Thursday, November 12, 2020

Combining information on others’ energy usage and their approval of energy conservation promotes energy saving behaviour

Households reduced their electricity use the most when they learnt both that they were using more energy than their neighbours and that energy conservation was socially approved. This suggests that efforts to use social information to nudge conservation should combine different types of social feedback to maximize impact.

Messages for Policy
  • The content of social information messages determines their impact on energy conservation.
  • Combining descriptive information on neighbours’ efficient energy usage and injunctive social approval for energy efficiency maximizes the effectiveness of social information.
  • Delivering inconsistent descriptive and injunctive information reduces the impact of each piece of feedback.
  • Simply adding more pieces of feedback of the same type has a limited effect.
Based on J. Bonan et al. (2020).

The policy problem
Home Energy Reports (HER) are a popular means of encouraging energy conservation, reaching millions of energy utility customers across many countries. HERs typically rely on social information about the energy usage of a customer’s neighbours (descriptive feedback) and their social approval of energy conservation (injunctive feedback) to nudge recipients toward more energy-efficient behaviour. The specific content of both types of feedback depends on how the recipient’s energy usage compares to that of their neighbours (Fig. 1). Available evidence indicates that the impact of HERs on energy consumption varies significantly both across countries and across individuals. This raises the question of whether the heterogeneity in the effectiveness of HERs can be attributed to how social information feedback is conveyed. Answering this question could inform the design of more effective communication campaigns relying on social information.

Fig. 1: Home Energy Report.
a–c, Layout and content of a Home Energy Report for a user receiving three thumbs-up (a); and a user receiving two thumbs-up (b). Both versions of the report contain injunctive feedback, that is, the thumbs-up (top), and descriptive feedback, that is, the bars displaying actual energy consumption (bottom). The figure also displays the position of the randomized descriptive or injunctive norm primes, whose text is shown in (c). Reproduced from Bonan, J., Cattaneo, C., d’Adda, G. & Tavoni, M. Nat. Energy (2020). Copyright 2016-2020

The findings
Energy customers who received two different types of social feedback (descriptive and injunctive) encouraging them to save energy reduced their consumption more than low-energy users for whom conforming with the descriptive feedback would entail consumption increases, at odds with the injunctive feedback praising energy saving. The addition of a second piece of information of the same type (for example, adding a second descriptive messages that encouraged energy saving) had a limited impact. When feedback was inconsistent, the piece of feedback delivering the strongest message prevailed, where strength reflected the difference between the user’s energy consumption and that of their neighbours (descriptive feedback) and the intensity of social approval conveyed through visual cues (injunctive feedback). These results suggest the significance of synergies between different types of feedback, rather than the superiority of any one type of feedback. The results may be specific to the precise wording and graphical representations used to provide feedback in our HER (Fig. 1), and may not generalize to the whole customer base.

The study
We carried out a randomized controlled experiment in Italy in which households received HERs. We disentangled the impact of descriptive and injunctive feedback in two ways. First, we exploited the discontinuities in the injunctive feedback, which changed discretely as users’ consumption crossed certain thresholds, for instance shifting from one to two ‘thumbs-up’ as a user’s consumption dropped below the average of their neighbours. Second, we randomly assigned customers to receive a message at the bottom of the HER emphasizing either a descriptive or an injunctive norm of energy conservation (Fig. 1). Using data on the content of the HERs received by users and on their energy consumption, we were able to evaluate the impact of each piece of feedback in isolation, and when combined with others of the same or of different types.
by Jacopo Bonan, Cristina Cattaneo, Giovanna d’Adda & Massimo Tavoni 
Nature Energy Policy Brief
Volume 5, Published: 02 November 2020; Pages 832–833 (2020)

The main article "The interaction of descriptive and injunctive social norms in promoting energy conservation" by the same authors published on the same day at (pages900–909) notes
the magnitude of the average savings from the programme (−0.353%) is outside the range of those generated by similar ones in the United States (minimum = 0.88%, maximum = 2.55%), they are in line with the existing evidence from Europe. Various factors, such as lower average consumption in Europe than that in the United States, the specific features of the programme we studied or differences in beliefs across contexts, may be responsible for these differences. The heterogeneous effects, although not robust and only marginally statistically significant, are qualitatively in line with the existing evidence on the larger impact of social information on high electricity users and on the absence of boomerang effects among low users

These results provide initial, albeit weak, support for our conceptual framework. For high users, normative and injunctive feedbacks pull behaviour in the same direction, which results in a reduction in electricity almost twice as large as that in the average treatment effect. For low electricity users, conforming to the reference groups’ behaviour motivates a consumption increase (’boomerang’), but the injunctive feedback included in the eHER counterbalances the negative effect of the descriptive feedback. 
The injunctive feedback therefore induces stronger behavioural reactions among high electricity users, who are also exposed to the supporting descriptive feedback, than that among low electricity users, for whom the two types of feedback are at odds
Figure 3 presents the results in terms of level changes in electricity usage. Although there are no statistically significant changes in the effect of the eHER when crossing the threshold between the one and two thumbs up (Fig. 3a), the discrete shift in the injunctive norm reduces electricity use when moving from the two to three thumbs up (Fig. 3b). The corresponding empirical estimates are presented in Table 1 (columns 1 and 2).

Figure 3 - Impact of the injunctive feedback on electricity usage.

a,b, Each dot represents the average daily electricity usage in the 3 months after the receipt of the April–May 2018 eHER around the two versus one thumbs-up cutoff (n = 216,328) (a) and three versus two thumbs-up cutoff (n = 130,466) (b) within evenly spaced bins. The solid line represents the local linear fit, estimated separately on either side of the cutoff and the shaded area shows 95% confidence intervals. The number of bins was selected through the integrated mean squared error (m.s.e). The running variable (horizontal axis) reports the individual difference between each customer’s monthly consumption in the period reported in the eHER and the relevant cutoff. The cutoff is then represented by the vertical line, which is set to zero. For positive values of the score, customers get an extra thumbs up with respect to those with negative values of the score. In a, customers on the right of the cutoff consume less than the average neighbour and more than the top 15% most efficient and get two thumbs up. In b, customers on the right of the cutoff consume within the 15% most efficient neighbours and get three thumbs up. Bandwidths (BWs) and kernel are set following the data-driven process described for formal RD estimations of impacts reported in Table 1.


Fig. 4: Heterogeneous impact of the normative primes at different injunctive feedback cutoffs.

No comments:

Post a Comment