Do social norms trump rational choice in voluntary climate change mitigation? Multi-country evidence of social tipping points

Abstract

The rational choice model of voluntary public good provision predicts that an individual's contribution to climate change mitigation responds negatively to larger contributions by others whereas social norm theory maintains that one's own contribution is positively related to that of others. This paper tests the competing hypotheses empirically using representative data for about 30,000 individuals from 23 European countries. The paper finds that, up to a threshold percentage of others perceived to engage in mitigation, individuals' willingness to engage in mitigation themselves is lower the more other individuals are perceived to engage in such behavior, whereas the relationship is positive when the threshold is passed. Since the actual percentage of others perceived to engage in mitigation is lower than the estimated threshold (30 to 56%) in a number of countries, marginal increases in the percentage of others perceived to behave in a climate-friendly way may backfire by enhancing free-rider behavior. For the social norm to take grip, policy-induced non-marginal increases of perceptions may be required to shift the system past the tipping point.

Introduction

The past three decades have seen a departure from the standard rational choice model of pro-social and pro-environmental behavior. While the rational choice model is of a consequentialist nature, assuming that people do what yields the individually optimal (expected) outcome, alternative approaches are deontological, maintaining that individuals evaluate behaviors per se and choose among them based on conformity with moral norms – inner feelings of what is right or wrong – and social norms – what most people do (descriptive social norms) or approve of doing (injunctive social norms).¹

Conformity to moral and social norms is generally understood as a mechanism that enhances the voluntary provision of public goods by countervailing the free-rider incentives highlighted by rational choice theory (e.g., Nyborg, 2018). With respect to voluntary public good provision there is, however, an important difference between moral-norm and injunctive-social- norm accounts of behavior, on the one hand, and descriptive-social-norm accounts, on the other. While the former are logically independent of rational choice theory, descriptive-social-norm accounts are logically opposed to rational choice theory, that is, the two make opposite behavioral predictions: While rational choice theory predicts that an individual's contribution to the public good responds negatively to others' contributions (e.g., Buchholz and Sandler, 2021), conformity to a descriptive social norm – what most others do – implies that an individual's contribution is positively related to the contributions provided by others.

Focusing on climate change mitigation, this paper tests the competing predictions empirically, using representative survey data for about 30,000 individuals from 23 European countries. Specifically, based on the idea of social tipping points (see below), the hypothesis will be explored that there are threshold levels of others' behavior that need to be passed for the descriptive social norm mechanism to dominate the rational choice mechanism.

The literature on the behavioral theory and empirical evidence on social norms as applied to pro-environmental behavior is concisely reviewed by Farrow et al. (2017). The review identified 42 studies, of which 11 referred to energy use, 11 referred to recycling, and 20 referred to green consumption, littering, water conservation, hotel guests' towel reuse, and pesticide use. These studies covered both behaviors and behavior intentions and involved descriptive and injunctive norms. The behaviors studied were found to be significantly positively related to descriptive norms (others' actual or perceived pro-environmental behaviors) in 31 out of 39 studies. An early example where clues about others' behavior prompted pro-environmental behavior change refers to the reuse of bath towels in hotels (Schultz et al., 2008). A more recent study (Falk et al., 2021) found that individuals' donations to curtail climate change were significantly positively related to the percentage of people they believed to engage in climate change curtailment. Farrow et al. (2017) take such findings to suggest that people take information or beliefs about what others do (descriptive norms) as a signal of which behaviors are socially approved and which are not (injunctive norms).

While the evidence found in this literature is often context-specific and local in nature, Nyborg et al. (2016) note that recent history has seen large-scale changes of norms and behaviors, for instance with respect to fertility behavior or the cessation of smoking in public places. However, for such changes to be self-reinforcing, they argue, it is necessary that social tipping points are passed, that is, for social conformity to induce behavior changes a sufficiently large number of people must already have changed their behavior so that a “new” social norm emerges. Below such thresholds, social conformity implies vicious rather than virtuous circles.

Though fertility and smoking behaviors do not necessarily or to a large extent refer to public goods, the notion of threshold values or tipping points may be even more relevant when public goods are involved. In such cases “new” norms of behavior (e.g., environment-friendliness) do not only compete with “old” ones (e.g., “keeping up with the Joneses”) but with incentives to free-ride on others' contributions to the public good. This makes it likely that public good provision by others may not only be ineffective in triggering an individual to behave similarly, but counter-productive unless a tipping point is passed where the “virtuous” conformity mechanism starts to dominate rational-choice free-rider considerations.

While the rational-choice/free-rider model is the standard framework that predicts a negative relationship between own mitigation behavior and that of others, non-standard mechanisms have been proposed that lead to the same prediction. One is so-called moral licensing. This involves the idea that people's sense of duty to contribute to a public good is inversely related to the level of the public good: When morally motivated people learn that the level of a public good increases, due to larger contributions by others, their sense of duty and the corresponding contribution may diminish, possibly in favor of other collective goals/goods (Mullen and Monin, 2016; Lasarov et al., 2022). The notion of moral licensing combines consequentialist (outcome-focused) and deontological aspects since the (collective) outcome of behaviors (or perception of outcome) influences an individual's behavior (as in the rational choice framework) even though the behavior's motivation may be of a deontological nature.²

Another mechanism that implies a negative relationship between own and others' behavior involves social distinction: Low overall levels of a desirable behavior may raise the adoption of this behavior through motivations to seek distinction (e.g., Bénabou and Tirole (2006). If, conversely, the overall level of the behavior increases (from a low level), the behavior becomes less attractive as a vehicle for attaining distinction and may thus be reduced. Consistent with a distinction as well as a moral licensing account it was found that informing people about the energy consumption of their neighbors had a boomerang effect for some individuals who learned that they outperformed the average (Allcott and Rogers, 2014; Schultz et al., 2007).

Focusing on climate change mitigation – by means of buying the most energy efficient home appliances – this paper contributes to the literature by studying the potentially opposing effects of others' behavior in a large-scale multi-country representative sample, the European Social Survey. Controlling for variables that capture the activation of a moral norm of climate change mitigation, income, and a battery of sociodemographic correlates, the paper finds that, up to a threshold level of others perceived to engage in mitigation, individuals' willingness to engage in mitigation themselves is lower the more other individuals are perceived to engage in such behavior, whereas the relationship is positive when the threshold (tipping point) is passed. The u-shaped relationship between own and others' perceived behavior is found to exist both in the overall sample and in the majority of individual countries as well as in all subgroups of the population (by age, gender, level of education, ideological position, and sense of moral duty). It is robust to including close to a dozen additional covariates that may determine both individuals' own behavior intentions and their beliefs about what others intend to do.

Since the percentage of others perceived to engage in mitigation is lower than the estimated threshold (30 to 56%) in a number of (but not all) countries, marginal increases in the percentage of others perceived to behave in a climate-friendly way may backfire by enhancing free-rider (or moral-licensing) behavior. For the social norm to take grip, policy-induced non-marginal increases may be required in such cases. If the actual level of the relevant behavior is sufficiently large relative to the perceived level, informing people about the actual level may trigger an increase in the behavior studied.

The paper proceeds as follows. Section 2 lays out the conceptual framework and empirical approach and describes the data and variables used. Section 3 reports the results. Section 4 addresses issues of identification. Section 5 provides a general discussion, and Section 6 concludes.