Contents lists available at ScienceDirect

Ecological Economics

journal homepage: www.elsevier.com/locate/ecolecon

Behavioral Economics and Energy Conservation – A Systematic Review ofNon-price Interventions and Their Causal Effects

Mark A. Andor⁎, Katja M. FelsRWI, Hohenzollernstraße 1-3, 45128 Essen, Germany

A R T I C L E I N F O

JEL Classification:D10D12L94L95Q41Q48Q58

Keywords:Systematic reviewBehavioral economicsEnergy demandEnergy efficiencyEnvironmental certificationSocial norms

A B S T R A C T

Research from economics and psychology suggests that behavioral interventions can be a powerful climatepolicy instrument. This paper provides a systematic review of the existing empirical evidence on non-priceinterventions targeting energy conservation behavior of private households. Specifically, we analyze four nudge-like interventions referred to as social comparison, commitment devices, goal setting, and labeling in 44 in-ternational studies comprising 105 treatments. This paper differs from previous systematic reviews by solelyfocusing on studies that permit the identification of causal effects. We find that all four interventions have thepotential to significantly reduce energy consumption of private households, yet effect sizes vary immensely. Weconclude by emphasizing the importance of impact evaluations before rolling out behavioral policy interventionsat scale.

1. Introduction

Climate change mitigation programs are on the political agendaworldwide. As a result of ambitious CO2-reduction goals, policymakersare increasingly interested in non-price interventions targeting privatehousehold energy consumption. Research from both economics andpsychology has shown that behavioral interventions – also referred toas nudges – can be powerful tools in shaping people's behavior in avariety of domains (see, among others, the influential publication byThaler and Sunstein 2008).1 Non-price measures are relatively in-expensive to implement and do not interfere with people's choice sets asstrongly as, for example, taxes or bans on certain products. Conse-quently, policy makers are now exploring nudges as a cost-effectiveapproach for reducing energy consumption (Allcott 2015). If proveneffective, these interventions could be established as integral andcomplementary components of climate change policy (Allcott andMullainathan 2010, Benartzi et al. 2017). This is why researchers areincreasingly interested in understanding the effect of non-price mea-sures on residential energy consumption.

This paper presents findings of a systematic review on the

effectiveness of behavioral interventions to induce energy conservation.We study the following four interventions: social comparison, com-mitment devices, goal setting, and labeling. Furthermore, the reviewfocuses on causal in contrast to correlational effects. To this end, weonly include those studies that employ an empirical estimation strategyenabling the identification of a causal relationship between a policyintervention and consumption behavior. To our knowledge, this is thefirst study that systematically reviews all published results from beha-vioral economics and related areas of research that are based on arigorous evaluation of causal effects.

Our study builds on a few earlier reviews that only focus on a subsetof our interventions. Many of these point to potential problems of in-cluding effects from correlational studies in their sample, i.e. studiesthat are not able to disentangle causation from correlation. Abrahamseet al. (2005) evaluate the effectiveness of some interventions aiming toencourage households to reduce energy consumption. They concludethat information has an influence on knowledge, but does not ne-cessarily result in behavioral changes or energy savings. Rewards haveeffects on energy conservation, but they are rather short-lived. Feed-back, in particular when it is given frequently, can also be effective.

https://doi.org/10.1016/j.ecolecon.2018.01.018Received 20 July 2017; Received in revised form 18 November 2017; Accepted 15 January 2018

⁎ Corresponding author.E-mail address: andor@rwi-essen.de (M.A. Andor).

1 The facts that the book “Nudge” (Thaler and Sunstein 2008) has already been cited more than 9000 times (Google Scholar, checked 11/15/2017) as well as that Richard Thaler wonthe Nobel prize “for his contributions to behavioral economics” in 2017 (Nobelprize.org 2017) can be seen as two indicators of a growing academic interest in behavioral interventions.

Ecological Economics 148 (2018) 178–210

0921-8009/ © 2018 Elsevier B.V. All rights reserved.

T

More recently, Karlin et al. (2015) conducted a meta-analysis on theeffect of feedback on energy usage. They conclude that feedback is ef-fective but with significant variation in effects. Furthermore, Delmaset al. (2013) analyze the effect of information strategies on energysavings and find a substantial reduction effect on average. However, ina similar vein as Abrahamse et al. (2005), they conclude that the effectdiminishes with the rigor of the study, indicating potential methodo-logical issues in the considered literature. In particular, none of theexisting reviews takes into account whether the considered studiesapply a method that has the potential to identify the causal effect of theintervention, which is critical to the question of its policy relevance(Imbens and Wooldridge 2009).

Consequently, our systematic review differs from previous researchby solely focusing on studies that have the potential to identify causaleffects between the intervention and the outcome. Furthermore, weinclude articles published up to May 2017 in working paper series aswell as peer-reviewed journals to provide the most comprehensive andup-to-date account of research in economics and psychology. This isparticularly important because there has been a growing number ofhigh-quality studies in the recent past. Hence, our review comprisesseveral very recent large-scale randomized controlled field experi-ments. As an additional contribution, our systematic review is the firstto account for labeling as a non-price intervention, which has beenapplied worldwide on a large scale and potentially affects millions ofhousehold decisions each year.

The paper proceeds as follows. In the subsequent section, we defineand motivate the four considered interventions. Section 3 explains themethodology of the systematic review. In Section 4, we synthesize anddiscuss the results. Section 5 concludes with recommendations for re-searchers and policy makers.

2. Behavioral Interventions and Energy Conservation

A considerable percentage of annual emissions in industrial coun-tries is induced by residential energy consumption. In addition, privatehouseholds are a prime target for behavioral interventions (Karlin et al.2015). Households may conserve energy in two ways: First, they canchange their consumption of energy services, for example by reducinglighting use. Second, they can modify their purchasing behavior andinvest in energy efficiency, for example by buying a highly efficientwashing machine.2 Both, the purchase decision and the consumptionbehavior, can be targeted by policy interventions. Non-price interven-tions are usually justified with so called internalities, i.e. externalitiesthat the agent imposes on herself by making suboptimal choices,measured by her own experienced utility (Chetty 2015).

According to Allcott (2016), six main internalities are responsiblefor consumer mistakes in the domain of energy conservation: presentbias, bias toward concentration, biased beliefs, costly information ac-quisition, exogenous inattention, and endogenous inattention. Ourstudy selected those non-price interventions that are most common andsuitable to address each of these internalities (see Table 1).3 The re-sulting four non-price interventions and the internalities they addressare explained in more detail below.

2.1. Social Comparison

Social comparison refers to the process of giving households in-formation about their energy consumption in relation to the

consumption of comparable households. Such a comparison is closelyconnected to also receiving feedback about one's own behavior. Thechosen reference group should be relevant for the treated household(Abrahamse et al. 2005) and can be, for instance, consumers of thesame energy provider or households within the same postcode-level.Moreover, the choice of the reference level is important: the house-hold's consumption can either be compared to the average consumptionlevel of the reference group or to a more ambitious group, e.g. the mostefficient 10%.

Social comparison addresses biased beliefs about one's own con-sumption behavior in comparison to others. For example, a personmight consider herself an environmentally friendly energy consumerand underestimate her actual consumption level when compared toother consumers. This biased belief can be corrected by a social com-parison.

The potential energy conservation effect of a social comparisonmight be triggered by three phenomena. First, many people exhibitreference dependent preferences (Kahneman 2003). Accordingly, socialnorms can constitute a reference point. Complying with these normsincreases most individuals' utility whereas deviating from it typicallyleads to disutility caused by social disapproval (Schubert andStadelmann 2015). Second, in situations of uncertainty, individualsmay use other peoples' behavior as orientation by implicitly assumingthat those others have more information about the socially desiredbehavior (Allcott and Mullainathan 2010, see also Delmas et al. 2013).Consequently, people tend to adjust their actions according to theprevalent group behavior. Third, social comparisons evoke feelings ofcompetition (Abrahamse et al. 2005). This is especially important whenthe household's consumption level lies above the average or above somethreshold that the household perceives as desirable (for example, be-longing to the most efficient 10% of costumers).

2.2. Commitment Devices and Goal Setting

Commitment devices are “a set of interventions that allow in-dividuals to lock themselves today into the action that they want to taketomorrow” (Allcott and Mullainathan 2010, p. 2). Examples of com-mitment devices are oral or written pledges or promises to conserveenergy (Abrahamse et al. 2005). The commitment can either be apromise to oneself, or alternatively be made public. Goal setting com-bines commitment with a concrete reference point. Instead of pledgingto conserve energy, a household specifically promises, for instance, “toreduce energy consumption by 10 percent within the next year”. Notonly setting a reduction level but also a deadline for achieving this goalfacilitates an evaluation of success or failure. This increases pressurebut also motivation by making satisfaction conditional on a desiredlevel of performance (van Houwelingen and van Raaij 1989). A goal canbe chosen by the household itself (being a form of commitment device)or be externally set (for example by institutions).

