Policy Design for Humans
Governments from Canada to Singapore are embracing findings from Behavioural Economics to improve the lives of their citizens.
FROM REVENUE COLLECTION to trade policy to infrastructure investment, the varied mandates of government are complicated and diverse. Yet at its core, the role of government is simple: To maximize the welfare of citizens by improving their wellbeing, creating fair and efficient marketplaces and planning for the future.
Governments attempt to accomplish this by helping citizens, organizations, their own agencies and local businesses make good choices. As such, the government — just like every other organization that exists — is in the business of behaviour change. Indeed, almost every activity that a government undertakes boils down to the need to encourage or discourage certain behaviours. There are four particular types of behaviour change that policymakers focus on:
1. COMPLIANCE. Getting people and businesses to behave in accordance with prescribed standards and by certain deadlines.
2. CHOICE SWITCHING. Encouraging citizens to perform certain tasks online or to save for the future.
3. CONSUMPTION. Promoting consumption takes many forms — from getting seniors to consume their medications to getting young people to eat healthily.
4. ACCELERATION OF DECISIONS. In many situations, officials want to accelerate decision making in important areas — e.g., for businesses to start implementing environmentally-friendly policies.
Given the centrality of behaviour change to virtually everything governments do, it is a matter of great surprise that until recently, most governments had scarce capabilities in the science of behaviour — or, as it is referred to nowadays, Behavioural Economics. Sure, every government has a chief economic advisor and cadres of traditionally-trained economists who develop and implement policy; but since the ultimate goal of every governmental policy
is to influence behaviour, we are surprised at how few governments have invested in hiring a chief behavioural scientist.
The reason for the dominance of traditional Economics in government can be explained by using the language that Richard Thaler and Cass Sunstein introduced in their 2008 book, Nudge. In it, they make a distinction between two completely different types of agents: ‘Econs’ and humans. Econs are highly-sophisticated decision makers who consume vast quantities of information with ease and have infinite computing abilities — much like the robot on the cover of this issue. They also maximize self-interest, are forward-looking and consider the future impact of every decision they make — never letting emotions get in the way. In short, they obey all of the laws of Economics.
In contrast, an abundance of research shows that humans are emotional, impulsive, cognitively lazy, and have difficulty dealing with large quantities of information or choice options. As a result, a number of commentators have referred to Econs as ‘rational’ and to humans as ‘irrational’ — as if to suggest that human decision making is inherently flawed.
Our view is slightly different. As one of us ( Dilip Soman) noted in his 2015 book [ The Last Mile: Creating Social and Economic Value from Behavioural Insights], the fact that humans do not obey the laws of Economics is not a surprise. Humans were never designed to solve complex inter-temporal maximization problems or to sift, curate, analyze and act on large volumes of data. The very assumption that humans actually behave like Econs is itself an example of irrationality.
If citizens were indeed robot-like Econs, the task of behaviour change for governments would be relatively easy and could involve three simple instruments:
1. RESTRICTIONS. Bans, legal restrictions and other forms of regulation limit access to certain options, thereby creating a behavioural shift towards the desired alternative.
2. INCENTIVES. These can be either positive incentives in the INCENTIVES. form of ‘carrots’ (i.e. subsidies or fee waivers) or negative incentives in the form of ‘sticks’ (i.e. surcharges or penalties).
3. INCREASED INFORMATION. The provision of additional information and sometimes, more options, is widely believed to improve decision making.
The problem is, governments struggle with making policy decisions work because these three tools are designed for Econs rather than humans — and the research is filled with examples of the problems that result. For example, the Canada Learning Bond — a welfare program that supported children’s education with ‘free money’ — garnered a take-up rate of only 16 per cent in the two years after its launch; and in the U.S., several welfare programs have suffered from similarly-low take-up rates.
Elsewhere, attempts to get citizens to pay their taxes online — or to get flu shots, donate organs, eat more vegetables, or read privacy policies designed to safeguard their online information — have all fallen on seemingly-deaf ears, despite large expenditures on advertising and communication. The reason is simple: The vast majority of these policies and programs are designed for Econs, rather than for humans who are forgetful, emotional and impulsive; influenced by their peers; confused by too much choice; and loathe to consume too much information.
