The Institutional Consequences of Nudging – Nudges, Politics, and the Law

Abstract

In this article we argue that a widespread adoption of nudging can alter legal and political institutions. Debates on nudges thus far have largely revolved around a set of philosophical theories that we call individualistic approaches. Our analysis concerns the ways in which adherents of nudging make use of the newest findings in the behavioral sciences for the purposes of policy-making. We emphasize the fact that most nudges proposed so far are not a part of the legal system and are also non-normative. We propose two ideal types: “law-as-normative” and “law-as-instrumental”, that allow us to understand and evaluate the relation of nudges and the law. We stress the importance of law as a safeguard for the possible negative consequences of nudges and conclude with proposals that could complement nudging policies.

1 Introduction

What are the institutional consequences of the widespread adoption of nudging policies? Does the application of findings in the behavioral sciences through nudging policies influence our understanding of the law? How does law look like in a nudge world?¹ Our aim is to provide answers to these questions. We show that the consequences of nudging policies for law and political institutions have not been discussed sufficiently.

In this article we reconstruct claims formulated by the proponents of nudging, analyze their policy proposals, and demonstrate implications. We propose two ideal types “law-as-normative” and “law-as-instrumental” in order to understand and evaluate the relation of nudges and the law. What we do not intend to do here is to discuss (so-called) libertarian paternalism. The next two sections explain why.

2 The Individualistic Framing of the Debate Thus Far

Thaler & Sunstein (henceforth T&S) can be credited for developing and advocating a comprehensive program to implement behavioral findings into public policy (Thaler and Sunstein 2001; Thaler and Sunstein 2008; Sunstein 2014a; Sunstein 2014b; Sunstein and Thaler 2003a; Sunstein and Thaler 2003b; Sunstein 2000). They defend nudging as a “liberal paternalist” tool of public policy that allows institutions, by using findings derived from the behavioral sciences, to steer people towards welfare-promoting outcomes while not limiting their freedom of choice.

Most frequently discussed in the current debate on nudges are the complex relations between the concepts of autonomy, freedom and rationality – how they are understood, how they change as a result of nudging policies, or how they should be redefined. Philosophical debates focus almost exclusively on moral limits to nudging and the defensibility of libertarian paternalism (Mitchel 2005; Veetil 2011; Bovens 2008). Critics such as Hausman and Welch (2010) warn us that “shaping people’s choices for their own benefit seems to us to be alarmingly intrusive” (p. 131), and are concerned that certain instances of nudging exploit human weakness. Some investigate whether or not nudging violates liberal principles (Grüne-Yanoff 2012), or look at how nudges impact the “fundamental rights of citizens to freedom of expression, privacy, and self-determination” (Alemanno and Spina 2014, p. 431). Others respond to these concerns by modifying the concept of nudging (Saghai 2013), but abstain from considering institutional implications of their rescue attempts.²

The perspective on nudges that is present in most debates thus far we call the individualistic approach. Such an individualistic perspective on nudging holds that the relevant attributes of a nudge can be described by looking only at the interaction between a nudger and a nudgee. In contrast, an institutional perspective challenges this and also considers factors external to the interaction, particularly the legal and political institutions in which a nudge is embedded. The work on nudging by T&S is interspersed with remarks about the institutional order that is favored by them; here, we start from the premise that a coherent institutional perspective is missing from the work of T&S, as well as from subsequent debate. We argue that when looking at the impact of nudging we should not look only at what a nudge does or does not do to a nudgee (e.g. whether it is paternalist or not), but also whether the adoption of nudging as a policy impacts institutional structures (which then in turn impact individuals). We argue that it does, but that this cannot be accounted for using the individualistic perspective prevalent in the literature. We further argue why we think the current debate on nudging should be complemented by a more institutional orientation; we then propose safeguards for the implementation and design of nudging policies.