The idea behind voluntarily binding one's own future behavior isthat some people are aware that they sometimes have time-inconsistentpreferences (O'Donoghue and Rabin 1999). For instance, asO'Donoghue and Rabin (2008) point out, many people procrastinate,sometimes to the extent that the desired action is never taken. Thisdistortion of one's own preferences is induced by present bias, the

Table 1Internalities and chosen interventions.

Internality Chosen intervention(s)

Present bias Commitment devices and goal settingBias toward concentration LabelingBiased beliefs Social comparison and labelingCostly information acquisition LabelingExogenous and endogenous inattention Labeling

2 The purchase of an energy efficient appliance will ultimately result in reduced energyconsumption when expected energy savings are not completely offset by an increase inthe use of the appliance, which is known as the rebound effect (see, for instance, Frondeland Vance, 2013).

3 A further important intervention in this regard is feedback. Yet, because the com-prehensive study of Karlin et al. (2015) provides a recent account of the existing researchon the intervention, we do not consider this intervention in our review.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

179

tendency to discount more strongly between today and tomorrow thanbetween two future dates (Laibson 1997).

A commitment device helps individuals to overcome such time-in-consistent preferences by providing a source of motivation: it comparesthe present situation with a desired future state (van Houwelingen andvan Raaij, 1989). When the commitment is a pledge to oneself, it ap-peals to a personal norm (the individual wants to satisfy expectationstoward itself). A public commitment creates expectations by others,thereby leading to social pressure (Abrahamse et al. 2005). In additionto that, a specific goal targets reference-dependent preferences: in-dividuals aim at a pre-determined level and judge their performanceaccording to this reference point. Achievement of the goal provides afeeling of accomplishment, whereas failure creates disutility even if thegoal level was externally set (Bandura 1986). This is mainly due to lossaversion (losses loom larger than gains), one of the key concepts ofprospect theory (Kahneman and Tversky, 2013; for an overview seeReisch and Zhao 2017).

2.3. Labeling

A label is a tag that summarizes information on a good in an easilyaccessible way. This can be achieved by presenting a selection of in-formation about the product's attributes on the label or by visualizingthe most relevant information in a graphical manner. In the domain ofenergy consumption, labels can, for example, comprise information onenergy usage levels of appliances or energy efficiency standards ofhouses.

In principle, there are two different sorts of labels. Either the la-beling program is voluntary and an appliance is awarded the label forsatisfying certain criteria like the US Energy Star Label, or a label ismandatory for all appliances, as with the EU Energy Label. In the lattercase, goods can be ranked according to their performance along thecriteria specified by the label, for instance the EU energy efficiencyclasses.

Labeling addresses several of the internalities that are relevant inthe domain of energy conservation. First, consumers potentially re-spond to labels by ascribing more attention to certain features of thegood and by this overcome inattention. As a label renders selected cri-teria (more) salient, it aims at the availability heuristic, i.e. a simpli-fying rule that gives highly accessible features a stronger influence ondecisions while information of low accessibility will largely be ignored(Kahneman 2003). By this, a label also helps to reduce costly informationacquisition. However, the mechanism of the availability heuristics re-mains true regardless of whether the costs of gathering relevant in-formation, for example on average usage levels or life expectancy ofappliances, is high or low (Schubert and Stadelmann 2015). Moreover,labels target biased beliefs about the value of different energy con-servation actions. If, for example, a consumer underestimates the sav-ings' potential of an appliance, she might underpurchase energy effi-ciency based on her own preferences (Allcott 2016). A label correctingher beliefs by pointing out the potential savings in energy costs ad-dresses this internality and might trigger more energy efficient pur-chases. The same is true for the bias toward concentration that describesthe phenomenon that consumers might downweight energy cost savingsthat accrue in small amounts compared to a bigger sum like the salesprices for an energy efficient appliance, which has to be paid im-mediately (cf. Kőszegi and Szeidl 2012).

3. Methodology

Following the guidelines for systematic reviews suggested by theCampbell Collaboration (2014), we take five successive steps to identifyand analyze relevant studies: setting up criteria for including studies inthe review, literature search, selection of studies, data extraction, anddata analysis.

3.1. Inclusion Criteria (PICOS)

As a first step, we developed a set of criteria along the so-calledPICOS – an acronym for participants, interventions, comparisons, out-comes, and study designs (Campbell Collaboration 2014). The PICOSguided the selection of studies in the further process. We include studiestargeting private households or individuals living in an industrialized oremerging country where the living situation (especially with regard toenergy consumption) is comparable to an industrialized country. Stu-dies focusing on energy consumption in official buildings or enterprisesettings are excluded.

Our focus is on the four interventions social comparison, commit-ment devices, goal setting, and labeling. We also consider combinationsof these interventions with each other or with additional non-pricemeasures (e.g. feedback, energy saving tips). In contrast, all studiescombining these interventions with financial incentives (for instance,dynamic pricing) are excluded from the review, as long as they do notallow identifying the effect of the non-price intervention separately.

Regarding the outcome, we include all studies that report an in-dividual's or household's actual or self-assessed usage level of energy,gas, or water. For labeling, we also consider secondary outcomes likethe perception of and the willingness to pay (WTP) for energy effi-ciency. These secondary outcomes influence usage levels via differentchannels (as discussed in Section 2).

We include all studies that permit the identification of the causalrelationship between the intervention and the outcome of interest.Hence, studies are considered if their methodology is based on a causalinference design like randomized controlled trials (RCT), matching,difference-in-differences, instrumental variable estimation, and regres-sion discontinuity design, or if they control for self-selection with al-ternative methodologies.4 Studies that do not employ methods suitablefor identifying the causal effect of an intervention are excluded from thereview. Overall, we apply no time restriction: all study results publishedin a journal or as working paper up until May 2017 are considered inthe review.

3.2. Literature Search

We used two main avenues to identify relevant studies: first, akeyword search in databases, and second, a backward search on re-levant review studies. Before the database search, we pre-determined asystematic combination of keywords for each of the four interventions(see Appendix B1). These keywords were employed on two databases:EconLit, consisting of more than 785,000 articles from peer-refereedjournals and acknowledged working paper series in economics, andScienceDirect, accessing more than 13 million articles from journals indifferent disciplines, from which our study chose the disciplines“Economics, Econometrics, Finance”, “Psychology”, “Social Sciences”,“Environment”, and “Energy”. This returned a total of 1,121 results.

Additionally, we conducted backward searches in the literature ofthe following four review studies: Abrahamse et al. (2005), Delmaset al. (2013), Karlin et al. (2015) and Lokhorst et al. (2013). Of the 147articles identified, we included 71 studies after ruling out duplicates.This garnered a total of 1,192 studies for the screening process of thisreview (see Fig. 1).

3.3. Selection of Studies

All studies were screened by independently reading the title andabstract and assessing whether a study satisfied all inclusion criteria setup in the PICOS via an inclusion decision form (see Appendix B2). Weexcluded 1,040 studies due to this procedure, leaving 152 articles for

4 For an excellent overview of causal inference designs see Angrist and Pischke (2009)or Imbens and Wooldrigde (2009).

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

180

full text screening (see Fig. 1), which again was conducted in-dependently. In cases of differing assessments, we consulted a third(independent) scientist and solved discrepancies by discussion. Acommon reason for exclusion was that studies discussed strategies forenergy conservation behavior theoretically or approached it with aqualitative design without actually measuring the effect of an inter-vention. When screening the selected full text articles, the main reasonfor exclusion was that studies neither applied a causal inference designnor controlled for self-selection. In contrast, at the beginning of thescreening process a majority of papers was completely off-topic. Finally,44 articles satisfied the criteria of this systematic review.

3.4. Data Extraction

We developed a detailed coding sheet (see Appendix B3) based onthe guidelines of the Campbell Collaboration (2014). Two reviewerscoded each study independently of each other according to the samecriteria. In cases in which the extracted information was ambiguous (forinstance regarding the method of causal inference design), we consulteda third (independent) scientist and discrepancies were solved by dis-cussion. The results from the data extraction are summarized in a re-sults table (see Appendix A1).

4. Results

4.1. Brief Characterization of Included Studies

4.1.1. Research InterestAmong the considered interventions, social comparison has at-

tracted the most research attention. While 245 independent studiestested at least one social comparison treatment, only ten evaluate theeffects of goal setting and commitment. Additionally, nine studies as-sess the impact of labeling. In terms of available treatment effects, theanalysis of our systematic review can rely on 47 documented effects forsocial comparison, 30 for goal setting and commitment, and 28 for la-beling. The difference in numbers between articles and treatment ef-fects is due to the fact that many studies consider more than one re-levant intervention or evaluate multiple versions of the intervention inquestion.

