Behavioural public administration: evidence from around the world

Andrei Turenko: In our final presentation today, we have Mia Drazilova and Haris Khan from the OECD, who are going to flip the script a little bit and discuss the idea of behavioural public administration. Mia's background is in behavioural insights and she's currently managing an international network of behavioural science experts and policy makers, which aims to share knowledge and global best practice on behavioural science and public policy. Haris, on the other hand, provides technical advice on behavioural public policy. He's a project lead at OECD. He has previously worked in Canada, where he was part of setting up behavioural science units and applying evidence-based decision-making to a wide variety of topics. Thank you, Mia and Harris for joining us. Over to you.

Haris Khan: Thanks Andrei, just checking that everyone can hear me. Thanks for the warm introduction and thanks to the BETA team for the invitation today. We're joining you from Paris where it's still pitch black outside, but we were listening into the earlier conversations and I think I speak for Mia and myself when I say that they were all fascinating. Again, it's a pleasure to be here today. This is actually my second time presenting to BETA, the last time was in 2018 at the BX conference. So happy to be back, even if virtually. And so as mentioned, our plan today is to talk about behavioural public administration, which is very much kind of behavioural science, turned inwards on ourselves as public servants and we want to talk a bit about the tools that we have, in order to make government more efficient and more effective using behavioural science. We'll share some evidence from around the world, some examples that we've worked on here at the OECD with partner countries. And then quickly we want to talk about a new initiative that's coming out of our public governance committee, which is called Government Unstuck, which has a lot to do with how governments can use tools like behavioural science to help deliver impactful reforms and make sure that we're having the impact that we'd like to see on citizens around the world. So this is our plan today. If that sounds good, I'll hand it over to Mia to give a quick introduction to our team and what we do, and then we'll get into the content itself.

Mia Drazilova: Perfect. Thank you so much and also welcome, hello everyone from my side as well. So I'll begin firstly by outlining our work on behavioural science at the OECD. So we always say we have three missions, so we convene, we enable and we advise.

And what this means in practice; we convene a global community of behavioural science experts around the world through our BRAIN network. So that's short for behavioural research in action international network. We enable governments through various tools and training to start to deepen the use of behavioural science and we advise governments on concrete behavioural challenges. So to support our international community, we also try to provide various interactive tools. So for example on our website we offer an interactive map with different behavioural insights units or different projects, so you can, if you are interested, find a project on a certain topic or see what other units exist around the world, you can visit the page or if you start an exciting project in the area of behavioural science, you can use the pre-registration function to share those designs and see what others are doing and this really helps to lift the quality of transparency in our international behavioural science community. If you go to the next slide, Haris, as I mentioned, we convene this international network. We try to convene and provide meetings regularly, so one is hybrid and one is online each year, and besides that we organise various work streams on different themes so these include for example sludge, artificial intelligence, and also one on the topic of today's presentation, which is the behavioural public administration.

If you go to the next slide, as I mentioned, we also try to enable governments around the world with various tools and publications, guidance, toolkits, etc. So some of them, just to mention a few, is we provided a toolkit called good practice principles for ethical behavioural science in public policy. Our basic toolkit, which really helps policymakers to apply behavioural science when they are starting a project or for policymakers who are keen to find out more about different methods, we provide them with a toolkit called 7 roads to experimentation, which guides them to use the correct method in their concrete policy issue. Our recent work has been the LOGIC report, which essentially sets out good practice principles for policy makers, guiding them on how to embed behavioural perspectives in their routine policy making, and in their administration. As part of this report, we've also analysed the status quo and how behavioural science is currently applied around the world. And we've noticed that over 300 institutions now apply behavioural science to public policy. Now, of course, this is changing every time we look at it, but some trends that we've noticed is that the field is broadening as more and more countries are establishing behavioural science units and also deepening. So the behavioural science functions are now embedded more in various functions and teams across governments, and that's why it's also very important for us to, from the OECD’s perspective, to be able to stay up to date with what's happening around the world. And some of these resources that we provide online, help us to keep up with what's happening around the world and also to be able to see what the governments around the world currently are working on. So this was also useful for us, for the work that we are going to talk about more today, which is the behavioural public administration and Haris, over to you to introduce this in more detail.

