Under Pressure! Nudging Electricity Consumption within Firms. Feedback from a Field Experiment.

• Author: Charlier, Christophe

1. INTRODUCTION

   The practical difficulties of implementing Pigouvian taxes on energy consumption and the development of behavioral sciences have raised interest for non-price energy conservation policies. Behavioral economics has made major contributions to the development of this literature assuming more realistic individual behaviors than the one usually supposed. In particular for the interest of this paper, agents are supposed to take account of social norms when choosing pro-environmental actions (Benabou and Tirole, 2006; Brekke et al., 2003; Nyborg et al., 2006).

   Within this field, the use of nudges has been prominent among policy proposals since the pioneering work of Schultz (1999). Nudges can be presented (Thaler and Sunstein, 2008) as rules of thumb modifying choice architecture to induce individuals to choose a wished course of action. Nudges can consist in choosing the right default option (taking advantage of the inertia generally characterizing individual behavior), delivering information to correct misperceptions, or providing information on what social psychologists call "descriptive norms" and "injunctive norms". Cialdini et al. (1990) define "descriptive norms" as specifying what most others do ("the is"), and "injunctive norms" as specifying what most others approve or disapprove of ("the ought"). Nudges based on the former exploit the property that individuals tend to behave in accordance with behavioral norms, whereas a nudge based on the latter uses the weight of moral motive in individual decisions. (1)

   Several experiments have already been conducted at varying scales testing nudges on energy consumption and have generally focused on households' electricity consumption in residential sites (Allcott, 2011; Ayres et al., 2013; Houde et al., 2013; Sudarshan, 2017; Kendel et al., 2017). Their results illuminated behavioral strategies to increase energy conservation (Asensio and Delmas, 2015). Allcott (2011), for example, showed that the effect of a social comparison nudge used by the company Opower in a very large randomized field experiment with households, was equivalent to a 11 to 20% short-run price increase. Allcott and Roger (2014) focused on the persistence of the effects of Opower's program. They showed that the marginal effect of the program decreases because of habituation, but that long-term effects should not be underestimated because of the formation of a "capital stock" of habits and technologies facilitating energy economies.

   Nudging behavior has also been considered within firms, although this type of study is less common. Thaler and Sunstein (2008) surveyed a number of "private nudges" of this type. Egebark and Ekstrom (2016) conducted a field experiment in a Swedish public university comparing two types of nudges on the saving of paper by employees when printing. The first one took the form of a "green default option" to print on both sides of a sheet of paper (rather than one side). The second one was a "moral appeal" nudge consisting in sending messages to employees trying to convince them to reduce their use of paper. The authors found that the first nudge reduced the consumption of paper by 15%, whereas the second one had no effect. The results of a large-scale field experiment on fuel usage in aeronautics are presented by Gosnell et al. (2016). To induce captains to improve fuel consumption efficiency, they explore three treatments delivering information on: recent individualized fuel efficiency ("information" treatment); the achievement of specific efficiency targets ("personal targets" treatment); and donation to a chosen charity conditional on achieving the efficiency targets("pro-social incentives" treatment). Distinguishing three phases of fuel consumption during a flight, they show that the captains use fuel differently for each treatment. The pro-social incentives is the only treatment that has significant effects on the three fuel consumption phases. To recapitulate this narrow field of literature relating to nudging workplace energy saving, we can argue that: the default option treatment is likely to operate well; moral treatment and pro-social incentives have mixed results; and social comparison treatment, widely used to address households' energy saving, has not been explored yet.

   The field experiment we explore in this paper contributes to this literature by considering employees' energy consumption. Compared to experiments with households, experiments involving employees' energy consumption are original for several reasons. First, employees have no financial incentives to reduce their energy consumption, contrary to households who pay the bill, except if a part of their income is indexed to the firm's profit. (2) Even in this case, the pecuniary motive of energy usage may be assumed to be small. Second, an employee's peers are his coworkers. They can share environmental values, especially when the firm has a Corporate Social Responsibility (CSR) policy or has received an environmental certification. The pride or guilt a worker can feel because of the amount (or lack) of his energy saving are thus directly linked to the pressure of his peers. (3) This form of individual interaction might be facilitated by the fact that within a firm the "energy consumption place" is common to workers, so that observing each other is possible. Therefore, individual energy saving actions may result from peer pressure conveyed by descriptive and social norms. (4) Finally, from a quite pragmatic point of view, an important part of the energy consumption takes place at workplace. For example, in our field experiment we focus on electricity consumption of companies in the French services sector, which accounts for 75% of the country's labor force. In 2013, 45% of the total consumption of electricity was attributable to the combined residential/service sector market sector. (5) The service sector alone had a heated surface of 957 million m (2) and mean annual consumption of 240 kWh/[m.sup.2]. (6) Offices (i.e. the surfaces of interest in our study) represent 23% of this total heated surface. These data show the important role played by offices in applying behavioral strategies to reduce electricity consumption.

   We observe the electricity consumption of 47 company sites during 12 weeks. Our measures distinguish between electricity consumption for heating and electricity consumption for other purposes. The field experiment explores the extent to which three different nudges affect workers electricity consumption. The first treatment provides information on good practices on energy consumption with the help of stickers. The second treatment consists in weekly reports comparing the electricity consumption of the site with the consumption of the other sites participating in the experiment. Finally the third treatment delivers weekly messages linking electricity consumption to the natural and human consequences of global warming. These treatments are experimented alone during four weeks. The latter two treatments, which make use of descriptive and injunctive norms respectively, are coupled with the stickers during the four remaining weeks. This enables us to focus more specifically on the complementarity of two different environmental nudges.

   Employing more than 3700 observations, our difference-in-difference analysis suggests that private environmental nudges have no significant impact on workers' energy conservation when implemented alone, but become significant as soon as they are combined with the use of another nudge. In particular, the combination of injunctive norms and stickers has significant effects on electricity consumption on company sites participating in the experiment. To our knowledge, the adoption of conservative electricity behaviors within firms has rarely been studied. By investigating this issue, our study has important implications for "green" business strategies, indicating how a manager can use nudges to encourage employees to engage in responsible energy consumption.

   The structure of the paper is as follows. The next section presents the experimental design of our field experiment and describes the nudges used. We present our results in Sections 3 and discuss them in Section 4.

2. EXPERIMENTAL DESIGN

   2.1 Design and procedures

   We conducted a field experiment at 47 French companies' sites. This experiment aimed at measuring the effect of three energy saving programs. The field experiment sought to establish whether three types of nudges: stickers, moral appeal, and social comparison (the independent All participating companies were equipped with a Building Management System (BMS), thereby allowing us to obtain the daily energy consumption of individual equipment such as ventilation, lighting, power systems, heating, etc. Companies had neither labor contracts with specific clauses about employees' use of electricity, nor any mechanism of control. We also asked companies not to implement any parallel actions during the field experiment aimed at encouraging energy conservation behavior among employees. A profit-sharing arrangement was implemented for employees on 37 sites. This arrangement links individual salaries to any cost saving. However, the impact of individual electricity conservation on remuneration can be considered as very small as well as the resulting financial incentives. (7)

   Each of the 47 companies was randomly assigned to either one of the three treatment groups (N = 11 or 12 for each group) or the comparison group (N = 13). Table 1 outlines the total number of companies within each group. All of the companies in the experiment are service companies. All employees work mainly on computers, have a collective rest area with a coffee maker, kettle, fridge, etc. A series of characteristics distinguishes them, i.e. area, number of employees and location (see Table 2).

   The experiment was carried out in three stages over 12 weeks (see...