Regarding the studies' publication dates an interesting patternemerges. While the peak of research interest in goal setting and com-mitment was in the 1980s, followed by a longer neglect and a recent

Fig. 1. Overview of results in the searching, screening and selection process.Note: The articles by Ferraro et al. (2011) and Ferraro and Price (2013) document short-term and long-term effects of the same study. They are thus summarized in the final results table and depictedas one study in the last stage of this figure. The study by Kurz et al. (2005) investigates a social comparison treatment as well as labeling. Here it is categorized as a social comparison study. The studyby Jaeger and Schultz (2017) evaluates treatments that include social comparison and commitment. As they do not provide a statistical analysis for the social comparison treatment effects, we listthem as a commitment study.

5 As Jaeger and Schultz (2017) do not report any statistical analysis for their socialcomparison interventions, we do not include them in the social comparison analysis.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

181

rediscovery,6 labeling has been largely neglected up till 2011. Sincethen, we observe a growing academic interest in analyzing the impactof the intervention. Similarly, research interest in the causal effects ofsocial comparison on energy consumption behavior is fairly recent: theoverwhelming majority of studies was published after the year 2004,with 80% of studies in this sub-sample (19 articles) being more recentthan 2010.

4.1.2. MethodsFor social comparison and goal setting/commitment, one dominant

method is applied, namely the evaluation via randomized controlledtrial. This methodology not only produces a high internal validity butalso has advantages in terms of external validity as it (in most studies)observes real-life behavior. In contrast, most of the studies on labelingare conducted as laboratory experiments or choice-experiments withinonline-surveys, which have low external validity. Specifically, re-searchers ask subjects, for instance, which appliance they would choosein a set of choices and attribute differences in choice patterns to the waythe choice set is presented (which differs between control and treatmentgroup). Obviously, such choices are not real purchases but rather de-cisions within a hypothetical and limited choice set. It is thus not cer-tain whether the reported effects would also occur in a real worldsetting, and results should be treated with caution despite their highinternal validity. However, a few labeling studies also exploit a policychange as a natural experiment or conduct an RCT with a labeling in-tervention.

4.1.3. Geographical LocationMost of the studies identified by the systematic review are located in

the US. This is especially true for goal setting/commitment, which up todate only has been evaluated in the US and the Netherlands. For socialcomparison and labeling, studies are conducted in several countries, forinstance Japan, Australia, Finland and Great Britain. The majority ofpublished research (social comparison) or at least a great share (la-beling), however, also targets a study population in the US.

4.1.4. Combination of InterventionsFor social comparison and goal setting/commitment, almost all

studies combine the considered interventions with other treatmentsthat we do not study. This poses a challenge for identifying the actualeffect of the intervention. It moreover makes it difficult to compareeffects from different studies, as they do not evaluate the same treat-ment. Fig. 2 shows that energy saving tips and a comparison of usagehistory are most popular among treatment combinations. Within thesub-sample of studies that allow identification of the pure effect by notcombining the intervention with additional treatments, other limita-tions apply. For instance, except for the study by Dolan and Metcalfe(2013), the relevant social comparison studies suffer from methodolo-gical shortcomings (e.g. no reported effect sizes). For goal setting andcommitment devices, there are substantial differences regarding theintensity of the treatment, because the pre-determined level of the goalvaries as well as the criterion whether the level was self-selected orexternally set. All in all this prohibits arriving at a coherent pictureregarding the existing empirical evidence on the pure effects of theconsidered interventions.

4.1.5. OutcomesThe vast majority of studies on social comparison and goal setting/

commitment focuses on actual or self-assessed energy consumption astheir outcome of interest. For labeling, the case is different. Most stu-dies evaluate the effect of labels on the WTP for energy efficient ap-pliances, followed by studies on the estimation of energy saving

potential or hypothetical purchase decisions regarding energy efficientappliances (see Fig. 3). Only one labeling study looks at actual energyconsumption. Consequently, the quality of empirical evidence differsbetween the interventions: while the majority of labeling studies arebased on stated preferences approaches (with their well-known lim-itation of hypothetical nature compared to revealed preferences), forthe other interventions most studies analyze real behavior.

4.2. Synthesis of the Evidence

4.2.1. Social ComparisonOverall, the empirical evidence suggests that social comparison

triggers energy conservation. All but two of the 24 studies presentstatistically significant results in at least one of their analyzed treatmentgroups.7 A social comparison intervention results in a reduced energyconsumption of private households ranging from 1.2% to 30% com-pared to the control group (see Fig. 4). The only study showing an in-crease in energy usage level is Schultz et al. (2007) in one of theirtreatment groups. Yet, the respective treatment group was deliberatelyrestricted to low users only, demonstrating the so called boomerangeffect, i.e. a social comparison can lead to adverse effects for the groupthat performs well in the comparison. They additionally show that theboomerang effect can be eliminated by adding an injunctive message.With regard to other outcomes than actual energy consumption,Komatsu and Nishio (2015) find a significant (positive) effect of thesocial comparison treatment on the motivation to conserve energy.

The heterogeneity of effects may be attributed to the medium viawhich the household receives the social comparison. As a tendency, weobserve that online/e-mail and In-Home-Display (IHD) treatments seemto result in a higher effect than letters. However, social comparisonsusing IHDs are still underresearched. Another noteworthy fact is thatstudies with higher sample sizes (above 80,000) find smaller effects(around 2%). Both observations are compatible with existing evidencefrom previous reviews (Karlin et al. 2015).

One specific intervention design, the social comparison based“home energy reports” (HER) by the private company Opower, hasattracted major research attention. Opower cooperates with numerousUS-energy suppliers and mails the HER to more than ten millionhouseholds in the United States with the aim of reducing electricityconsumption. The HER is a two-page letter with a bar graph comparingthe household's energy consumption to its geographically nearestneighbors in similar house sizes on the first page and energy saving tipson the second page. Several high quality studies in this review (Allcott2011; Allcott and Rogers 2014; Ayres et al. 2013; Costa and Kahn 2013)report significant, yet modest reduction effects. The empirical evidencefor this specific intervention is outstanding as the internal and externalvalidity for the US is high, long-term effects are documented and cost-benefit analyses are conducted.8 In a very recent study, Andor et al.(2017a) show that social comparison based HERs are most likely not acost effective climate policy instrument in many countries, in particularin Europe, due to lower electricity consumption levels and carbon in-tensities. For a study population in Germany, they additionally measurea substantially smaller treatment effect of social comparison basedHERs. The numerous high quality studies can be seen as best practicefor the evaluation of an intervention at large scale.

In all studies, in which a comparison of two similar combinations of

6 Almost two thirds of studies were published between 1978 and 1989, the remainingshare since the year 2002.

7 The study of Hakaana et al. (1997) is not considered in this regard, as the article doesnot report any significance levels at all.

8 Yet, on the basis of 111 RCTs that evaluate the HER, Allcott (2015) shows that evenfor the exceptional case that there exists many replications, program evaluations can stillgive systematically biased out-of-sample predictions due to a “site selection bias”. In thisexample, predictions from the first 10 replications substantially overstate efficacy be-cause, among others, utilities in more environmentalist areas are more likely to adopt theprogram, their customers are more responsive to the treatment, and utilities initiallytarget their treatment at higher-usage consumer subpopulations.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

182

interventions is possible, the effect increases when social comparison isadded to the treatment (Ferraro and Price 2013; Mizobuchi andTakeuchi 2013; Tiefenbeck et al. 2016). Moreover, Schultz et al. (2015)find a significant reduction in energy consumption of 9% compared toan insignificant treatment effect, when social comparison is added tothe initial treatment intervention. In addition, the reference groupseems to matter: while Shen et al. (2015) do neither find a significanteffect of a social comparison with the entire neighborhood nor withnext-door-neighbors, the comparison with street neighbors results in areduced consumption by 5.4%.

Two out of three studies that aim to identify the pure effect of socialcomparison have methodological limitations. Haakana et al. (1997)find that a pure social comparison treatment shows similar or smallereffects as a social comparison with energy saving tips. The study,however, provides no information about significance levels for any ofthe effects. Komatsu and Nishio (2015) analyze the self-assessment ofone's own consumption in regard to the neighbors' consumption and theown motivation to conserve energy. They find ambiguous effects of four

social comparison treatments, but do not report effect sizes. Moreover,the analysis is not based on actual behavior but solely relies on self-assessments. In contrast, Dolan and Metcalfe (2013) show that biannualfeedback with a comparison to the neighbors' consumption triggersenergy conservation of around 4%, while the combination of this in-tervention with tips to save energy increases the effect to almost 11%.

