The Hidden Costs of Behavioural Interventions

ETHOS Issue 17 June 2017

Introduction

Few social science ideas have been adopted by policymakers as quickly as the behavioural science revolution. The main attraction of behavioural interventions is the low cost of implementation: simple changes in messaging or policy workflows can meaningfully affect behaviour, often more cost effectively than taxes or regulation would. Motivated by the promise of low-cost, high impact policymaking, governments including the United States, the United Kingdom, and Singapore have set up behavioural intervention offices reporting directly to their highest officials. But the success or failure of behavioural interventions is usually defined simplistically: “Did it work? Did it change behaviour?”

Rarely do we ask: “Is society better off?”

Behavioural science is no magic bullet. While many behavioural interventions are fiscally cost-effective, even successful interventions may impose implicit costs on individuals and society. Consider a simple example. To improve public health, many workplaces now nudge people to take the stairs by highlighting the benefits of exercise and the risks of a sedentary lifestyle. But not all are made better off. Some may exercise when they should not, degrading their performance at work; others may fail to exercise, but will feel guilty for not making the effort. Is society truly better off when guilt is subtly imposed on people who were simply making their way to work?

When we consider the well-being of the entire community or societal welfare, we should weigh both the benefits from people changing their behaviour in positive ways, but also the real costs imposed on people who are made to feel guilt, social impropriety, or discomfort with their day-to-day actions.¹

In general, behavioural interventions which help direct people towards their own desired outcomes more efficiently are likely to promote welfare. For example, interventions which help people understand and interpret their CPF account statements, highlight the real costs of borrowing on credit cards, and nudge people into attending their medical appointments, are almost certainly welfare promoting.

In contrast, behavioural interventions which steer people towards the outcomes desired by policymakers may not always promote welfare, even if the prescribed behaviours are well intended. Behavioural interventions that evoke negative emotions, encourage social comparisons and discrimination, or leverage cognitive biases may impose societal costs that could outweigh the benefits from improvements in the targeted behaviours.

Behavioural interventions should not be abandoned as policy tools. As legal scholar Cass Sunstein points out,² our everyday choice environment already shapes behaviour regardless of whether interventions are consciously applied, and policymakers have a responsibility to shape this environment in the best interests of the public. However, policymakers may overuse behavioural interventions, because they are easier to implement than conventional policy. Conventional interventions that rely on regulation, taxation or subsidies are highly visible and require strong support from stakeholders because they impose costs or use coercion. However, behavioural interventions are usually incorporated to improve the effectiveness of an existing policy, making them less visible to the public. Policymakers therefore receive less direct feedback on behavioural interventions and should exercise discretion, and even restraint, when leveraging on behavioural insights for their work.

Behavioural Interventions that Enhance Welfare

Most people often lack the expertise or time to make careful deliberations based on complete information. Interventions that ease rational decision making — by providing information, de-biasing decision making, or reducing the costs of decision processes — are likely to be welfare enhancing, because they do not compromise autonomy. Instead, they help decision makers achieve their own objectives more efficiently. In Sunstein’s terms, such interventions ‘increase navigability’: they guide decision makers much as a good map does, but the decision maker ultimately chooses the destination.

The best interventions target specific problems of decision making such as our inability to compute complex financial sums, and our tendency to procrastinate and fail to plan for the future. For instance, the Central Provident Fund Board (CPFB) in Singapore has employed behavioural interventions to help members become more informed about their retirement savings, and to encourage members to plan for retirement. In 2016, the CPFB simplified CPF members’ account statements, using visual cues and graphical summaries to highlight the most salient information. They also added a salient, gain-framed financial tip to nudge members to prepare for retirement: Members were told exactly how much additional interest they would earn by the age of 65 for each $1,000 they transfer from their Ordinary Accounts to their Special Accounts.³ Since most people have difficulty understanding compound interest,⁴ such nudges could motivate CPF members to take early action to prepare for retirement.