3 What are Nudges for? The Politics of Nudges

T&S argue that nudges help people to achieve outcomes that they would reasonably want themselves. Yet we can distinguish two different objectives that nudges fulfill (or, put differently, two rationales why a planner improves the choice architecture, i.e. the context within which people make decisions). First, some nudges aim to improve individual decision-making or to steer the behavior of people “into rational directions” as judged by individuals. Second, some nudges aim to achieve desirable social outcomes, irrespective of their effects on the capacity of individuals to make informed decisions (for example, default opt-out for organ donations).³ The debate on paternalism concerns those nudges that aim to optimize individual welfare, understood as the satisfaction of individual preferences that are in people’s best interests. What our distinction makes clear is that the function of nudges in the eyes of nudging advocates is not solely to address some form of individual behavioral failure or deficiency and to enable people to get what they might want. An ulterior motive is to make policy more efficient and achieve states of affairs that are socially desirable in the eyes of nudging advocates, such as, for example, public health. There should be an institutional (legal) framework that subjects such policy instruments to public scrutiny: the goals of policymakers, which concern public and societal objectives, should not be concealed. Such an institutional framework, however, is also necessary for nudges of the first kind. The need for public scrutiny is justified differently in both cases. Regarding the first kind, policymakers sometimes make mistakes when identifying individual preferences and implementing nudges that aim to increase individual decision-making capabilities. This may cause harm to individuals. Nudges of the second kind, which directly aim to achieve public or social objectives, should undergo public scrutiny because these objectives should not be decided by policymakers alone in a democracy.

The problem here is also that nudging proponents provide only a selective reading of the scientific literature to justify their policy proposals. This can best be demonstrated with the concept of bias. The concept is consistently relied upon by nudging enthusiasts, despite there being an ongoing, foundational, debate in the behavioral sciences on whether a) certain patterns of behavior are biases or b) whether debiasing is possible.⁴ Adherents of nudging are also not consistent in their treatment of biases. At times, nudges should liberate people from biases – yet on other occasions, biases are actively used to serve policy goals. Sometimes they propose to take advantage of a bias that stems from availability heuristics⁵ in order to influence people to make a healthy choice (Sunstein 2013, p. 130–134). Sometimes they propose to eliminate a bias so that people are “helped” to maximize their self-interest or welfare (Thaler and Sunstein 2008, pp. 142–144; Jolls and Sunstein 2006, pp. 207–211). These responses to biases are not just innocent comments on the state of the literature in the behavioral sciences: they shape how legal and political institutions are to be designed. Institutions that attempt to eliminate biases differ from institutions which take advantage of biases. The interpretation and selection of behavioral sciences and their translation into policy has institutional consequences. Therefore it is important to debate the consequences of reliance on a selective reading of the scientific literature and to scrutinize proposals for nudges that rely on it.

4 Nudges and the Law

4.1 Nudges and the Legal System: Apart, and Estranged

In order to grasp the impact that nudges have on institutional configurations in society, we propose to distinguish between nudges introduced as a part of a legal system and those which are extra-legal, i.e. not a part of a legal system. We understand here the legal system as the complete body of codified legal norms in a given society and state.

Using this distinction we quickly see that default rules are one of the few,⁶ yet widely and controversially discussed, examples of nudging policies that are part of a legal system. Examples are the so-called “opt out” options,⁷ such as retirement savings programs for employees, or default rules for regulating organ donations. Default rules are supposed to influence people’s behavior effectively because of the status quo bias (akrasia) which is the tendency to place higher values on options perceived as status quo, explained as a consequence of loss aversion (Tversky and Kahneman 1991).

Other nudging policies such as simplification, uses of social norms, increases in ease and convenience, disclosure, warnings, precommitment strategies, reminders, eliciting implementation intentions, informing people of the nature and consequences of their own past choices, are not part of a legal system.⁸ This means that they are not codified and thus that the behavioral change of the addressees of nudges is not effected by law.⁹

We believe that a reason for why most current nudging policies are not part of a legal system is this: decisions about implementing nudges are driven by the effectiveness of the instrument. Nudges, introduced not as a part of a legal system, and not as legal norms, are less visible and hard to contest. They do not require political support, parliamentary procedure and debate, and, thus, enable policymakers to influence people’s behavior more quickly, more effectively, and without putting effort into cumbersome legislative and deliberative processes.

4.2 Two Ideal Types

We distinguish two ideal types of conceiving of the law: “law-as-instrumental”, and “law-as-normative”. Following Weber, we understand ideal types as concepts which highly stylize some aspects of the analyzed phenomena.¹⁰ We construct ideal types for heuristic purposes in order to analyze and understand what can be learned through ideal types of law about nudges, as well as for evaluative purposes in order to assess nudges and to understand to which ideal type they approximate.