Some studies suggest that social comparison might also result inadverse effects. Schultz et al. (2007) show that low-consumptionhouseholds significantly increased their energy consumption afterhaving learned that they are below-average users. This raises importantquestions regarding a tailored application of social comparison treat-ments. Furthermore, Tiefenbeck et al. (2013) investigate side-effects ofa social comparison treatment. They find a significant reduction inwater consumption of 6% after they provided weekly feedback aboutwater consumption levels per capita, accompanied by a social com-parison with the most efficient 10% of users, and tips about how toconserve water. At the same time, the electricity consumption of thetreatment group increased by 5.6% compared to the control group. Theauthors argue that their “findings are consistent with the concept ofmoral licensing” (Tiefenbeck et al. 2013, p. 160), the phenomenon thatpast good deeds favor a positive self-perception which in turn createslicensing effects, leading people to engage in behavior that is less likelyto be in line with their moral values (Nisan and Horenczyk 1990). Yet,other phenomena could also explain this, for example, attention shiftsfrom conserving electricity to conserving water (as noted by one re-viewer). However, if such side-effects occur, it is not clear whether anintervention induces sustainable behavior in a broader sense, even ifthe estimated treatment effect suggests a reduction of the “direct”outcome (in this example: water consumption). Therefore, we agreewith Tiefenbeck et al. (2013, p. 160) who advocate the “adoption of amore comprehensive view in environmental program […] evaluation inorder to quantify and mitigate these unintended effects”.

In regard to long-term effects, no clear picture emerges. Severalstudies document a reinforcement of effects in the long-run for some oftheir treatment groups (Alberts et al. 2016; Allcott and Rogers 2014;Delmas and Lessem 2014; Dolan and Metcalfe 2013; Schultz et al.2007). In Staats et al. (2004), two initially insignificant short-term ef-fects increase and become significantly different from the control groupwhen measured two years after the intervention stopped. In other stu-dies, treatment effects decrease (Ferraro et al. 2011; Schultz et al. 2015;

30

16

0

18

50

15

14

7

20

40

212 6

7

210

1020

3040

50N

umbe

r of T

reat

men

ts

Energy Saving Tips

UsageHistory

GraphicalEnhancement

Framing Reminder Finan./Comp.Incentives

None

Social Comparison Goal SettingLabeling

Fig. 2. Additional interventions combined with social comparison, goal setting/commitment, and with labeling treatment.Note: For Jensen et al. (2016) we refer to model 1.

13

7

6

1

WTP for energyefficient appliance or house

Correct estimationof energy savings

Purchase probability ofenergy efficient appliance Energy consumption

Fig. 3. Analyzed outcomes in studies with a labeling treatment.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

183

Schultz et al. 2007; Staats et al. 2004), remain or become insignificant(Anderson et al. 2017; Alberts et al. 2016; Delmas and Lessem 2014;Dolan and Metcalfe 2013; Tiefenbeck et al. 2013; Schultz et al. 2007). Itwould be desirable to identify the driving factors for these hetero-geneous effects. Yet, based on the existing empirical evidence, we donot observe any obvious indications.9 As long-term effects are crucialfor the cost-effectiveness of the interventions, their analysis should beone focus of future research.

In sum, social comparison presents an effective treatment. Yet, re-searchers and policy makers should closely monitor potential adverseeffects on certain subgroups of households (below-average users) andon consumption patterns regarding other goods (moral licensing effect).Furthermore, the analysis of long-term effects and cost-benefit analysisshould be an inherent part of the impact evaluation.

4.2.2. Goal Setting and Commitment DevicesMany of the identified studies on goal setting and commitment

suffer from methodological shortcomings, despite satisfying the cri-terion of applying a methodology suitable to identify a causal effect. Forinstance, they test several treatments in their RCT despite relying onvery small study samples, or they do not report significance levels ofresults. Thus, the picture regarding an assessment of treatment effects isnot quite clear: exactly one half of the documented effects are notsignificantly different from zero or cannot be depicted as such becauseno significance level is reported (see Fig. 5). This may either be becausethe treatment actually has no effect, or that the study's features (in

particular a not sufficiently powered sample) contribute to the result.Interpreting the results of the identified studies, self-set goals seem

to result in a significant reduction effect when they are chosen realis-tically (Harding and Hsiaw 2014; Jaeger and Schultz 2017; McCalleyand Midden 2002; Winett et al. 1979). Harding and Hsiaw (2014) re-port based on a field study with more than 12,000 households that thesub-sample selecting the achievable goal of energy conservation be-tween 0% and 15% reduced their consumption by 11%. The authorsstill measure a significant reduction effect after 18months. At the sametime, the treatment group that selected a more optimistic goal(15%–50%) also saved energy shortly after the start of the program.Yet, their effort vanished after they received feedback about their usagedevelopment, presumably because they realized that it would be im-possible to achieve the ambitious goal. In contrast, both sub-sampleschoosing either a zero-goal or a more-than-50% goal showed no be-havior change at all. As we deem it important for policy advice, wewould like to clarify the interpretation of the results: while it is a validconclusion to state that people who set themselves realistic goals ex-hibit a conservation effect of 11%, the results of Harding and Hsiaw(2014) do not allow the conclusion that setting people realistic goalswill end up in substantial energy savings (of around 11%) because thiseffect is only valid for the self-selected group. People who are in generalmore motivated to conserve energy could have selected themselves intomore realistic goals, while people who are generally unwilling tochange their behavior chose zero. Similarly, Jaeger and Schultz (2017)only document (significant) effects for compliers in their study. How-ever, in order to identify an average conservation effect of the treatmentgoal setting, it would be necessary to estimate the average treatmenteffect for the full treatment group.

Goals that are externally set by the experimenters have resulted ineither insignificant effects or energy savings up to 22% compared tobaseline consumption (Fig. 5). Many of the reported significant effectsstem from Winett et al. (1982), who conducted two RCT studies, but

Fig. 4. Estimated effects of social comparison by medium.Note: Insignificant effects and effects of studies that do not report significance levels are depicted as nil-effects in this graph. Studies are sorted by treatment-medium and in alphabetical order. “sg” and“avg” stand for “study group” and “average”, respectively. Comment: The studies of Komatsu and Nishio (2015) and Kantola et al. (1984) are not included in the graph as they do not mention themagnitude of the measured significant conservation effects.

9 In a recent study, Brandon et al. (2017) explore the underlying mechanisms of long-run reductions in energy consumption induced by a social comparison (specifically theHER by Opower). Using data of 38 natural field experiments, they find that the physicalcapital channel (e.g. insulating the house or buying a more energy efficient washingmachine) is more important for the persistence of effects than habit formation within thehousehold.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

184

with a relatively small sample size of in total 83 households for fivetreatment groups. After having taken part in a meeting, researchersasked participants to sign a commitment to reduce their energy con-sumption by 15% within the next 35 days. In addition to frequentfeedback about their progress, households received an audio tape withtips on how to save energy. One study was conducted during winter, theother one during summer. All but one of six treatment groups showedsignificant and relatively high reduction effects ranging from 11% to22%.

In another study, an externally set goal of 10% resulted in a sig-nificant energy conservation effect of about 12.3% (van Houwelingenand van Raaj 1989). The effect remained significant one year after theintervention. Abrahamse et al. (2007) measure a smaller, but still sig-nificant effect of a 5%-goal in their RCT. In addition to these findings, alaboratory study by McCalley and Midden (2002) points to energysaving potential of externally set goals of around 20%.

On the other hand, Becker (1978) as well as Katzev and Johnson(1984) do not detect significant effects of external goal setting in thefield. Yet, in another study, Katzev and Johnson (1983) find that in-significant effects in the short-run become significant in the long-run.

In sum, goal setting and commitment still require substantial re-search efforts. The significant results point to a relatively high energyconservation potential of around 10%, while at the same time wecannot rule out that many of the insignificant results are due to un-derpowered samples. It seems to be a promising avenue for furtherresearch to evaluate the effect of energy conservation goals with a well-powered study set up, especially goals that are externally set.

4.2.3. LabelingAll but two of the identified studies on labeling report significant

results in at least a subsample of their study population. Regardingeffect sizes, they also point to potentially pronounced effects of la-beling.

Six studies evaluate the effect of labels under real-world conditions.Brounen and Kok (2011) document significantly higher sales prices forhouses with a “green” energy efficiency EU-label (classes A, B, C) thanfor comparable houses without such a label. In the same vein, Jensenet al. (2016) show that the price mark up for houses with efficient EU-labels (A to C) increased from 2.4% to 10.1% compared to houses withinefficient labels (D to G) after the energy performance rating was mademandatory in real estate advertisements in 2010. Houde (2014) exploitstwo natural experiments concerning the Energy Star Label. He observesa significantly higher WTP for refrigerators when they are awarded thelabel. His analysis is based on a comparison of the same models of re-frigerators before and after the criteria of the Energy Star Label weretightened. In the second study, the measured positive effect of the En-ergy Star Label on the WTP for refrigerators is not significant. In a si-milar vein, Allcott and Taubinsky (2015) do not find a significant effectof a label on the probability to purchase an energy-saving lightbulb intheir RCT in a big electrical store in the US. In the study by Kurz et al.(2005), which provided the treatment group with labels on the con-sumption levels of the household's own appliances, effects range fromzero to a reduced energy consumption of 23%.

The choice experiment of Heinzle (2012) shows that the WTP for amore efficient TV significantly increases when consumers are giveninformation on absolute operating costs over the course of ten years.Yet, it significantly decreases when the label provides information onannual operating costs. Newell and Siikamäki (2014) find in a furtherhypothetical choice experiment that insufficient information can lead toconsiderable undervaluation of energy efficiency. Simple informationon the monetary value of energy savings from a more efficient warmwater processor was the most important element guiding a cost-efficientenergy efficiency investment, with information on physical energy useand carbon dioxide emissions having additional but lesser importance.