Behavioural interventions can also help people who fail to take advantage of public benefits they are entitled to, or fail to make choices that would benefit themselves. In Singapore public hospitals, up to three in 10 patients fail to attend appointments in the specialist outpatient clinics. This contributes to operational inefficiency, and increases patient risks from untreated health conditions.⁵ Each public healthcare cluster now implements behavioural interventions, through SMSes and physical mail, to increase the salience of upcoming appointments and reduce no-show rates.

Such behavioural interventions efficiently enhance societal welfare because they do not force people down any particular path, and do not leverage biases to get people to act; they simply guide people to make rational and informed decisions that enhance their welfare.

Behavioural Interventions that Need to be Used with Care

Interventions that leverage emotions

Behavioural interventions that exploit emotions to change behaviour are often considered less coercive than traditional policy tools such as taxes and regulations. Evoking negative emotions — fear, disgust, guilt and so on — is effective, because risks and dangers tend to be more salient and easily recalled when making decisions. However, such nudges impose social costs on the public who experience negative emotions, and may also lead people to make choices that are inconsistent with their own preferences or interests.

For instance, while graphic anti-smoking advertising is widely credited with raising awareness of the hazards of smoking, there is also substantial evidence showing people are mentally and emotionally affected by graphic images printed on cigarette packs.⁶ Whether public shock and disgust is a worthwhile trade-off depends on whether graphic anti-smoking campaigns actually discourage smoking, which is more challenging than simply raising awareness of its hazards.

While the public health hazards of smoking may justify emotionally stressful interventions, negative messaging is increasingly being considered to address other health-related issues, such as sedentary lifestyles, obesity, and the consumption of sugar, meat, and fat. Those who have no serious health problems will likely resent being made to feel guilty or shameful at the occasional indulgence. Others, who are trying hard to improve their habits, may likewise feel shame and stigma⁷ if the interventions are not carefully calibrated. Negative messaging imposes costs, just as taxes do, and should be used with the same caution.

Interventions that leverage cognitive biases

A wide variety of cognitive biases, associated with “System 1” (quick, subconscious, emotive) decision making, affects how we make choices every day. Policymakers recognise that emotions and biases may cloud rational judgment — which is why a ‘cooling-off’ period applies for purchasing insurance and political voting. But the same concerns apply to policy interventions which leverage cognitive biases, when people may be nudged into decisions inconsistent with their own welfare.

Default policies, which rely on our tendency to passively accept the status quo, are highly effective. Nearly all adults are organ donors in countries where consent is presumed by default, whereas less than two in 10 consent to be donors in countries where people must actively choose to become a donor.⁸ However, leveraging the default bias can fail to enhance welfare because the choice made by the policymaker may differ from the choice preferred by an informed decision maker. For example, employees tend to invest their retirement savings according to the default asset allocation set by their pension fund,⁹ regardless of whether the default allocation meets their needs.

Changing policy outcomes is also not the same thing as changing the minds of decision makers. A 2014 survey on organ donation in Singapore suggests that only 60% of those surveyed were willing to donate their organs, even though very few actually actively opted out.¹⁰ For the policy default to be a real nudge, policymakers need to make a concerted effort to ensure that individuals are aware of the real consequences of their decision (or lack of a decision). Otherwise, a policy default is a shove rather than a nudge, and may have repercussions if the public believes that a decision has been made against their own preferences.

Interventions that leverage social norms and alter perceptions of what is socially acceptable

Advertisers and governments have long recognised the impact of social norms.¹¹ However, the recent wave of policy enthusiasm for social norms is motivated by systematic evidence showing that social norms can change behaviours in a cost-effective manner. Social norm-based nudges have since been used to encourage citizens to pay taxes on time,¹² to pay their foreign domestic worker levies,¹³ and to act more graciously on public transit. By and large, these measures have nudged behaviour in the direction intended by policymakers. However, as governments apply social norms to more domains, they should also consider the risks from overuse.

