Richard Thaler, of the University of Chicago, just won the Nobel Memorial Prize in economics for his contribution to behavioral economics — the subfield known for exploring how psychological biases cause people to act in ways that diverge from pure rational self-interest.
Policymakers are more likely to know him for a different reason. Together with his co-author, Harvard Law’s Cass Sunstein, Thaler is responsible for developing and popularizing the notion of “nudges” as a policy tool. Over the past decade, policymakers around the world have taken up Thaler and Sunstein’s ideas, setting up government nudge units and other programs intended to guide people toward choices that are in their best interests. Nudging has become fashionable.
Nudges can be used by both businesses and government to shape the behavior of employees, customers, and citizens. A classic nudge would be when a company automatically enrolls employees in a decent 401(k) plan — but lets them opt out of contributing, if they wish. The non-nudging status quo would require the employees to actively decide to sign up for a retirement plan. Studies have demonstrated that opt-out framing leads to higher enrollment.
Thaler and Sunstein argue that nudging is a win-win. Unlike traditional regulation, it doesn’t force people to make choices that they don’t want to make. Yet unlike a “laissez faire” approach it doesn’t assume that people should be left to make their own choices free of outside interference. Instead, their approach structures choices so that people are going to be nudged into making the choice that is probably best for them. In Thaler’s description:
if you want to get somebody to do something, make it easy. If you want to get people to eat healthier foods, then put healthier foods in the cafeteria, and make them easier to find, and make them taste better. So in every meeting I say, “Make it easy.”
So what’s not to like?
The problem — as Carnegie Mellon’s Cosma Shalizi and I have discussed elsewhere — is that government-by-nudging amounts to a kind of technocracy, which assumes that experts will know which choices are in the interests of ordinary people better than those people know themselves. This may be true under some circumstances, but it will not be true all of the time, or even most of the time, if there are no good opportunities for those ordinary people to voice their preferences.
Traditional forms of democratic policymaking rely on expertise too, but they have better correction mechanisms than nudgeocracy, since people who are sufficiently angry at a particular rule may have the incentive to complain and organize against it. Nudgeocracies, in contrast, are insulated from the feedback that would help them get things right.
This is even clearer if we call nudgeocracy by another name. Chris Hayes has written on Twitter about what he describes as the “hassleocracy” — the tendency to make something a hassle, so that fewer people will do it. Hayes was writing about the ways that Republicans make it harder for many groups to vote — mandating voter ID’s, and the like — but his description applies just as aptly to insurance companies doing everything they can to stop you from filing claims, or magazines that make it very easy to subscribe, but require you to notarize, and sign a form in triplicate, with the asservations of seven witnesses in red ink, if you want to stop the subscription from automatically self-renewing. (Thaler and Sunstein briefly discuss that example.)
Structuring choices to make things that you want people to do easy, and things that you don’t want them to do look like hassle carries over to contexts very different from the benevolent paternalism that Thaler favors.
Behavioral economics has inspired new approaches to policymaking
Nudgeocracy is a natural spin-off of Thaler’s economic arguments. Departing from standard microeconomics and game theory, behavioral economics takes it for granted that real-life decision-making is flawed in systematic ways, and that people often do not recognize what is in their best interest. (And even when they do know, they may find it hard to exercise appropriate self-restraint to make the right call.)
These insights lead to very different policy prescriptions than traditional economics. If you believe that human beings are perfectly rational, you can leave them alone to make the choices that will best satisfy their needs and self-interest. If, in contrast, you believe that human beings’ ability to know and act upon their interests is flawed, then you may want to help them make the choices that are best for them (so long as you are a benign administrator), even if those are not the choices that they would make unaided.
However, you face a potential ethical problem. By deciding what is good for people, rather than letting them decide for themselves, you take away some of their autonomy. You also run the risk of being wrong. Perhaps the choice that you think is best for the majority of people is obviously wrong for some minority of them.
This is why Thaler and Sunstein opt for nudging rather than benevolent dictatorship — or, less dramatically, for a system of governance in which legislatures pass laws designed by technocrats that force people to do the “right” thing. You do not compel people to make the right choices. Instead, you set up their choices so that they will find themselves guided to make the “right” choice unless they have strong alternative preferences. Creating a default option, as in the 401(k) example, is one option.
All of this sounds innocuous and even compelling, given that even neo-classical economists admit, when they are forced to, that human beings are not perfect calculating engines.