The results of Waechter et al. (2015), moreover, indicate that en-ergy efficiency classes induce people to make decisions based on the

Fig. 5. Goal setting/commitment effects by self-set or externally set goal.Note: Insignificant effects and effects of studies that do not report significance levels are depicted as nil-effects in this graph. Studies are sorted by externally set and self-set goals/commitments and inalphabetical order. ‘sg’ stands for ‘study group’. Comment: The study by Jaeger and Schultz (2017) solely documents the effect for those individuals that comply with the commitment device and notfor the full treatment group.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

185

energy efficiency class even when concrete and readily apparent con-sumption information contradicts it. Indeed, in a recent study based onactual efficiency class ratings, Andor et al. (2017b) find that a majorityof participants value efficiency classes beyond the economic value ofthe underlying differences in energy usage. Furthermore, they showthat adding annual operating cost information to the EU energy labelpromotes the choice of energy-efficient durables. Their results also in-dicate that displaying operating cost affects choices through two dis-tinct channels: it increases the attention to operating cost and reducesthe valuation of efficiency class differences. Ölander and Thøgersen(2014) provide evidence for the positive effect of visualizing informa-tion in form of a simple label: changing an energy efficiency scale froma complex “A+++ - D”-system to a simpler “A – G”more than doubledthe probability that an energy-efficient device is chosen.

In sum, even though up to now there is not a vast amount of re-search on labeling, we identify a remarkable potential of the inter-vention. Not only do hypothetical purchase decisions in choice ex-periments confirm their effectiveness, but also evaluations of labels inthe field. Future research should focus on different elements of labels,like the framing of costs or the mode of ranking, and test their separateand combined effects in the field.

5. Discussion and Conclusion

This paper conducted a systematic review of behavioral interven-tions to induce residential energy conservation. In contrast to the ex-isting literature, this review focused on studies that permit the identi-fication of the causal relationship between the intervention and theoutcome of interest. In addition, it is the first review to cover labeling,an intervention that affects millions of people worldwide.

We find that all four reviewed interventions – social comparison,commitment devices, goal setting, and labeling – have the potential tosignificantly reduce energy consumption of private households. Whilethe vast majority of studies documents a significant reduction effect forthe social comparison intervention, results for the other interventionsare mixed. Social comparison has been the most researched interven-tion, in terms of both quality and quantity. In particular, the several“Opower studies” that investigate the causal effect of social comparisonbased home energy reports (HERs) on energy consumption deliverbroad empirical evidence and can be seen as best practice for programevaluations of energy conservation interventions. Yet, even for thisintervention, many open questions remain: What is the actual pureeffect of social comparison? Under which circumstances do socialcomparisons cause adverse effects? How much does the effect of socialcomparison interact with the medium by which it is delivered? Firstevidence on these questions indicate that adverse effects seem to matterand that it can make a difference if the social comparison is deliveredby a letter, electronically online or via an In-Home-Display (IHD).

Commitment devices and goal setting have not yet been researchedextensively and existing studies show major methodological short-comings mainly in terms of underpowered samples. Yet, studies thatdocument significant effects show conservation potential of around10%. It is thus a promising avenue for further research to evaluate theseinterventions with a sufficiently large study sample, preferably in thefield.

Although energy labels are applied worldwide, labeling is a veryrecent field of research. First evidence shows that labels can be effectivewith regard to the perception of and the willingness to pay (WTP) forenergy efficiency. Early results by laboratory experiments are con-firmed by some first field studies. Because the existing empirical evi-dence is mostly based on stated preference approaches, though, apromising field of future research is the evaluation of the effects oflabels by analyzing revealed preferences.

The review moreover clearly shows that the amount of studies that

satisfy the quality criterion of causal inference has increased in the lastfew years. There are presumably three reasons for this development: theimproving quality of empirical research, an increasing interest in be-havioral economics, and the need to find effective interventions totrigger energy conservation. However, there is still a lot to be done.Although we focused on studies that are potentially able to identifycausal effects, only few studies within the sample contain evidence toback up concrete policy recommendations. Obstacles are, for instance,insufficiently powered studies, poor reporting of statistical tests, andstudy populations that are different from the target population.Furthermore, the minority of studies discusses the benefits and costs ofthe intervention, a prerequisite to give policy recommendations.

To sum up, we are surprised how little we know. We therefore callfor systematic evaluations of these and similar interventions potentiallyable to trigger energy conservation before a large-scale rollout. It seemssurprising that an intervention such as labeling is applied worldwidebut there is little knowledge about the actual impact.

Future studies should focus on at least four crucial points: first, theyshould assess the causal effect of the intervention with a suitablemethodology and a sufficiently powered sample that enables identifi-cation of even small effects with statistical precision. Second, re-searchers should refrain from combining too many different treatments.From a research perspective, we need evidence on the pure effect of agiven intervention before we begin to potentially reinforce this effect byadditional treatments. Third, long-term effects of the interventionshould be analyzed and the intervention costs should be documented inorder to enable cost-benefit analyses on the treatment's effectiveness aswell as on its relative effectiveness compared to traditional policy in-struments, such as taxes or bans (cf. Benartzi et al. 2017).

Fourth, a growing literature discusses that for a final decisionwhether an intervention is welfare enhancing and thus should be ap-plied at large-scale, a classic cost–benefit analysis is not sufficient but awelfare analysis that takes into account all benefits and costs of an in-tervention is needed (cf. Allcott and Kessler 2015, Andor et al. 2017a,Ito 2015). Behavioral interventions might address externalities as wellas internalities but can also cause hidden costs (Damgaard and Gravert,2018). Allcott and Kessler (2015), for instance, show that while themajority of households has a positive WTP for continuing to receiveHERs in the future, a certain percentage of households reveals a ne-gative WTP. This indicates a disutility from receiving the HER. Futuresystematic evaluations should therefore aim at providing a compre-hensive welfare analysis. At best, interventions should be evaluated ex-ante before they are rolled out at large-scale, for instance by rando-mized field experiments within the target population.

Acknowledgements

This article is a further developed and extended extract of a projectreport published in German (Andor and Fels 2017). We are grateful forinvaluable comments and suggestions by the editor Dale S. Rothman,two anonymous referees, Gunther Bensch, Andreas Gerster, JörgLangbein, Jörg Peters, Christoph M. Schmidt, Stephan Sommer, andColin Vance. We also thank Marvin Gleue, Nadine Kneppel, MarcTeipel, and Maximilian Zettler for excellent research assistance. Wegratefully acknowledge financial support by the National Academy ofScience and Engineering (acatech). Furthermore, this work has beenpartly supported by the Collaborative Research Center “StatisticalModeling of Nonlinear Dynamic Processes” (SFB 823) of the GermanResearch Foundation (DFG), within Project A3, “Dynamic TechnologyModeling”. Fels gratefully acknowledges the support of a special grant(Sondertatbestand) from the German Federal Ministry for EconomicAffairs and Energy and the Ministry of Innovation, Science, and Re-search of the State of North Rhine-Westphalia.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

186

App

endixA

Theiden

tified

articles

have

been

catego

rizedacco

rdingto

theinterven

tion

ssocial

compa

rison,

commitmen

t/go

alsetting,

labe

ling,

andareorde

redba

sedon

theirpu

blicationda

tewithineach

catego

ry.n=

samplesize;n

.a.=

notav

ailable;

CG=

controlgrou

p;**

*,**

,*sign

ificanc

eleve

lsof

1%,5

%an

d10

%,r

espe

ctively.

TableA1

“Final

resultstable”

Stud

yTy

peof

interven

tion

[+co

mbina

tion

withad

dition

alinterven

tion

s]an

dinterven

tion

design

Effect(in%)in

compa

risonto:

controlgrou

p/ba

selin

e

nDep

ende

ntva

riab

leCou

ntry

Metho

dof

causal

analysis

Rem

arks

Social

compa

riso

nKan

tola

etal.(19

84)

(1)Fe

edba

ck,S

ocialCom

parison

[+Fram

ing,

Energy

saving

tips]

Hou

seho

ldswereinform

edon

ceby

letter

that

they

show

edan

abov

e-av

erag

eelectricity

consum

ptionde

spiteha

ving

stated

inan

earlier

survey

that

they

feel

anob

ligationto

save

electricity(the

message

pointedto

the

disson

ance

betw

eenan

swersin

thesurvey

and

theiractual

consum

ption)

+userswere

prov

ided

withpa

mph

lets

andcardson

energy

saving

tips

(1)

sign

ificantly

lower

electricity

consum

ption/

n.a.

118

Electricity

AU

RCT

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips]

Hou

seho

ldswereinform

edon

ceby

letter

that

they

show

edan

abov

e-av

erag

eelectricity

consum

ption+

userswereprov

ided

with

pamph

lets

andcardson

energy

saving

tips

(2)insign

if./

n.a.

