Policymakers are keen on encouraging people to recycle, conserve resources, and reduce pollution. One popular method is to automatically enrol individuals in programmes—such as renewable-energy programmes—in order to boost participation. In some places, for example, households must actively choose to opt out when they sign a new energy contract, or else they will receive renewable energy. However, automatic enrolment is unlikely to always be the optimal policy. New research suggests that whether a policy breeds guilt or resentment helps determine if it ultimately will be effective at increasing participation or not.

Do you bring your own reusable bags when you do groceries? If you do: have you ever felt an urge or inkling to reward yourself for your efforts to reduce your environmental footprint? A recent study of consumer behaviour in a supermarket by Bryan Bollinger and Uma Karmarkar found that people who brought their own bags were more likely to buy ice cream, candy, and dessert.1 Psychologists refer to this phenomenon as “moral licensing”, which describes the idea that people tend to reward themselves when they feel good about something that they have done.

While treating oneself with sugars and fat for bringing textile bags to the supermarket is a benign act and its consequences hardly pernicious (except maybe for the waistline), moral licensing has been found to have much more problematic effects in other instances. A study of a water-conservation campaign in Massachusetts, for example, found that while the campaign was successful in reducing the targeted households’ water consumption, the same households simultaneously increased their average electricity consumption.2 The most likely explanation is that the people who reduced their water usage felt so good about their eco-friendly actions that they allowed themselves to be more lavish with heating and not as careful to turn off electric appliances.

Is guilt a stronger motivator than reward?

Although it may sound like a glum proposition, making people feel bad can make them do good things. Complacency can lead to a self-rewarding behaviour that is counterproductive. Guilt, by contrast, has repeatedly been found to be a powerful motivator for social change.3 An experiment by Aristeidis Theotokis and Emmanouela Manganari on whether or not to reuse towels when staying in hotels, for instance, revealed that feelings of guilt were strongly related to choosing to conserve the environment over comfort.4 Participants who felt more guilty were more likely to say that they wanted to reuse towels. For policymakers who are interested in promoting socially desirable outcomes—such as environmental conservation—it could therefore make sense to try to induce some sense of culpability among citizens to prod them to do the “right” thing.

Policies can prompt guilt in several ways. One of them is to use so-called “choice architecture”, which refers to the way that choices are presented.5 Including nutritional labels on restaurant menus is one example of this; showing the number of calories in a bucket of fried chicken does not alter a customer’s ability to choose whether or not to buy one, but it may nudge him or her towards a healthier option. Another example of choice architecture is to enrol people in a scheme by default and allow them to opt out—automatic magazine subscription renewal provides a simple illustration of this.

Changing how we choose

Enrolling people by default is a popular policy and it stands in contrast to “opt-in” policies (people have to explicitly make the choice to enrol or else they will stay out) and “active choosing” (people are either forced or prompted to make a choice). Opt-in policies are common. When you open a checking account you may also be offered a savings account, but if you do not explicitly state that you want one, for example by ticking a box on a form, nothing will happen. Active choosing means that no alternative has been selected by default and so a choice has to be made. This can be forced upon people for example by requiring them to fill out a form about whether they want to donate organs or not when they visit the Department of Motor Vehicles to pick up a driver’s licence.6 In order to get their licence, they have to actively choose to check either the yes box or the no box on the organ-donation form.

Automatic enrolment has been found to have sometimes surprisingly large effects.7 This is in part because of inertia and a tendency to procrastinate, which means that most people will stick with the default rather than investing effort in changing.8 But another reason is that automatic enrolment can make people feel guilty. Indeed, this is what Theotokis and Manganari found in their towel experiment.9 Some of the participants were automatically enrolled in a fictitious towel-reuse programme whereas others were merely given the option to reuse their towels. Notably, the former group felt more guilty about not reusing towels than the latter.

These results make sense. A hotel owner who wants to reduce water consumption might automatically enrol guests in a towel-reuse programme and then allow them to opt out. If every guest who stays at a hotel is assumed to want to reuse their towels, one thus has to actively opt out to get a new towel by specifically asking for it. This can easily induce guilt since most people are aware that this will increase water usage and that the norm that the hotel has set is to reuse towels.