Haris Khan: Sure. Thanks, Mia. And it's worth mentioning, we'll have a QR code at the end of the presentation with a link to all the resources that we're mentioning today. So you don't have to worry about Googling them just yet, but you'll get to see them at the end quite easily. But we're here to talk about behavioural public administration, which, as discussed, is the idea that we can turn the tools of behavioural science that we're perhaps comfortable with using in the public setting, on government officials themselves to make our work more effective, more efficient and, basically more impactful for citizens. I assume I'm talking to people who know a lot about these kind of biases. But I just want to talk about two really important biases that I think affect us as public servants in particular. We're all humans, obviously. So they all apply. But in general, I think for our purposes, confirmation bias and sunk cost fallacy, I think are particularly impactful or important in the context of government decision making and government service delivery. So just for a bit of review, confirmation bias is the idea that individuals seek, remember and prefer information in a manner that confirms their prior views and there's a couple of great examples of this. The Jelveh paper in 2015, found that economists find statistically significantly findings in their papers that confirm their prior policy beliefs.

And you know, people like me who are raised in economic schools, maybe we're taught that it's a hard science that is not susceptible to these kind of biases, but we see it obviously in the publications that are made. And Cass Sunstein and his colleagues found an interesting result with judges in the US who found that basically, judges are more likely to make ideological based decisions when they're in a group full of people who share the same ideology as them, from a political point of view, this was in the US, so it was very much Republican or Democrat. And again, I think this is super interesting because often times, we as public servants, find ourselves in rooms full of other public servants who are making decisions about people who are not in the room, and providing policy advice on topics that don't affect us personally. And while we're obviously capable of doing that, as much as possible, I think we really should be trying to get those diverse perspectives into our thinking and our decision making and then that will help us make better decisions and better recommendations for our Ministers moving forward. So it's important again to have a diversity of lived experiences and thoughts in a room while we're deliberating.

If we move to sunk cost fallacy quickly, this again is the tendency to continue a project once an investment of resources has been made. We do this all the time in our personal lives, but the World Bank and the UK Department for Foreign Affairs and International Development, did a really interesting study where they basically asked their policy makers a question which had to do with continuing on a biodiversity project, that was $50 million worth of investment, in the situation in which there was a provincial governor who was going to build the dam that would, in the end, make that biodiversity project obsolete, would basically flood the area in which the biodiversity project was occurring. And the question was, given X percentage of funds already spent, how much more money would you invest in this project? And it was really interesting to see that despite the fact that the rational actor would invest zero extra dollars into the project because it was going to be flooded, in every situation, there was a significant portion of policy makers who invested more in the project. And the people and the percentage of investment correlated with the amount of money that was already spent, as part of the project. So for example, if someone was told that 40% of the budget had been spent, the average response was that people would invest more, but if people were told that 80% of the project budget was spent, that investment amount on average was higher. So not only are we making less than perfectly rational decisions, it's affected by supposedly irrelevant factors, which I think constitute a classic bias. So this is just to say that we are just like everybody else, we're susceptible to these biases and perhaps there's things that we can do to guard against these biases, and indeed use the biases that we have, to make our work more effective and more efficient. Here is a quick table that shows what we're talking about when we're talking about behavioural public administration.

Basically, the idea is that the target people that we're trying to affect are civil servants, it's about making the public administration process more effective, improving it in some way and supporting effective decision making and implementation. Oftentimes this means that we're working on policy development and decision making, project management, HR, procurement, kind of the classic functions of government with the intent of using evidence better, fighting corruption in some cases, and making efficient internal processes. And I just want to make a point here. I was talking to a colleague about this yesterday and you know, while the Republican administration is about making the machine of government work better, I think it's still important to remember that the goal of the machine of government is to create value for citizens. And so just making a process more efficient for the sake of making a process more efficient is not the goal here. The goal is to make the process more efficient because it results either in better outcomes for citizens or less expenditure on actually administrating the process and therefore frees up resources for other points. So while you know impact on citizens is very much a second order effect, it's really important to keep that in mind, when we're thinking about where the most effective ways to leverage behavioural public administration are.

Now, I would like to showcase two projects that we're doing. Unfortunately, they're both kind of in the stages where we can't share results, but Mia will share results from other previous projects in a second. We have two projects with Italy and with Greece both tackling behavioural public administration from different points of view. In Italy, the idea is very much around helping individuals and organisations set better performance objectives. Making sure that these performance objectives are more geared to the public value, to measurable outcomes and they're not process based. So for example, we want to say a person was able to increase customer service by X percent rather than, you know, simply going through a number of cases at a time.

And this actually informs a lot of the Blueprint report, an actual plan that we'll be releasing very shortly, at the beginning of next year, which is an overview of how behavioural science can be applied to behavioural public administration.

In Greece, we are working on a couple of different things. Basically, we're helping their Ministry of the Interior use behavioural science to improve a set of HR related reforms and most interestingly, that's around increasingly attractiveness of the Greek public sector as an employer. We recently did an experiment that found basically that appealing to the civic duty aspect of a public service career, so appealing to the idea that joining the public service is a way of helping your fellow citizens and a way of improving society, was found to be super effective in driving interest in a career in the Greek public service.