(3)Con

trol

Group

[+En

ergy

saving

tips]

userswereprov

ided

withpa

mph

lets

andcards

onen

ergy

saving

tips

(3)insign

if./

n.a.

(4)Con

trol

Group

Participan

tsreceived

aTh

ank-You

letter

for

taking

part

intheexpe

rimen

t

(4)CG/n

.a.

Haa

kana

etal.(19

97)

(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips]

Participan

tsreceived

feed

back

andothe

rinform

ationacco

rdingto

theirwishe

s.Most

househ

olds

optedforco

mpa

risons

oftheirow

nen

ergy

consum

ptionwithsimila

rho

useh

olds

inFinlan

din

addition

tope

rson

alized

tips

andan

energy

conserva

tion

vide

o(2)

Feed

back,S

ocial

Com

parison

[+En

ergy

saving

tips]

Asin

treatm

ent(1)bu

twithtailo

red

inform

ationin

written

form

insteadof

avide

o

a)(1)n.a./

(−7)

(2)n.a./(−

5)(3)n.a./(−

5)(4)n.a./(+

1)

b)(1)n.a./

(−4)

(2)n.a./0

(3)n.a./(+

1)(4)n.a./(+

1)

105

a) Electricity

b)Water

c)Gas

FIRCT

Noinform

ationab

outsign

ificanc

eof

effects.

Long

-run

effects:

n.a.

Costs:

n.a.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

187

(3)Fe

edba

ck,S

ocialCom

parison

Asin

treatm

ent(1),bu

twitho

utan

ytailo

red

inform

ationor

energy

saving

tips

(4)Con

trol

Group

Participan

tsof

theCon

trol

Group

didno

tkn

owthey

werepa

rtof

anexpe

rimen

t

c)(1)n.a./

(−9)

(2)n.a./(−

7)(3)n.a./(−

4)(4)n.a./(−

3)

Staa

tset

al.(20

04)

(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips]

Group

sco

nsisting

ofsixto

tenpe

ople

met

as„E

coTe

ams”

once

amon

thto

discuss

environm

entally

releva

ntbe

havior.D

uringthe

sessions

they

received

feed

back

abou

ttheir

energy

saving

s.To

pics

atthemeeting

swere:

garbag

e,ga

s,electricity,

water,transpo

rtan

dco

nsum

erbe

havior

a)n.a./(−4.6)

[insignif.]

b)n.a./

(−20

.5)**

c)n.a./(−2.8)

[insignif.]

482

a) Electricity

b)Gas

c)Water

NL

Diff-in-Diff

Long

-run

effects:(2ye

arsafterstop

oftheEc

oTeam-m

eeting

s)(−

7.6)**

(−16

.9)**

(−6.7)**

Costs:

n.a.

Schu

ltzet

al.(20

07)

(1)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips]

Han

dwritten

message

swithinform

ationab

out

how

muc

helectricitytheho

useh

olds

consum

edin

theprev

ious

week+

descriptiveno

rmative

inform

ationab

outtheen

ergy

consum

ptionof

theav

erag

eho

useh

oldin

theirne

ighb

orho

od+

prep

rinted

sugg

estion

sforho

wto

conserve

energy

;sam

plerestricted

tohigh

users

(1)n.a./

(−5.68

)*28

7Electricity

US

RCT

Long

-run

effects:

(−4.61

)[insignif.]

(+9.66

)**

(−5.97

)*(+

1.0)

[insignif.]

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips]

Sameinterven

tion

asin

(1)bu

tsample

restricted

tolow

users

(2)n.a./

(+8.57

)*

(3)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips,G

raph

ical

enha

ncem

ent]

Sameinterven

tion

asin

(1)+

sadface,b

ecau

setheho

useh

olds'con

sumptionwas

abov

eav

erag

e

(3)n.a./

(−8.34

)**

(4)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips,G

raph

ical

Enha

ncem

ent]

Sameinterven

tion

asin

(2)+

happ

yface,

becausetheho

useh

olds

consum

edless

than

averag

e

(4)n.a./

(+2.32

)[insignif.]

Peschieraet

al.(20

10)

(1)Fe

edba

ck[+

Com

parisonto

usag

ehistory,

Rem

inde

r]Online-feed

back

abou

tindividu

alpa

stan

dpresen

telectricityco

nsum

ption+

threeweekly

reminde

rsto

checktheon

line-tool

(1)(−

16)

[insignif.]

/(−

30)

[insignif.]

83Electricity

US

RCT

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Rem

inde

r](2)(−

6)[insignif.]

/

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

188

Asin

treatm

ent(1)+

compa

risonof

the

individu

alelectricityco

nsum

ptionwiththe

averag

eco

nsum

ptionof

therest

ofthebu

ilding

(−22

)[insignif.]

(3)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Rem

inde

r]Asin

treatm

ent(2)+

inform

ationab

outthe

electricityco

nsum

ptionof

“peers”(i.e.

occu

pantsof

thebu

ildingwho

inapre-survey

thetreatedho

useh

olds

classified

as“kno

wn”

)

(3)(−

21)***

/(−

34)***

(4)Con

trol

grou

pPa

rticipan

tskn

ewthey

werepa

rtof

anexpe

rimen

t

(4)CG/n

.a.

Allc

ott(201

1)(1)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips]

Mon

thly/b

imon

thly/q

uarterly

feed

back

via

mail(H

omeEn

ergy

Rep

ortwithpe

rson

alhistoryof

consum

ption,

compa

risonto

neighb

orsan

dtips

tosave

energy

)(2)Con

trol

grou

pNoinform

ationab

outpa

rticipationin

stud

y

(1)

(−2.03

)***

/n.a.

(2)CG/n

.a.

588,44

6Electricity

US

RCT

Short-term

effectmeasuredon

eye

arafterinterven

tion

.

Long

-run

effects:

According

totheau

thor

thereis

noev

iden

ceof

aredu

ctionof

theeff

ect

withthetreatm

entlastingfortw

oye

ars,

butno

specificnu

mbe

rsstated

(altho

ugh,

seeAllc

ottan

dRog

ers

(201

4))

Costs:

Costsof

prod

uction

andshipping

ofthe

repo

rtsdivide

dby

kWhof

save

den

ergy

:3.31cents/kW

hFe

rraroet

al.(20

11)

+Fe

rraroan

dPrice(201

3)

(1)Con

trol

grou

p[+

Energy

saving

tips]

One

timeletter

withtips

tosave

energy

(1)

(−0.66

)[insignif.]

/(−

8.41

)

106,66

9Water

US

RCT

Long

-run

effects:

(measuredtw

oye

arsafter

interven

tion

)(+

0.9)

[insignif.]

(−0.22

)[insignif.]

(−1.3)**

Costs:

$0.57

5pe

rsave

d10

00ga

l(=

3785

.41L)

ofwater

(2)Fe

edba

ck[+

Energy

saving

tips,F

raming(soc

ialno

rm)]

One

timepe

rson

alized

letter

withhe

avily

norm

-ba

sedlang

uage

abou

tsaving

water,feedb

ack

from

theco

nsum

ptionbill,

tips

tosave

energy

(2)(−

2.7)**

*/(−

10.08)

(3)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,F

raming(soc

ialno

rm)]

Like

(2)plus

aco

mpa

risonto

themed

ianwater

consum

ptionof

thepreced

ingsummer

anda

classification

oftheho

useh

oldinto

oneof

the

consum

ptiongrou

ps

(3)(−

4.8)**

*/(−

12.01)

(4)Con

trol

grou

pNoinform

ationab

outpa

rticipationin

expe

rimen

t

(4)CG/

(−7.83

)

Peschieraan

dTa

ylor

(201

2)(1)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory]

Participan

tsge

taccess

totheirelectricity

consum

ptionda

tathroug

han

onlin

e-tool

+inform

ationab

outco

nsum

ptionof

the

(1)0

[insignif.]

/n.a.

88Electricity

US

RCT

Effects

arecalculated

forab

ove-

averag

euserson

ly.Ifbe

low-ave

rage

usersareinclud

edin

thesample,

the

effects

turn

insign

ificant.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

189

last

seve

nda

ys+

aco

mpa

risonwiththe

averag

ereside

ntialelectricityuse+

weekly

reminde

rsto

checktheon

line-tool

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory]

Sameinterven

tion

asin

(1)+

data

abou

tthe

electricityuseof

‘peers’"(i.e.the

seoc

cupa

ntsof

thebu

ildingwho

thetreatedho

useh

olds

men

tion

edto

"kno

w"in

apre-survey

)"

(2)(−

8.8)**

/n.a.

(3)Con

trol

grou

pPa

rticipan

tskn

ewthey

werepa

rtof

anexpe

rimen

t

(3)CG/n

.a.