The “choice architect” does not always know best
Indeed, there are many circumstances under which nudges are a good idea. But the fad for nudgeocracy has hidden implications. Thaler and Sunstein describe the philosophy that underlies nudging as “libertarian paternalism” — libertarian because it lets people make the choices that they want to, paternalist because it provides them with a father’s guiding hand. Behind nudgeocracy lies the assumption that daddy knows best.
For Thaler and Sunstein, daddy is a “choice architect” — a skilled and intelligent technocrat who uses good data, good social science and his own intelligence to figure out what people would really want to do, if only they were as smart and well informed as the choice architect.
The problem is that daddies — consider those stern yet kindly men who smoke pipes and deliver shopworn aphorisms in in deep and manly voices — may have good intentions but don’t necessarily know best at all. They themselves may not have good information about what their charges want, what they should want, or how their charges will respond to nudges. Their understandings of their underlings or users may be skewed by preconceptions or poor information.
In one study, a power company tried to “nudge” its customers into consuming less energy, by telling them how their power consumption compared to their neighbors, and telling power hogs that they had room to improve. The nudge worked as intended for environmentalist Democrats, but non-environmentalist Republicans instead took it as license to increase their power consumption.
Of course, the choice architect has tools to try to remedy misinformation and misperceptions. Sometimes Thaler and Sunstein suggest that the choice architect can figure out what people should choose by simple introspection — “Libertarian paternalists would like to set the default by asking what reflective employees … would actually want.” Sometimes, they argue that the choice architect can carry out experiments, varying choices to see what consequences they have for human behavior. Choice architects could also turn to other sources of information such as surveys, or focus groups.
It is not clear, however, that choice architects will get good information from any of these sources. After all, choice architects are almost by definition going to be political, social, or business elites, and people are very often unforthcoming to those who can control their choices, fearing retaliation if they say the wrong thing.
Even when people provide honest information, it is not clear that the choice architects will have much incentive to listen to them if they say things that the choice architects don’t want to hear. Thaler and Sunstein do not really talk about whom the choice architect is supposed to be responsible to, perhaps because they want to talk both about the private sector (in which the architect must ultimately be responsible to owners or shareholders), and the public sector (in which the architect is a bureaucrat, and therefore responsible, directly or indirectly, to politicians). This means that they have little to say about the role of accountability in ensuring that choice architects accept new information, where it is available, and use it well.
Hard-and-fast rules can lead to democratic pushback. Technocratic “nudges” may not.
The above is another way of saying that nudgeocracy has a big weakness: It is not very good at receiving and incorporating feedback. Thaler and Sunstein have a lot to say about how choice architects can provide better feedback to ordinary people, but they have little to say about how ordinary people can provide feedback to choice architects. This is hard to see if your starting point is economics. While markets provide all kinds of feedback, standard micro-economic models assume that market actors have complete information over all the possible ways that other actors might respond to them; from that perspective, the idea of a genuinely unexpected response is a big stretch.
It is much easier to understand if you start instead by comparing nudgeocracy to traditional democracy. If “libertarian paternalism” sounds oxymoronic, so too does the way that traditional democracies run: through coercion. A democracy is a society in which the choices of a majority of citizens are binding on everyone else. These choices may be made directly or indirectly, and through a variety of mechanisms. The makeup of the majority may also vary, but what does not vary is that the choices are binding — those in the minority are forced to abide by them. This is why democracies are coercive — and why some libertarians dislike them. You do not have the option not to pay taxes; nor can you opt out of Social Security.
There are never-ending debates over the benefits and disadvantages of democracy. However, relatively little attention is paid to the informational advantages that coercion provides in a genuinely democratic setting.
Imagine that you live in a democratic system that imposes some new rule that requires you to do something that you don’t want to do — say, wear a seatbelt in your car. This rule will be visible to you, since you will be punished if you break it. If you are very unhappy with this rule, you can use your democratic voice. If you can persuade enough of your fellow citizens, you can then push to have the rule overturned.
That’s a natural feedback mechanism. Democratic coercion may sound like a contradiction in terms, but it builds a feedback loop through which suboptimal rules can get fixed.
Nudgeocracy, in contrast, lacks such a loop. First of all, the rules governing choice are likely to be unobtrusive. Much of the time, people won’t even be aware that they are being guided to a particular choice. In contrast to what happens in a coercive regime, people will be less likely to know to complain, or to mobilize against choice architectures that guide them to decisions that aren’t the best decisions they could make. Even when they do complain, it is not clear that their complaints will make a difference, since the choice architects might or might not be publicly accountable.