So should we just automatically enrol everyone in programmes that promote environmentally-friendly behaviour? Not necessarily. First of all, there may be ethical concerns about automatically enrolling individuals in programmes that cost a lot of money. Those who are sceptical of using choice architecture to promote certain behaviours see it as manipulation. I think this is an unfortunate misperception. Whenever a choice is presented it will by definition be framed in one way or the other. So the only question is whether the person who designed the choice you face (e.g. in a food court, a marketplace for health insurance, or an offer to sign up for a rewards card) is steering you in a certain direction deliberately or arbitrarily. However, this does not mean that framing choices is always unproblematic; I would personally be concerned if poor people were automatically enrolled in very expensive programmes, for example.

In addition, automatic enrolment may backfire. Several studies suggest that automatic enrolment, in fact, can lead people to opt out from a programme that they otherwise would have chosen had it not been for the policymaker’s “nudge”.10 An example is a programme that aimed to reduce electricity consumption in California. In a paper, Dora Costa and Matthew Kahn presented suggestive evidence that some of the participants who were automatically enrolled in the programme not only opted out, but also increased their electricity consumption.11 Within these households, the scheme appears to have had a negative effect, leading to the opposite result to what was intended.

When policies lead to reactance

This form of negative response by individuals against constraints placed on their freedom is often referred to by psychologists as “reactance”.12 Of course, automatic enrolment with an offer to opt out is, unlike a mandate, not a true constraint on an individual’s freedom, because the decision is still yours to make. But as with so many other things in life, what matters most is perception. And so if people perceive a policy as interfering with their freedom to choose for themselves, it is not unreasonable to expect some of them to show a form of resentment.

To illustrate what I mean by reactance against policies that attempt to promote a certain outcome, I will use another example unrelated to environmental issues. In New York City, many taxis provide tip suggestions by default. The way it works is that credit card touchscreens will show the amount due at the end of the ride with suggested tip amounts (e.g. 15%, 20%, and 25%) that the passenger easily could click on in order to tip (some restaurants have also adopted this system of default tip suggestions). Because some taxi companies provide lower suggested tip percentages than others, it is possible to discern what effect higher default tips have on the likelihood of tipping. An analysis of 13 million taxi rides in New York found that although higher default tips led to higher average tips (as one might expect), the likelihood of tipping was lower.13 In other words, the participation rate was lower. The most plausible interpretation of the data is that when some customers saw the higher suggested tipping amounts, they displayed a form of reactance and chose not to tip at all. These customers would most likely have given a tip either if the defaults were lower, or if there were no default suggestions at all

The key takeaway here is that reactance can act as a counterpoint to guilt. Of concern to decision-makers is therefore to what extent the policies they design will cause guilt or reactance. These two emotional responses may ultimately have an important impact on whether the desired outcome will be achieved or not. Focusing specifically on the previously mentioned choice-architecture policies, an important question is whether to automatically enrol people or not based on the potential effect that such a rule may have on guilt and reactance.

A closer look at the effects of guilt and reactance

Currently we do not yet know much about the relationships among choice architecture, guilt, and reactance. To try to shed some light on this issue, I conducted an online experiment together with Cass Sunstein, Professor at Harvard Law School and co-author of the bestselling book on behavioural economics - Nudge.14 We were specifically interested in how different ways of presenting the same choice to people could affect their interest in joining a hypothetical green-energy programme.

For the experiment, we recruited 1,245 Americans from Washington D.C. and all of the states with the exception of Wyoming. They were randomly assigned to read short vignettes, or prompts, about an opportunity to enrol in a programme that would provide more environmentally-friendly energy than the programme that they currently had. The vignettes were of three different types: “green-energy default” (respondents would automatically be enrolled in the new green programme unless they opted out), “standard-energy default” (they were offered to join the green programme, but would keep their current energy provider if they did nothing), or active choosing (they were to assume that the government had cancelled their energy plan and they now had to choose either the green or the standard provider). After reading one of the vignettes, the participants were asked whether they wanted to enrol in the renewable or the standard-energy programme; whether they felt guilty about not choosing the green-energy programme, on a scale from 1 to 6; and whether they approved of the policy that they had been assigned to (i.e. green-energy default, standard-energy default, or active choosing), on a scale from 1 to 7.