Both these results are very much in process, but we're hopeful that we can share tangible results very soon next year. Before I hand to Mia, I just have one last framework to present. This comes out of the Blueprint report that I mentioned earlier that will be released shortly. What we're trying to say here, is that basically, behavioural public administration is relevant across the policy cycle, on the top row you can see that when we're analysing and designing policies, having a guard against the biases that we may have, can be useful when we're implementing the policies, things like sunk cost fallacy, can really affect decision making at that time. And we can make better decisions if we're guarding against biases like that.  And then on the monitoring evaluation side, I know earlier in the day we talked a lot about evidence generation and how to make sure that we're able to use evidence to the maximum and so I think again we could use behavioural science on ourselves, to make sure that we're doing that monitoring and evaluation as effectively as possible. We think that there's these three key areas across the policy cycle behavioural science can be very helpful. One is with people improving recruitment and ensuring that the governments have the skills and the people that they need to effectively do their work.

Integrity is a big area, and Mia will present a great example about that shortly, and then lastly sludge and simplifying administrative processes and making sure that we're spending as fewer resources as possible in terms of time and effort and resources in general, in order to make our processes work and then hopefully using that to improve customer outcomes and make the government more efficient. So with that as a bit of an overview, I'd like to hand over to Mia and talk about some key examples.

Mia Drazilova: Perfect. Thanks, Haris. And if you can move to the next slide. So as Haris outlined the framework, we would just like to give you a short taste for how this is built up and what kind of examples we provide in the report. So if we focus on one of the key areas that we see, that is the problem analysis and policy design, we looked into what could be the factors based on research and applications by different governments and what are the actionable recommendations that policy makers should do if they would like to apply behavioural public administration.  And if you go to the next slide, Haris, some things that we found, is that it's really important to, for example, reduce the messenger effects. So for instance, in situations where the identity of the messenger is not relevant, it's really important to anonymise the source of evidence or recommendations, so people respond to the message itself and its content, and not the messenger, is one of the examples. Or for instance, we know talking about confirmation bias, it's really important to use the evidence deliberately. So for example, in order to guard against the confirmation bias, we could ask the teams, the governments, to consider the opposite, and ask them to state the best reasons that their first view could be wrong. So in a way this can help present evidence in ways that limit the motivated interpretation. Another example that we mentioned here is, for instance, that it's important to frame risk in a mindful way. We're aware that loss frames could push risk seeking and gain frames could push the risk aversion. So really choosing the right frame here can help policymakers match the policy goal that they have in mind. And then one more example maybe, is that it's important to widen the perspectives. So for instance, behavioural public administration tells us here that it's important to engage the policymakers early in order to counter the illusion of similarity or, for example, the curse of knowledge that many policymakers may have, and to explain some of these insights, or to back up some of these insights with examples, I would like to give you an example from the Netherlands. If you can go to the next slide. So this is from one of the teams in the Dutch Government, and basically the challenge here was that policy makers often overestimate the behavioural effects that the policies may have, in the society, and underestimate the difficulties that people may face when they're complying with different rules or accessing government services, and so in practice, what this means is that when we know that people know what they have to do, even though they know what they have to do, they will act differently, due to different emotional, contextual barriers, etcetera, and so to address this, the Dutch Government has decided to implement a behavioural approach that is known as capacity to act and it's now also formalised as a standard in the Dutch policy making. The goal of this behavioural public administration approach that they've chosen, is to ensure that the policy and the legislation that they have, more accurately reflects the real world circumstances and really reflects how people operate and helps policy makers to provide adequate measures, and help enable the citizens to act. So this capacity to act, what it does in practice is to prompt, it is a test. So it's called the capacity to act test, and it prompts policy makers to analyse the feasibility of the proposed policies in the context of people's lives. So what does this means? Could it could be, for example, mapping citizen journeys, consulting target groups and frontline staff over possible piloting measures to evaluate the behavioural impact that they may have. And so what this does overall, essentially it helps policymakers to look beyond the knowledge that we've talked about and the technical capability and it prompts them to systematically consider the different behavioural and situational factors that may limit the citizen's ability to respond as intended. So this was the first example for the problem analysis and design. But if we now move on, for example, to one of the enablers in the framework which Haris spoke about, it's the integrity and what we argue here is that behavioural public administration can help to underpin and strengthen the public sector integrity and how we also provide in the forthcoming paper, practical examples on how the public administration can help do so. So, for example, we say that to promote ethical behaviour of policy makers, it's important to clarify both descriptive, so what others do, and the injunctive, what is approved social norms. Then we also, for example, encourage policy makers to leverage group leaders and to really focus on reinforcing norms and themes within organisations. Then, we recommend, based on the various research and examples we've looked into, to foster civil servants' social identity. So here it's important to use the frame as defenders of the public interest. For example, are strengthening the pro-social motivations.