We understand the ideal type “law-as-normative” as a conception of law with the following features: First, law in this ideal type is characterized as a requirement to behave in a particular way. Second, addressees of a legal norm (agents) respond to law after they recognize a requirement to behave in a particular way, prescribed by a legal norm. They treat this requirement as a considered reason for action (or not – in which case the agent does not conform to the law). Third, law needs to be cognitively accessible. This is a necessary requirement for effective responding to the law.

We understand the ideal type “law-as-instrumental” as a conception of law with the following features: First, law in this ideal type is characterized as an instrument of influencing decisions not by requiring from agents to behave in a particular way, but by changing the context of individual decision-making. Second, agents respond to law through reaction to different non-legal factors (perception of risk, emotions, interests). Third, law needs not be cognitively accessible to be effective. In other words, the non-cognitive reaction of an agent is a sufficient condition for effective responding to the law.

Our ideal types should be useful for comparing and contrasting ways of influencing people’s behavior. Therefore our ideal types include not only characteristics of the law, but also of the agents’ responses to law and the conditions for law’s effectiveness.

Two things should be stressed here. First, we do not engage in philosophical debates on the normativity of law. We merely say that law that is normative (binding) provides reason for action, and that this is one of the characteristics for one of our ideal types. We do this without defending a theory of normativity that may underlie it. We simply examine what acceptance or elimination of normativity of law leads to for our understanding of agency, as well as for a legal and political system. We are aware of philosophical difficulties related to explaining in what sense law can be a reason for action. But we think that the idea of law being a reason of action is sound for any approach which aims at understanding a human action as driven by law (legal norms).

Second, our ideal type “law-as-instrumental” differs from typical instrumental accounts of the law.¹¹ These accounts do not exclude the possibility that law can be a norm (or can be normative) while also being an instrument of achieving a certain social, policy, or economic goal. The ideal type proposed by us, however, represents a non-normative and non-cognitive way of influencing people’s behavior by law- and policymakers. It is non-normative because the agents’ behavior is not influenced by legal norms (requirements to behave in a particular way) but by different factors. It is non-cognitive because the agents’ behavior, in order to be influenced, does not have to be a conscious, deliberate response.

We argue that nudges approximate an ideal type “law-as-instrumental”. By arguing this we emphasize that nudges represent a novel approach to regulating behavior in law and policy settings, since nudges are non-normative and they impact people’s behavior in a non-cognitive way. Nudges are non-normative because, first of all, most of them are not a part of a legal system and hence cannot provide a reason for action through the legal requirement to behave in a particular way. But even if some of them are a part of a legal system, such as the above-mentioned examples of default (opt-out) rules, they do not impose a requirement to behave in a particular way, but affect situations of agents without requiring their actions.¹² Nudges influence behavior in a non-cognitive way because for their effectiveness, non-cognitive reactions to features of a “choice architecture” are sufficient.¹³

Influencing behavior in a way which approximates an ideal type “law-as-instrumental” represents a view of human agency where individuals are held by law- and policymakers to act out of emotions, sentiments, automatic reactions. But these individuals never deliberate about norms, whether they (should) bind them, and, thus, to give them reasons to obey norms and accept stipulated requirements. The result of this is first, an abandonment of an idea of influencing people’s behavior through and by legal norms, and second, a deprivation of the chance for society to engage in self-legislation.¹⁴ Law which approximates this ideal type impacts behavior due to accommodation and exploitation of behavioral regularities. In such a nudge world there is no need for publicly accessible, codified laws in order for law-and policymakers to effectively influence people’s behavior.

An alternative understanding of law (“law-as-normative”) leads to a view of human agency where law is a reason for action and where behavior can be influenced by legal norms. Law which approximates “law-as-normative” may hence fulfill a dual function of aiming at motivating individual behavior, as well as at providing a social ideal towards which action should strive. Thus this understanding of law is also compatible with self-legislation and democratic deliberation over the rules that society gives itself. Here law, in order to fulfill its functions, has to be publicly accessible and visible.