Social norms work because people judge behaviour against a social yardstick, and feel implicit and explicit pressure to conform. In the short run, social norms interventions impose mental costs on people who — for whatever reason — resent having their behaviour measured and compared to others. Research on Opower, a social norms intervention to lower energy consumption by comparing energy usage levels between neighbours, suggests that mental costs incurred by individuals who dislike the intervention eliminate half of the gains to society.¹⁴ While lowering energy consumption has great social benefits, policymakers should be mindful of the mental costs.In the long run, the persistent use of norm-shaping interventions may increase social polarisation, and prime society to discriminate more harshly against minority behaviours. In Taiwan, priority seats on public transit have caused conflict when seemingly able-bodied commuters who take the priority seats are bullied by bystanders aggressively enforcing the social norm. In late 2016, the Taiwanese government received a public petition to remove priority seats, on the grounds that designating priority seating is counterproductive to improving overall civility and graciousness on public transport. Using social norms to promote civic-mindedness may eventually replace altruism with social obligation; people may feel forced to act kind, rather than feel any intrinsic motivation to be kind. While one social norms intervention is unlikely to change society fundamentally, the overuse of social norms messaging across different domains may have a multiplier effect in accelerating social polarisation. People may be gradually primed to discriminate against minority behaviours not only in the domains where social norm messaging is applied, but also in other societal issues.

Interventions that are too narrowly focused on agency goals

Agencies naturally seek to use behavioural interventions to solve policy challenges in their domain. But there is a risk that behavioural interventions could be overused in pursuit of, say, agency performance indicators, rather than outcomes that truly enhance societal welfare. Take one of the fastest growing uses of nudges: bill collection by public agencies. Bill collection nudges have cost-effectively increased on-time payment rates in the UK and Singapore. These nudges are welfare enhancing when delayed payment is due to simple procrastination or absent mindedness. The agency gets payments on time, and the payer avoids late penalties.

However, there are people who cannot afford to pay on time because they are in financial distress, or have low incomes. They face a ‘cognitive tax’ from having to manage competing needs on a very limited budget,¹⁵ and may make poor financial decisions when nudged for immediate payment. They may, for instance, resort to borrowing money at high cost to pay for a bill when nudged, instead of seeking financial assistance, or they may prioritise paying creditors who nudge them the most effectively, rather than settle their most critical debts. While introducing nudges may improve specific policy outcomes, they could also unintentionally reduce the welfare of certain groups in society.

Every behavioural intervention that is narrowly focused on a specific policy goal may impose a small, but cumulative, cognitive tax on members of the public. A barrage of such nudges may gradually deplete our capacity to plan and make good decisions.¹⁶ Society accepts that constant exposure to commercial interventions may be harmful, which is why many cities, including Singapore, regulate billboards and outdoor advertising. But policy interventions are not as easily ignored as commercial advertising: they make use of official government channels to reach members of the public wherever they are. Policymakers must therefore regulate themselves; they must ask whether an intervention is truly necessary, given the potential cognitive costs on society.

Conclusion: Weighing behavioural interventions

Behavioural interventions are a powerful — even disruptive — tool for policymakers. But they do not relieve the policymaker from the need to carefully consider the costs and benefits of implementation. We suggest three principles for policy thinking in the age of behavioural interventions.

First, policymakers should act more cautiously when their interventions are designed to nudge people towards making the choice desired by the policymaker, rather than to just make better choices overall. Designs that leverage cognitive biases or impose mental costs to achieve behavioural change are particularly likely to generate welfare losses, because they infringe on people’s autonomy to make the choices that best suit themselves. The best interventions are those that preserve the autonomy to choose as far as possible, while using behavioural insights to help people make choices that are better for society as a whole.

Second, policymakers should evaluate behavioural interventions to consider the broader impacts on societal welfare, rather than just the narrow question of whether the policy ‘works as intended’. Societal welfare can be improved if policymakers use willingness-to-pay estimates to help to quantify the implicit costs of behavioural interventions, and use advances in analytics to target interventions on those who are the most likely to change their behaviour. For example, a more efficient method of implementing social norms interventions may be to focus on households or individuals who value conforming more, and reduce efforts to reach out to those who dislike conforming, and who would not conform in any case.