Of course, even the best democracies are not always responsive. (Notoriously, it is easier to mobilize against rules that have high costs for a small number of actors — like anti-pollution rules for power plants — than rules that have low costs for a very large number of actors.) Nonetheless, the closer that political systems approach true democracy, the more responsive to the public they are going to be. Nudgeocracy, even in its ideal form, lacks such accountability, and instead relies on a variety of more or less adequate technocratic substitutes.
In short, we may expect that bad laws and regulations — “shoves,” we might say — will generate protest or noncompliance. A bad nudge, in contrast will either escape notice, or be sidestepped by those who do notice it. We can expect such a system to be worse at self-correction than a system that combines traditional coercive rules with democratic accountability.
Both moderates like Steve Teles and progressives like Suzanne Mettler have argued that stronger nudging is likely to lead to weaker citizenship. Mettler is worried that it threatens to turn citizens into consumers. For democratic government to be of the people, by the people, for the people, people have to have a clear view of what government is doing, she argues. Nudging, in contrast, may produce efficient policy goals but leaves citizens uninformed and without real agency.
She singles out policy measures such as a stealth Obama tax cut that was intended as a stimulus to boost the economy. This cut was designed to be unobtrusive, so as to nudge people to spend more instead of saving. This may have made it more effective, but it also made it politically invisible to citizens. Teles agrees with these criticisms, and in a minatory passage, foresees “a prostrate citizenry—one insufficiently mobilized to actually challenge deep structural inequalities, let alone to explicitly recognize who has leased its consent.”
Advocates of “nudges” have thought more about economics than about politics
This does not mean that nudging is useless. It does suggest that it should be used only under limited circumstances. One can imagine two circumstances where nudging could work.
One is where there is some default decision that is a) non-obvious to people (or at least difficult for them to stick with even if it is obvious), b) the best plausible decision when you think about it more carefully, and c) does not have any complicated feedback effects. Here, it may be reasonable to guide people toward making the “right” decision, although choice architects still have to be careful.
This description strongly echoes what advocates of nudges say about when their approach should be deployed. But what they do not recognize is that the scope for such nudges is much smaller than it appears at first to be. In a profoundly interconnected society, even apparently simple choices can have complex and unpredictable consequences. As scholars like Scott Page (and Shalizi and I) have argued, technocratic experts are likely to be less able to foresee and account for those consequences than groups that draw on a wider set of diverse perspectives and forms of background knowledge.
In short, for complex issues, democracy (when it can usefully draw upon diversity) is likely to trump technocracy (which draws on more sophisticated, but narrower expertise). Thaler and Sunstein, for example argue that nudges can help us deal with climate change, since global climate is the “outcome of a global choice architecture,” where market pressures and the lack of feedback mean that people often make collectively bad choices. They suggest that experts can design a better choice architecture based on incentive schemes.
But as the different responses of Republicans and Democrats to power saving schemes illustrates, apparently straightforward incentives can backfire, or interact in complex and unforeseeable ways with other factors shaping human behavior. Technocracy may end up being as much part of the problem as the solution
The second is where there are a variety of other “choice architects” — possibly in the private sector — already working hard to shape people’s decisions to selfishly benefit themselves rather than the people making the choices. For example, investment planners who receive commissions for selling specific investment plans have an obvious economic incentive to shape their clients’ choices so as to maximize their own commission income rather than the long-term value of the investments. In such situations, “benign” choice architecture decisions can help keep people from making bad long-term choices.
Here, though, again, there are difficult problems. Purportedly “protective” structures can become magnets for bad actors who will want to influence, infiltrate and reshape them to malign ends. At this very moment, the Trump administration is turning the machinery of the regulatory state to the Republican Party’s ends — for example, by frustrating efforts to guide people toward good health care plans. It is easier for bad actors to undermine or subvert nudges (which are typically semi-formal policies) than to undo binding coercive rules (which need to undergo a complex review process and possible court challenges before they can be changed).
In short, nudge theorists have failed to consider the political consequences of their work. This is understandable, given that their ideas emerged from intramural fights within economics. The argument that Thaler and Sunstein wanted to start — and, in a sense, are still conducting — are with libertarian-leaning economists who do not understand why rational self-interested beings would need others to tell them what they ought to do.
But the truly hard questions don’t come from market fundamentalists. The greatest challenges to nudge theory comes from people who want to know what nudges mean for democracy, for accountability, and for relations between citizens and their governments. That’s a conversation that Thaler and Sunstein have at best started to engage with.
Henry Farrell is a professor of political science and international affairs at George Washington University. He is an editor of the Monkey Cage.
The Big Idea is Vox’s home for smart discussion of the most important issues and ideas in politics, science, and culture — typically by outside contributors. If you have an idea for a piece, pitch us at firstname.lastname@example.org.