The best policy design, then, would assume that people will likely forget, ignore, gloss over or misunderstand critical pieces of information — and build safeguards into the system against such behaviour. Fortunately, recent advances in the world of behavioural insights have provided governments with a new toolkit to achieve this, and it is being embraced by governments the world over.
The Basics of Choice Architecture
The term ‘choice architecture’ made its debut in Nudge, where Thaler and Sunstein argued that since we know from Psychology that context influences choice, it should be possible to design contexts to steer choices to a desired outcome. Choice architecture therefore refers to the conscious and careful presentation of different options available to a decision-maker, and interventions to change the manner of option presentation are called ‘nudges’.
Choice architecture draws upon findings from behavioural science to design environments in which humans make decisions. For example, every policy initiative comes to the attention of citizens with a pre-chosen default status: you either check the box to donate your organs, or you don’t. And studies have shown that changing that default has significant effects on behaviour.
Enrollment in 401(k) pension plans in the U.S. is a prime example. Signing up for a 401(k) can be a hassle, and retirement seems far off in time for many people. By using a default ‘opt-in’
Like businesses, governments and non-profits should be constantly iterating on their service offerings.
enrollment, employees have been automatically enrolled and participation rates have increased significantly. Between 2010 and 2014, the number of companies with an 80 per cent participation rate or higher rose by 14 per cent.
Elsewhere, the UK government sought to apply choice architecture to the decision its citizens made regarding organ-donor registration. The idea was that tweaks to processes and language — informed by behavioural science and tested for effectiveness — could significantly improve participation rates. This work was spearheaded by the UK’S Behavioural Insights Team (BIT), the world’s first behavioural insights unit within government. The most successful message out of those tested was the following: “If you needed an organ transplant, would you have one?” By invoking the concept of reciprocity, this simple question encourages potential donors to think a bit more about the decision. The result: BIT estimated that it would be able to add 100,000 names to the donor registry annually.
The Canadian province of Ontario has also succeeded in increasing organ-donor registration rates by harnessing two simple behavioural insights to design nudges: First, as seen in the UK, a message that evokes empathy was used to get potential donors to think a bit more about the decision; and second, simplifying the application form itself increased the likelihood that this greater thought would be converted to action.
The Role of Testing
The gold standard of applying insights from behavioural science involves the use of Randomized Controlled Trials (RCTS). While the name might sound intimidating, RCTS are no different from the trials used in the world of Medicine to test for the efficacy of new drugs, or the A/B tests used by online businesses to test layouts of webpages.
With an RCT, various options designed to encourage certain behaviours are tested amongst a sample population. This often entails very subtle changes to materials or to the context, such as creating multiple versions of an intervention (say, an application form, a brochure and an application process) and then trying all versions simultaneously.
One of the key strengths of applying behavioural insights is the ability to test nudges on a sample of real-life users, prior to the full implementation of a program. This allows an organization to receive valuable feedback on the effectiveness of its proposed changes and to gauge potential impact before widespread implementation.
Much like businesses, governments and non-profits should constantly iterate on their service offerings and procedures. Testing different nudges provides an outlet to review the status quo and look for new ways to improve interactions with the public. Few would argue with this logic of continuous improvement, but if this is the case, why have so few governments embraced this approach?
The answer is likely inertia and the need to change mindsets. Given that many policymakers have been conditioned to think about citizens as Econs, they are also conditioned to think that economic theory can predict the best way of creating behaviour change. Once a policy or program has been approved, the thought of having to test it for effectiveness in the field and designing a scientific experiment to do so may seem daunting, unnecessary or threatening.
The fact is, using behavioural science to uncover policy insights requires a certain degree of humility. Governments are often divided into silos, with subject experts operating in each area. The status quo expectation is that government branches inherently know how to improve or implement new programs because of their past experience, but when working for so many citizens — all of whom behave differently in different contexts — past experience does not necessarily predict future outcomes.
As a result, the dangers of not testing are significant. An example is the Scared Straight program of the 1970s in the U.S., whereby young people committing minor offences were taken to prisons and introduced to inmates, in hopes that the experience would scare them from committing future crimes. Little testing was conducted on the effectiveness of the program — which in hindsight, seems to have only normalized the idea of a life of crime with some of the young people. The result of implementing a flawed policy was disastrously costly: the Washington State Institute for Public Policy estimated in 2004 that every dollar spent on Scared Straight programs incurred a further crime cost of $203.51.