Ayres

etal.(20

13)

(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,c

ompa

risonto

usag

ehistory,

grap

hicalen

hanc

emen

t]Hom

eEn

ergy

Rep

orts

(HER

)ab

outelectricity

consum

ptionviamail,mon

thly

forhe

avyusers,

quarterlyforlig

htusers,

withtips

tosave

energy

,persona

lhistoryof

consum

ptionan

dco

mpa

risonto

neighb

ors(add

itiona

llylaug

hing

orsadsm

iley,

depe

ndingon

consum

ptionbe

ing

below-or

abov

e-av

erag

e)

(1)

(−2.02

)***

/n.a.

84,000

(1)+

(2):

Electricity

(3)Gas

US

RCT

Long

-run

effects:

n.a.

Costs:

4.94

Cen

tspe

rkW

hsave

d1.78

Cen

tspe

rkW

hsave

dn.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,C

ompa

risonto

usag

ehistory,

Graph

ical

enha

ncem

ent]

Like

(1),theHER

addition

ally

containe

dinform

ationab

outthega

sco

nsum

ption,

freq

uenc

yrand

om(m

onthly/q

uarterly)

(2)

(−1.22

)***

/n.a.

(3)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,C

ompa

risonto

usag

ehistory,

Graph

ical

enha

ncem

ent]

Like

(2),ga

sco

nsum

ptionmeasuredinsteadof

electricityco

nsum

ption

(3)

(−1.20

)***

/n.a.

(4)Con

trol

grou

pNoinform

ationab

outpa

rticipationin

expe

rimen

t

(4)CG/n

.a.

Costa

andKah

n(201

3)(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,C

ompa

risonto

usag

ehistory]

Hom

eEn

ergy

Rep

orts

(HER

)ab

outelectricity

consum

ptionviamail,mon

thly

forhe

avyusers,

quarterlyforlig

htusers,

withtips

tosave

energy

,persona

lhistoryof

consum

ptionan

dco

mpa

risonto

neighb

ors

(1)Libe

rals:

(−2.4)**

*/n.a.

Con

servatives:

(−1.7)**

*/n.a.

81,772

Electricity

US

RCT

Thestud

yespe

cially

analyzes

the

interven

tion

'she

teroge

neou

seff

ects

("Libe

rals"vs."Con

servatives").T

heav

erag

etreatm

enteff

ectis

(−2.1)**

*co

mpa

redto

theco

ntrolgrou

p.

Long

-run

effects:

n.a.

Costs:

According

totheau

thors,

the

interven

tion

might

beco

st-effective

,bu

tno

specificstatem

ent.

(2)Con

trol

Group

Noinform

ationab

outpa

rticipationin

expe

rimen

t

(2)CG/n

.a.

Dolan

andMetcalfe

(201

3)(1)(−

4.4)**

*/n.a.

569

Gas

GB

RCT

Expe

rimen

t2of

thearticlefits

them

atically

andis

gene

rally

suitab

le

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

190

(1)Fe

edba

ck,S

ocialCom

parison

Bian

nual

feed

back

andco

mpa

risonto

neighb

ors'co

nsum

ptionviamail

toiden

tify

causal

effects,bu

tthe

inform

ationin

theworking

pape

ris

contradictoryan

dis

thereforeno

tlistedhe

re.

Long

-run

effects:

(18mon

thsafterfirstInterven

tion

)(−

7.0)*

(−6.0)

[insignif.]

Costs:

According

totheau

thors33

3kW

hweresave

dforev

eryPo

undspen

t

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips]

Like

(1)+

Tips

tosave

energy

(2)

(−10

.8)***

/n.a.

(3)Con

trol

Group

Feed

back

abou

tpe

rson

alelectricity

consum

ption

(3)CG/n

.a.

Mizob

uchi

andTa

keuc

hi(201

3)(1)Fe

edba

ck[+

Fina

ncialIncentive,

Com

parisonto

usag

ehistory]

Mon

thly

feed

back

viamail,rewardof

200Yen

(2$)

for1%

redu

ctionof

energy

consum

ption

(1)(−

4.15

)**/

(−5.87

6)**

208

Electricity

JPRCT

Effects

asdifferen

cefrom

pre-

treatm

ent-co

nsum

ption.

Thedifferen

cebe

tweentheeff

ects

of(1)an

d(2)is

statistically

notsign

ificant.

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+Fina

ncialIncentive,

Com

parisonto

usag

ehistory]

Mon

thly

feed

back

viamail,co

mpa

risonto

othe

rpa

rticipatingho

useh

olds,r

ewardof

200Yen

(2$)

for1%

redu

ctionof

energy

consum

ption

(2)(−

6.48

)**/

(−8.19

6)**

(3)Con

trol

grou

pInform

ationab

outpa

rticipationin

stud

y(3)CG/

(−1.72

1)Tiefen

beck

etal.(20

13)

(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,F

raming(soc

ialno

rm)]

Weeklyfeed

back

abou

tthewater

consum

ption

percapita

andtips

tosave

water,p

artly

complem

entedby

asocial

compa

risonwiththe

10%

saving

themostw

ater

orasocial

appe

alto

contribu

teon

e'sshareto

themutua

lgo

alof

energy

conserva

tion

(1)a)

(−6.0)**

/n.a.

b)(+

5.6)**

/n.a.

154

a)Water

b) Electricity

US

RCT

Stud

yshow

sad

verseeff

ectof

(inten

ded)

saving

sin

water

consum

ptionforco

nsum

ptionof

electricity.

Sinc

etena

ntsprivatelypa

yforga

san

delectricity,

while

water

isbille

dco

llectively,

inco

meeff

ects

can

beexclud

ed.

Water

consum

ptionwas

measured

daily

,electricity

consum

ptionweekly.

Long

-run

effects:

Nomoresign

ificant

effects

measured

aftertw

oweeks:

Water:(−5.5)

Electricity:

(+0.3)

Costs:

n.a.

(2)Con

trol

grou

pInform

ationthat

water

consum

ptionwill

bemon

itored

intheco

urse

ofascientificstud

y

(2)CG/n

.a.

Allc

ottan

dRog

ers

(201

4)(1)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips]

Mon

thly

feed

back

viamail(H

omeEn

ergy

Rep

ortwithpe

rson

alhistoryof

consum

ption,

compa

risonto

neighb

orsan

dtips

tosave

energy

)

(1)(−

1.7)**

*/n.a.

234,00

0Electricity

US

RCT

Long

-run

effects:

2009

–201

3(followingthe

measuremen

tof

theeff

ectafter

receivingfour

repo

rts,

thetreatm

ent

grou

pswererand

omly

re-allo

cated):

Stop

oftheinterven

tion

aftertw

oye

ars:

(−2)**

*

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

191

Bian

nual

receiptof

Hom

eEn

ergy

Rep

orts:(−

3.1)**

*Receipt

oftheHom

eEn

ergy

Rep

orts

ininitialfreq

uenc

y:(−

3.3)**

*Costs:

Costsof

prod

uction

andshipping

ofthe

repo

rtsdivide

dby

kWhof

save

den

ergy

Assum

ingthat

thesaving

s-eff

ectdo

esno

tlast:

3.2–

4.44

cents/kW

hAssum

ingthat

theeff

ects

last

forthe

long

term

:1.35

–1.79cents/kW

h

(2)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Energy

saving

tips]

Qua

rterly

feed

back

viamail(H

omeEn

ergy

Rep

ortlik

ein

(1))

(2)(−

1.2)**

*/n.a.

(3)Con

trol

Group

Noinform

ationab

outpa

rticipationin

stud

y(3)CG/n

.a.

Delmas

andLe

ssem

(201

4)(1)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Rem

inde

r]Real-tim

efeed

back

andco

mpa

risonwithothe

rusersthroug

han

onlin

e-platform

,weekly

reminde

rviae-mail

(1)(−

5.68

)[insignif.]

/n.a.

66Electricity

US

RCT

Long

-run

effects:1

7weeks

afterfirst

interven

tion

(−6.5)

[insignif.]

(−24

.76)*

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+Com

parisonto

usag

ehistory,

Rem

inde

r]Add

itiona

llyto

(1)pu

blic

rank

ings

ofwhich

stud

entroom

sco

nsum

ebe

low-or

abov

e-av

erag

e(via

postersin

theen

tran

ceha

llan

dvia

e-mail)

(2)

(−19

.36)**

/n.a.

(3)Con

trol

grou

pNoinform

ationab

outpa

rticipationin

expe

rimen

t

(3)CG/n

.a.

Kom

atsu

andNishio

(201

5)(1)Fe

edba

ck,S

ocialCom

parison

One

timefeed

back

viamailan

dco

mpa

risonto

med

ianco

nsum

ption

(1)Noeff

ect

sizesgive

n;tend

encies

for

assessmen

tof

own

consum

ption:

high

er**

3,03

3Electricity

(ind

irect)

JPRCT

In(1)–(3)thepa

rticipan

ts'a

ssessm

ent

oftheirow

nelectricityco

nsum

ptionin

compa

risonto

thene

ighb

ors'

consum

ptionwas

analyzed

,in(4)–(6)

itwas

theho

useh

olds´motivationto

save

energy

.