5 Institutions and the Nudge World

The legal system – a body of codified, publicly accessible and debated legal norms – is a social institution which can make nudges more visible and accessible and thus minimize a non-normative and a non-cognitive impact of nudges on citizens’ behavior. Therefore, in the next section, we propose safeguards which should relate nudges to the legal system in a visible, recognizable way. This connection of nudges with the legal system should make agents reflect on the influences they are subject to, as well as being able to debate the (legal) preconditions of such influences.

Critical remarks emphasizing the threat of nudges to policy visibility, institutional transparency, and public deliberation have already been formulated by some scholars (Conly 2013; Glaeser 2006; Hansen and Jespersen 2013). Sunstein, in his response to this kind of critique, remarks that, since nudges are often highly salient, they are transparent and debated in public (Sunstein 2014a). He however also notices that nudges are especially worrying when they rely on unconscious cognitive processes, or emotions, or both (the so called System 1). Sunstein argues that since, in our decision-making, we cannot avoid being influenced by System 1 processes, therefore “so long as the initiatives are made public and defended on their merits, nudges should not be ruled off-limits merely because they work as a result of the operations of System 1” (Sunstein 2014a, p. 151).

This reference to nudges being public should however be discussed more critically. An important but under-appreciated aspect of T&S’s proposals is their stance on the publicity of nudges. In “Nudge”, T&S refer to John Rawls’s “publicity principle” as one of their “guiding principles” (2008, p. 244), arguing that this principle “is a good guideline for constraining and implementing nudges” (2008, p. 245).

T&S’s appropriation of Rawls’s publicity criterion might be a little misguided given that Rawls did not intend that it apply to all policies, but mainly to the “first principles” of justice (Rawls, 1999). Also, Rawls’s criterion relies squarely on an ideal of morally and rationally autonomous individuals that are owed justification for the laws they are subjected to. Citizens “must have knowledge of the moral bases of coercive laws”, and of the “real reasons for their social and political relations”; it is publicity of the first principles which should give “democratic citizens a common basis for political argument and justification” (Freeman 2007, p. 187). But attaining such knowledge is harder to achieve than T&S assume – on the contrary, nudges obscure this knowledge if instituted without safeguards and without being visibly brought into the legal system.

And what happens if nudges are not made public, and what if they are not part of a legal system? Can we rethink and propose a legal and institutional design which could prevent and protect us from such a possibility? In the next section we present our proposals.

6 How (not) to Nudge

There are very few safeguards to nudges today. Here, we would like to analyze some examples of popular nudges and propose ways in which legal norms could complement them.

Default rules (organ donations, retirement savings programs) are a part of the legal system. Obviously, default rules were present in law long before T&S started promoting opt-out rules. They exist especially in contract law, which regulates relationships, rights, and obligations between parties of contractual agreements in a way directed by law (unless parties decide to regulate their legal positions differently).

In the case of “nudging default rules” the choice of an agent has been already determined by a law-maker, the only available choice is to opt out. With nudges, the scope of choice is narrower than in the case of e.g. contract default rules¹⁵ - one cannot choose whether one prefers the proposed pattern, or whether one wants to opt-out; one can only opt-out.

We propose that nudging in the case of default rules should be complemented by a mechanism of the following kind: every time citizens are nudged towards a choice, via default rules which policy makers regard as leading to socially desirable outcomes, rules should stipulate that citizens ought to be required to make an active choice between default or opt-out option.¹⁶ Our second proposal states that in order to address possible abuse of default rules, lawmakers should be held liable for the introduction of default rules that infringe rights or are in other regards violations of the constitutional order. These default rules could be recommended as being valid or invalid by “nudging oversight bodies”. For example, a quasi-public, quasi-independent “nudging ombudsman” could be appointed by parliaments. The task of this body would be to represent a broad variety of societal and legal concerns, as well as diverse academic and scientific perspectives and to oversee the conformity of nudges generally (not just default rules) to constitutional and basic legal principles. This goes beyond calls for the rigorous evaluation of nudging instruments (through RCTs and “regulatory impact assessments”) as proposed by Alemanno and Spina (2014).