Challenges (and Solutions) in Conducting RCTS
Although RCTS are vastly beneficial in uncovering the effectiveness of proposed behavioural nudges, governments may face technical constraints, such as the availability of data.
Testing different nudges provides an outlet to review the status quo and improve interactions with the public.
Hasti Rahbar, research advisor at Employment and Social Development Canada (EDSC) told us that often, the data required for designing an appropriate nudge for a particular problem is not readily available — or is not even being compiled.
In British Columbia, where the provincial government recently launched its own behavioural unit, this was one of the key challenges it faced as it started on its initial roster of projects. For understandable privacy reasons, data is often held separately and securely, and this means that “the process to acquire data can take a longer than anticipated,” says Heather Devine, Head of BC’S Behavioural Insights Group.
Governments may also struggle with the existence of ‘touchpoints’, or points of contact between a government and its citizens, which can include mail, phone and face-toface. Behaviourally-informed approaches can most easily be implemented at these touchpoints. At the federal and provincial levels, there are limitations to the number and variety of touchpoints with citizens. Sometimes, the results of a proposed behavioural intervention cannot be analyzed simply because the touchpoints are not there.
Despite these limitations, the world of behavioural insights and choice architecture design offers a number of other avenues to test. If an RCT is not possible, perhaps a laboratory experiment, a series of design workshops or a natural experiment might be possible. As long as data is collected to compare multiple nudges with the status-quo (i.e control) condition, governments can learn, iterate, adapt and launch tested interventions.
Even though governments the world over have started to embrace the power of applying behavioural insights to policy
and the importance of testing, much more can be done to enable progress in this space. Two key areas of best practice are:
• Collaboration and joint initiatives between behavioural
units; and
• Research by and consultation with academics.
In many cases, the problems encountered in government are not unique to a single level or branch of government, so collaboration on projects can lead to shared learning and greater overall improvement. In Canada, hubs at the provincial level are working on projects in tandem with hubs at the federal level, pooling their resources and knowledge. There is also vast potential in establishing hubs at the municipal level: Municipalities have access to many more readily-available touchpoints, opening up a wide variety of opportunities to incorporate and test behavioural insights as they relate to policy improvement.
Another trend worldwide is the central role that academic institutions can play. Behavioural units in the UK, U.S. and elsewhere have tapped into the expertise of the academic community to identify and develop a framework for problems, to design trials and to analyze, interpret and iterate on the learnings. In Canada, Behavioural Economics in Action at Rotman (BEAR) collaborates with the Ontario government; Rotman Professor Nina Mažar was appointed as a behavioural scientist at the World Bank; and she and one of the authors [Prof. Soman] serve as advisory to the federal government’s Innovation Hub at the Privy Council Office.
The bottom line is this: Insights from Behavioural Economics can simplify procedures for citizens and better clarify what they are being asked to do and why they should do it. As the world becomes increasingly digital, governments could seek to add an additional channel of communication through mobile technologies, such as SMS. Behavioural insights can also help significantly in pressing policy areas such as poverty alleviation, education and public safety. In the end, by using approaches tailored to how citizens actually think and act — not how policymakers believe they should think and act — governments can provide better services at a lower cost.
In closing
Behavioural hubs in government are proving that innovation isn’t reserved for Silicon Valley or Fortune 500 companies. Along with better data and an improved ability to test, behavioural insights will play an increased role in improving policy and services to ensure a better life for every global citizen.
Dilip Soman is the Corus Chair in Communication Strategy and Professor of Marketing at the Rotman School of Management and co-director of Behavioural Economics in Action at Rotman (BEAR). He is the author of The Last Mile: Creating Social and Economic Value from Behavioural Insights (Rotman-utp Publishing 2015). Katie Chen is a research assistant at BEAR and a student at Western University. Neil Bendle is an Assistant Professor of Marketing at Western University’s Ivey School of Business. This article is based on a longer report entitled “Policy by Design,” available for download at rotman.utoronto.ca/facultyandresearch/researchcentres/bear