And to give you an example of this, it's an example that we've done a few years back where we worked with the Government Office of the Slovak Republic in order to support civil servants in Slovakia to follow and be able to follow the risk management process. So for context, the key factor in shaping the effectiveness of risk management is the behaviour and the decision maker and the decision making of the actors who are involved in the process. So therefore, understanding really the behavioural elements that affect corruption management, can inform the design and the implementation and the improvement of the risk management policies. So that's what we've tried here. To approach this, we applied the OECD basic toolkit, one of the tools that I've mentioned in the beginning. So it's a framework for applying behavioural insights in public policy making and we've applied this to the Slovak Republic's risk management system and as part of the diagnostic, we found that there has been a culture of silence. So many public servants, especially the older ones, they do not feel comfortable raising risk. And we've found two main barriers. So one was the lack of understanding of what constitutes a risk and the other one was insufficient leadership modelling, what the correct behaviour should look like. And so we've done an experiment based on these findings with more than 2,500 public officials in Slovakia to test different behavioural strategies that promote risk communication and increase the likelihood to report perceived risk behaviours and if you go to the next slide, we tested 2 messages that were informed by behavioural science. So one was to speak up, so encouraging policy makers to speak up, to report the corruption risks they observe, and the other one was more focusing on the leadership example. So stating that manager’s report risks and expect their teams to do the same. And what we found, if you click one more, is essentially that both messages were able to address these barriers. We found that the leadership message had a stronger effect, but overall, the main finding for us was that behavioural science can make public integrity procedures more effective and some of the implications we drew from the results were that it's important to empower leaders to set the good standards, and for their actions, ensured that their communication is really well understood and I think this really aligns well with the previous presentation we heard today. And to make sure that there is communication and the environment is safe and encouraged. And if you go to the next slide, just before we give to Haris to quickly wrap up. So basically what's next in our federal public administration work is that we plan to launch these findings that we've collected from the numerous examples around the world, but also from the Italian project that Haris outlined, and we hope that this work is really going to encourage policymakers to adopt more of these behavioural public administration practices in their work and I look forward to continue talking with countries about this topic closer. And now over to you, Haris.

Haris Khan: Sure. Thank you, Mia. And just before we close out and provide time for questions, I want to also talk about another initiative that we're doing here at the OECD. It's called government unstuck. The idea, or the problem definition, is that governments seem to be struggling to deliver impact for reforms due to various external pressures, including lack of political will or heavy processes and the public governance committee at the OECD, which is made-up of people from all OECD countries who work on internal governance issues, has taken this on as a key piece of work moving forward and we think that behavioural science has a lot to offer in this in this space, getting governments moving and getting governments delivering in ways that are super impactful for citizens. Obviously we're helpful in the way that we analyse barriers and identify behavioural and organisational blockers to buy in decision making. We talked about that earlier in our talk a little bit. We also think that sludge reduction and base tape and friction reduction of all kinds can be really useful both internally and externally. Making sure that once a policy is decided or once the service is being delivered, we're taking all the steps that we can take, to reduce the frictions along the way in that delivery process and make sure that it's actually getting to citizens and making sure that the impacts of our work is felt. And then lastly, I'm sure it's not a surprise to anyone on the call, we are able to test and measure through experimental designs or other kinds of empirical methods, to see whether or not these reforms are actually having the effect that they're supposed to have, and what are the adjustments that we can make in order to make sure that those effects are as strong as possible? So on top of what Mia mentioned earlier in terms of the behavioural public administration work, the OECD will be releasing best practices in countries and making sure that we're able to share these insights, as well as developing a measurement framework to assess capacity and support. This will be happening over the next 18 months or so I believe. Thanks again from both me and Mia, for inviting us and for listening into our talk, if you want to get in touch in any way, if you want to join the network that Mia mentioned or if you want to see any of the resources that we talked about, you can either e-mail us at behavioural.science@oecd.org or if you follow that QR code you'll end up on our website which has all the links to the to the relevant information. Thanks again.

Andrei Turenko: Thank you, Mia and Harris, super interesting. We do have a lot of public servants in the audience who I'm sure are trying to figure out what biases they're most susceptible to right now. Thank you for your time. I'm afraid we don't have time for questions this afternoon, but I do encourage everyone to follow that link and get in touch with OECD for more information. Thank you Mia and thank you Haris.