The second nudge we would like to discuss here are shocking health warnings which are not a part of the legal system and which work through reliance on availability heuristics. People (potential consumers) react to the warning, but not to the law (see footnote 9). Reliance on availability heuristics is not a conscious process and cannot be controlled easily. We think that in order to avoid misusing this kind of policy solutions, shocking health warnings should be complemented with information about the legal source of such warnings, as well as by the liability of law or policy-makers who misuse this kind of instruments. We are aware that these kinds of legal solutions may lead to a decreased effectiveness of the policy. Our safeguards may, though they do not have to, decrease the effectiveness of a nudge, especially of these nudges which “work in the dark”. Paradoxically, the decreased effectiveness of nudges is not so much an objection, but should count as one more argument in favor of having safeguards for nudges. If after being made aware of their presence and impact, people chose to abstain from nudges and act differently, this can be an indicator that there may be reasons why they do not want to be nudged.

Further, we think that an effective way of making nudges more visible is to provide a legal registry of nudges, especially of those which are not a part of a legal system. We are also sympathetic to proposals made that expand the traditional scope of judicial review for the purpose of assessing nudging policies (Alemanno and Spina 2014). This would require taking into account the broader context in which nudges are to be implemented and may also spur a debate about the constitutional limits to reliance on nudges for different bodies of law in society. Lastly, we note that nudges have an important diachronic dimension. Due to their invisibility, it is increasingly problematic if societies “inherit” nudges from prior public administrations. In addition to a nudge registry, we advocate expiration dates for nudges which would force lawmakers to deliberate on implementation in regular intervals.

Thus, the core part of our proposals revolves around the claim that policy-makers ought to try as far as possible to complement nudges with legal norms or to make nudges connect with a legal system (or in other words – bring them into the legal system). The rationale for our proposal is to make nudges accessible and public and in this way weaken their non-cognitive and non-normative impact on agents’ behavior. Ideally, the introduction of legal norms in modern constitutional democracies is preceded by public deliberation (e.g. in parliaments), that helps to give policy instruments and associated policy goals legitimacy.¹⁷

7 Concluding Thoughts

Our basic objection to nudging concerns the consequences of nudging policies for understanding and treating law and political institutions. In the concluding section we would like to recall our epistemic concern, voiced at the beginning of this piece, about the reliance of policymakers on scientific findings. The important question which should be asked here is this: what kind of reasons do policymakers have in order to treat the current state of knowledge as reliable, or at least reliable enough to use it as a base for formulating and proposing policies? While proponents of nudging believe that scientific knowledge (especially in the newly emerging disciplines such as cognitive psychology or cognitive sciences and neuroscience) is subject to constant testing, critique and refutation (see Sunstein 2014b), there is no deeper reflection on the reliability of this knowledge as a basis for policy recommendations. This is further complicated by the worry that there is no neutral way to rely on behavioral findings in policymaking (in addition to fundamental difficulties of the legitimate roles of technocrats, experts and policymakers generally). The danger of regulatory capture is especially problematic when it might befall a policy instrument that is by design less visible and contestable than traditional legal and policy alternatives. For these reasons, we must proceed with caution if we want to turn democratic societies into nudge worlds.

We advocate caution and severe limits to nudging, particularly in light of T&S’s promise of overcoming partisanship through nudging in public policy. Contrary to the proponents’ optimism, nudging is not a “viable middle ground in our unnecessarily polarized society” (Thaler and Sunstein 2008, p. 252). This is because the way in which nudges bypass the normative function of the law is a novel way of instrumentally influencing citizens’ behavior, thereby jeopardizing a society’s capability to self-legislate.

In this text we have attempted to underline the consequences, thus far overlooked, of nudging for, first, an understanding of law, and second, the current institutions within which nudges are implemented. We have tried to formulate why we think nudges should be a part of a legal system and what kind of legal framework can be proposed in order to make a public debate about nudging possible. Our account of “law-as-normative” introduced an ideal type of law that contrasts with “law-as-instrumental”. Creating this distinction has enabled us to indicate the change in the understanding of law as the result of applying behavioral sciences in policy. Here, we do not put forward a philosophical account which could complement, support, and strengthen our reasoning. We are nevertheless convinced of the need for such a contrasting account to be developed. We propose a range of legal safeguards for nudges today and hope to inspire readers to engage in both the philosophical and the legal discussion concerning the issues raised here.