Based on the existing literature, one might expect the green-energy default to cause the strongest feelings of guilt about choosing conventional rather than renewable energy, which will help produce the highest enrolment rate in the green programme. But we suspected that neither the former, nor the latter outcome would necessarily be true. In particular, we hypothesized that a crucial question would be whether the green-energy programme would cost more than the standard-energy programme. If the cost is the same, then it is easy to see how one would feel guilty about not doing what is good for the environment, and what most people would consider the “right” choice. By contrast, if the renewable-energy provider charges more, it is likely that people will feel less guilty about not enrolling since they would have to pay more for it. In addition, some may also think that automatic enrolment into a programme that costs more than what they currently pay is too strong of a nudge, which may cause reactance and reduce the enrolment rate. We therefore separated the participants already assigned to the three policies (green-energy default, standard-energy default, and active choosing) into two conditions, one where the cost of the renewable-energy program would be identical to what they currently paid and another where the cost of green energy would be an additional $25 per month, leading to a total of six groups that each read a unique vignette.

In accordance with previous studies, we found that the standard-energy default was by far the least effective in terms of leading to the desired outcome of enrolment to the green programme. Participants assigned to this policy (regardless of whether the renewable-energy programme came at an additional cost or not) were less likely to be interested in signing up for the renewable energy programme than those assigned to active choosing or green-energy defaults. As expected, they also expressed lower levels of guilt about choosing the conventional energy source, and regression analyses revealed that the level of guilt was significantly correlated with the likelihood of enrolling.

With respect to the other two policies, the green-energy default and active choosing, it came down to the cost factor. When the renewable-energy programme cost no more than what they currently paid, respondents were just as likely to opt for green energy when they were forced to make an active choice as when they were automatically enrolled. However, when the renewable energy cost $25 extra per month, there was a substantial and significant difference in enrolment rates between the two policy options. With an additional cost, active choosing led to an enrolment rate of 56% and the green-energy default just 39% (which, it should be noted, was still higher than the 26% yielded by the standard-energy default).

What explains the difference in effectiveness between active choosing and green-energy defaults when renewable energy costs extra? Sunstein and I reckon that it has to do with the effects of guilt and reactance. With the additional cost imposed, there was a clear difference in expressed guilt between participants assigned to the two policy conditions. Those assigned to the green-energy default were less likely to feel guilty about not enrolling, which suggests that some of them felt resentment towards being automatically enrolled into a programme that had them pay more for their electricity. It therefore seems plausible that some participants displayed reactance and felt less guilty about opting out of the renewable-energy programme.

When we looked at the approval rating of each policy, we found further evidence in support of this hypothesis. Focusing on differences and similarities between the green-energy default and the standard-energy default, we found that at an individual level, approval was positively related to the likelihood of enrolling. This makes sense. Presumably those who think that it is good to have the choice of joining a renewable-energy programme are also those who are most likely to sign up. And, as previously noted, the green-energy default produced a higher enrolment rate than did the standard-energy default. Yet we also found that when the renewable energy cost extra, and only then, the approval rating for the green-energy default was lower than for the standard-energy default.

What is going on here? If approval and likelihood of enrolling are positively related, and if the green-energy default is substantially more effective (39% vs 26% in enrolment rates), how can the approval rating of the green-energy default be lower? One potential explanation is that there may have been a small, but significant, minority that opted out from the green-energy default while giving the programme the lowest possible approval rating because the idea of automatic enrolment into a programme that cost extra made them upset. This could then explain why the approval rating of the green-energy default was lower even though it was, at the same time, more effective than the standard-energy default. We found some evidence in support of this hypothesis.

Policymakers should frame choices with care

Our results have two main implications. First, in accordance with previous studies, how choices are framed will affect both how people feel about the options and how they ultimately choose. Second, when the objective is a higher enrolment rate, imposing an additional private cost (e.g. an additional cost to sign up to a green-energy plan) seems to impact whether framing a choice in terms of automatic enrolment or forced active choice is most effective. If there is an additional cost associated with a pro-social behaviour, then active choosing can be the more effective policy because it can lead to higher levels of guilt and lower levels of reactance than a policy of automatic enrolment. This is an important result that challenges conventional wisdom regarding the superiority of automatic enrolment. As many governments around the world are trying to cut their budgets, the popularity of choice architecture is likely to grow as altering the way choices are framed is cost-effective. It is therefore crucial that policymakers get this right.