Finally, policymakers must exercise discretion and even restraint when considering behavioural interventions, particularly because they are relatively easy to implement. Greater openness and scrutiny of behavioural interventions from public agencies will be helpful; the commitment that many public agencies have made to share their findings publicly is an excellent start. In the longer term, we believe that creating an institutional process to coordinate and evaluate behavioural interventions across the whole of government will help to ensure that interventions continue to make everyone in society better off.

NOTES

1. See: G. Loewenstein and T. O’Donoghue, “We Can Do This the Easy Way or the Hard Way: Negative Emotions, Self-Regulation, and the Law,” The University of Chicago Law Review 73(2006): 183–206. Also, E. Glaeser, “Paternalism and Psychology,” The University of Chicago Law Review 73(2006): 133–56.
2. See: C. Sunstein, “The Ethics of Nudging,” Yale Journal on Regulation 32(2015): 413–50.
3. CPF contributions are divided into the Ordinary, Special, and Medisave Accounts. The Special Account is reserved solely for retirement. As of 2017, the Ordinary Account earns a minimum of 2.5% interest per annum while the Special Account earns a minimum of 4.0% interest per annum.
4. A. Lusardi and O. Mitchell, “Financial Literacy and Retirement Preparedness: Evidence and Implications for Financial Education,” Business Economics 42(2007): 35–44.
5. Ministry of Health Press Release, 20 March 2001, updated on 20 March 2015. Accessed at: https://www.moh.gov.sg/news-highlights/details/waiting-times-in-specialist-outpatient-clinics-of-restructured-hospitals-and-specialty-centres.
6. D. Hammond, “Health Warning Messages on Tobacco Products: A Review,” Tobacco Control 20(2011): 327–37.
7. See: J. Gruber and B. Koszegi, “Is Addiction “Rational”? Theory and Evidence,” Quarterly Journal of Economics 116(2001): 1261–303.
8. E. Johnson and D. Goldstein, “Do Defaults Save Lives?” Science 302(2003): 1338–9.
9. See J. Beshears, et al., “The Importance of Default Options for Retirement Savings Outcomes: Evidence from the United States,” in eds. J. Brown, J. Liebman and D. Wise, in Social Security Policy in a Changing Environment (Chicago, USA: University of Chicago Press, 2009), 167–95.
10. Although Singapore has adopted an organ donation opt-out default, a 2014 survey found that only 60% of those aged between 30 and 60 were willing to donate their organs. Since Ministry of Health statistics indicate less than 3% of the population has opted out, a sizeable minority of Singaporeans are now legally presumed to be organ donors even though they may not in fact be willing to donate. See: Emilia Tan, “Youth most open to organ donation after death: Poll,” Today, 18 October 2014, http://www.todayonline.com/singapore/youth-most-open-organ-donation-after-death-poll and reply from the Minister for Health to Parliamentary Question No. 720 by Dr Ahmad Mohd Magad on 28 February 2011, https://www.moh.gov.sg/.
11. Social norms are the unwritten rules of acceptable behaviour encapsulated in what people do, and what people think should be done. See: R. Cialdini, C. Kallgren and R. Reno, “A Focus Theory of Normative Conduct: A Theoretical Refinement and Reevaluation of the Role of Norms in Human Behaviour,” Advances in Experimental Social Psychology 24(1991): 201–34.
12. See: M. Hallsworth, et al., “The Behavioralist as Tax Collector: Using Natural Field Experiments to Enhance Tax Compliance,” Journal of Public Economics 148(2017): 14–31.
13. Peter Ong, “Gearing the Public Service for SG100,” Ethos 14(2016): 75–9.
14. See: H. Allcott and J. Kessler, “The Welfare Effects of Nudges: A Case Study of Energy Use Social Comparisons,” National Bureau of Economic Research Working Paper No. 21671, 2016.
15. See: A. Mani, et al., “Poverty Impedes Cognitive Function,” Science 341(2013): 976–80.
16. R. Baumeister, et al., “Ego Depletion: Is the Self a Limited Resource?” Journal of Personality and Social Psychology 74(1998): 1252–65.