellectual influence on BIT is the interest in behavioural public policy that emerged from the work of Thaler and Sunstein in the 1990s and 2000s that culminated in their book Nudge (Reference Thaler and Sunstein2008). Much of the work conducted by BIT is in the spirit of Nudge and the libertarian paternalism tradition (Thaler & Sunstein, Reference Thaler and Sunstein2003), whereby interventions based on behavioural research are used to influence people's behaviour while allowing them the freedom to avoid the influence. However, it is also clear that BIT is not wedded to this philosophical framework, and that the type of work they have specialised in conducting can apply equally in cases where mandates and other hard interventions are involved (House of Lords, 2010). This is a wider point for the behavioural public policy literature, and there is clearly a role for behavioural design in considerations of the merits of mandates as opposed to nudge-type interventions, rather than focusing solely on development and testing of nudge interventions.
The longer intellectual foundations of behavioural science also have relevance for thinking about the origins and development of BIT. As documented by Halpern and others, the success of BIT clearly involved the support and active involvement of Richard Thaler in particular and the wider interest in behavioural economics sparked by Daniel Kahneman and the success of the book Nudge (Halpern, Reference Halpern2015; Thaler, Reference Thaler2015). There is a broader intellectual context also. London, in particular, is the home of utilitarianism. There is an intellectual tradition that crosses areas such as law, economics and psychology that arguably makes London particularly suitable as a world centre for applied behavioural public policy. It would require a much longer study to isolate the various threads, but it is clear that many of the literatures that fuel behavioural public policy have their roots in British empiricism traditions that permeate many of the institutions BIT has drawn from. Tapping more into this history is a potentially fertile source of imagination and invention for the organisation and the field in general.
With regard to the near history of behavioural science, the recent award of the Nobel Prize in Economics to Thaler underpins the role he has played in shifting economics back to being an explicitly behavioural discipline aspiring partially to realistic micro-foundations. It is fitting that he is mentioned in the Sanders et al. (Reference Sanders, Snijders and Hallsworth2018) paper, along with Sunstein, as one of the chief intellectual inspirations for the work of BIT. It would also be interesting to think about how the work of BIT can be developed through consideration of other traditions in the literature. For example, the work of Herbert Simon has much resonance with potential future developments for BIT, most importantly in developing templates and models that could allow the development of interventions aimed at firms and organisations, as well as those focused on individual-level consumers and citizens (e.g., Simon, Reference Simon1966, Reference Simon1987, Reference Simon2000). As BIT expands to move from studies examining consumer behaviour, this work could provide a substantial source of ideas for influencing administrative behaviour across a wide range of contexts.
A number of books have emerged recently that give shape to academic training programmes in the broad area of behavioural public policy. Shafir et al.'s (Reference Shafir2012) extensive edited volume on behavioural science and public policy, Oliver's (Reference Oliver2017) overview of the foundations of behavioural public policy, undergraduate textbooks by Angner (Reference Angner2012) and Wilkinson and Klaes (Reference Wilkinson and Klaes2012) and Dhami's (Reference Dhami2016) extensive graduate textbook on behavioural economics all have begun to provide the basis for a pedagogy in the areas of behavioural economics, behavioural science and public policy. Having said that, the definition of the field of behavioural science still remains elusive, and more work is needed on clarifying the relationships between the disciplines involved.
Conclusion
In summary, BIT has been at the forefront of the development of the transdisciplinary field of behavioural science and public policy. The wider field itself has a great deal to grapple with in terms of developing measurement and evaluation structures that are rigorous and pragmatic. This will be increasingly important during a period where work in this area increasingly is conducted at scale. Debating ethical issues and questions of field definition and location of the current field within wider historical traditions can lead to abstractions that are not useful in ongoing policy efforts. However, the debate around such issues will increasingly provide oxygen for the future development of BIT and other groups in this area. My hope is that BIT resists the obvious temptation to ignore the more difficult elements of these questions and instead devotes capacity to explicitly leading on the development of wider structures in this field. Their recent programme of joint PhD projects is one potential mechanism to develop their thinking in this domain, and in general their wide networks within academia offer potential for them to lead in the integration of field development and applied practice.