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

Like

(1)plus

compa

risonto

high

estsaving

25%

ofpa

rticipan

ts

(2)high

er**

*

(3)Fe

edba

ck,S

ocialCom

parison

[+Fram

ing(soc

ialno

rm)]

Like

(2)plus

amessage

abou

tthesocial

acceptan

ce/d

isap

prov

alof

ownco

nsum

ption

(3)high

er**

*

(4)Fe

edba

ck,S

ocialCom

parison

Like

(1)

(4)Motivation

tosave

energy

:insign

if.

(5)Fe

edba

ck,S

ocialCom

parison

Like

(2)

(5)insign

if.

(6)high

er**

*

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

192

(6)Fe

edba

ck,S

ocialCom

parison

[+Fram

ing(soc

ialno

rm)]

Like

(3)

(7)Con

trol

grou

pSimplefeed

back

ofelectricityco

nsum

ption

(7)n.a.

Schu

ltzet

al.(20

15)

(1)Fe

edba

ck[+

Energy

saving

tips]

Real-tim

efeed

back

viaIH

Dan

dprov

isionof

a“clim

ateprotection

vide

o”

(1)(−

3.0)

[insignif.]

/n.a.

431

Electricity

US

RCT

Long

-run

effects:

(three

mon

thsafterInterven

tion

)(−

0.81

)[insignif.]

(+1.13

)[insignif.]

(−7.02

)**

Costs:

n.a.

(2)Fe

edba

ck[+

Energy

saving

tips]

Real-tim

efeed

back

viaIH

Dwiththe

consum

ptionbe

ingco

nvertedto

actual

costs,

vide

o

(2)insign

if./

n.a.

(3)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips]

Real-tim

efeed

back

viaIH

Dan

dco

mpa

risonto

othe

rpa

rticipan

ts,v

ideo

(3)(−

9.0)**

/n.a.

(4)Con

trol

Group

Onlyvide

o,no

feed

back

(4)CG/n

.a.

Seyran

ianet

al.(20

15)

(1)Con

trol

Group

[+En

ergy

saving

tips]

One

timeletter

withtips

tosave

energy

(1)CG/

(+40

.8)

[insignif.]

374

Water

US

RCT

Con

sumptionwas

measuredon

eweek

aftertheinterven

tion

Long

-run

effects:

(fou

rweeks

afterinterven

tion

)CG/(+

16.2)[insignif.]

(−11

.5)**/(+

2.6)

[insignif.]

(−12

.2)**/(+

2.6)

[insignif.]

(−12

.1)***

/(+

3.5)

[insignif.]

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,G

raph

ical

enha

ncem

ent]

Like

(1)plus

inform

ationab

outow

npe

rson

alco

nsum

ptionin

compa

risonto

averag

eco

nsum

ptionof

neighb

orho

od,co

mplem

ented

byalaug

hing

orsadsm

iley

(2)(−

29.74)*/

(−0.1)

(3)Con

trol

Group

[+En

ergy

saving

tips,F

raming(soc

ialno

rm)]

Like

(1)plus

aco

verletter

withthecity

logo

,em

phasizingtheho

useh

old'srole

asapa

rtof

theco

mmun

ityan

dstressingon

water

conserva

tion

asamutua

lgo

al

(3)(−

34.1)**/

(+1.4)

[insignif.]

(4)Con

trol

Group

[+En

ergy

saving

tips,F

raming(persona

lno

rm)]

Like

(1)plus

aco

verletter,s

etting

water

conserva

tion

asago

al,b

utaccentua

ting

the

househ

oldas

asing

leplay

eran

dno

tasapa

rtof

theco

mmun

ity

(4)(−

24.6)

[insignif.]

/(+

3.4)

[insignif.]

Alberts

etal.(20

16)

(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,G

raph

ical

enha

ncem

ent]

Weeklyem

aileden

ergy

repo

rtco

nsisting

ofen

ergy

conserva

tion

tips

andinform

ationab

out

one'sow

nco

nsum

ptionin

compa

risonto

(1)(−

20.9)**/

n.a.

89Electricity

andga

sUK

RCT

Long

-run

effects:

(twoweeks

afterstartof

interven

tion

)(−

30.8)**/n.a.

(−9.9)

[insignif.]

/n.a.

Costs:

n.a.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

193

averag

eco

nsum

ptionof

allne

ighb

orsan

dto

mosteffi

cien

t20

%;w

ithrating

as“great”,

“goo

d”or

“morethan

averag

e”acco

mpa

nied

byasm

ileyface

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,G

raph

ical

enha

ncem

ent,

Com

petition

incentive]

Like

(2)plus

inform

ationthat

thereis

aco

mpe

tition

tofind

thereside

ntwiththelowest

energy

consum

ptionov

ertheco

urse

ofthe

project,althou

ghno

specificpricewas

commun

icated

(2)(−

21.2)*/

n.a.

(3)Con

trol

Group

[+En

ergy

saving

tips]

One

-sho

te-mailat

thestartof

theexpe

rimen

tco

ntaining

aseries

ofsimpleen

ergy

saving

tips

(3)CG/n

.a.

And

ersonet

al.(20

17)

Stud

y1(gradu

atestud

ents):(1)

Con

trol

Group

[+Fe

edba

ck,E

nergysaving

tips]

Weeklymessage

sem

ails

withco

mmon

energy

usefeed

back

inform

ationalon

gwithafew

energy

conserva

tion

tips

(1)CG/n

.a.

146

Electricity

KR

RCT

Long

-run

effects:

(eleve

nweeks

afterstartof

interven

tion

)CG

insign

if./n.a.

CG

insign

if./n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,g

raph

ical

enha

ncem

ent]

Like

(1)plus

inform

ationab

outav

erag

eco

nsum

ptionof

othe

rsimila

rreside

ntsan

dof

themosteffi

cien

t10

%,a

ndasupp

orting

message

plus

starsde

pend

ingon

therelative

performan

ce.

(2)

perweekof

messaging

(−1.2)**

/n.a.

Stud

y2(und

ergrad

uate

stud

ents):

(3)Con

trol

Group

[+Fe

edba

ck,E

nergysaving

tips]

Like

(1)

(3)CG/n

.a.

118

Electricity

KR

RCT

(4)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,g

raph

ical

enha

ncem

ent]

Like

(2)

(4)insign

if./

n.a.

Tiefen

beck

etal.(20

16)

(1)Fe

edba

ck[+

Graph

ical

enha

ncem

ent]

Smartshow

ermeter

that

displays

water

usein

tenthof

liters,

water

tempe

rature

inde

gree

Celsius,en

ergy

consum

ptionin

kWh,

energy

efficien

cyrating

(from

Ato

G),an

dapo

larbe

aran

imation(ice

floe

that

prog

ressivelyshrink

sas

theam

ount

ofen

ergy

used

increases)

during

show

er

(1)

(−22

.03)**

*/n.a.

636

Electricity

CH

RCT+

Di ff-in-

Diff

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+Graph

ical

enha

ncem

ent]

Like

(1)plus

totalam

ount

ofwater

used

inthe

(2)

(−22

.52)**

*/n.a.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

194

prev

ious

show

er(soc

.com

p.ifmorethan

one

person

lives

intheho

useh

old)

(3)Con

trol

Group

Smartshow

ermeter

displays

water

tempe

rature

only

CG/n

.a.

Hah

net

al.(20

16)

(1)Con

trol

Group

[+”gain”

-Framing]

Letter

containing

$100

reba

teco

upon

for

purcha

sing

drou

ght-resistan

tplan

tsph

rased

“You

canha

veafree

$100

land

scap

eco

upon

”

CG/n

.a.

23,282

Water

US

RCT

Thestud

ylook

sat

threeou

tcom

es:

numbe

rof

reba

tesrequ

ested,

numbe

rof

reba

tesrede

emed

,and

chan

gesin

mon

thly

water

use.

Thistableon

lylists

theeff

ects

onwater

use.

Long

-run

effects:

n.a.

Costs:

n.a.

fordirect

water

use,

$10.64

perthou

sand

gallo

nsof

estimated

water

saving

sviareba

teco

upon

s

(2)So

cial

Com

parison

[+”gain”

-Framing]

Like

(1)plus

asocial

compa

risonph

rased“W

eha

veselected

yoube

causeyo

uused

XX,XXX

gallo

nsmorewater

in20

13than

theav

erag

eho

me…

”

(2)(−

1.4)**

/n.a.

(3)Con

trol

Group

[+”loss”-Framing]

Like

(1)bu

tthereba

teco

upon

swereph

rased

“Dono

tlose

your

chan

ceto

have

afree

$100

land

scap

eco

upon

”

(3)CG/n

.a.

(4)So

cial

Com

parison

[+”loss”-Framing]

Like

(2)plus

asocial

compa

risonph

rased‘W

eha

veselected

yoube

causeyo

uused

XX,XXX

gallo

nsmorewater

in20

13than

theav

erag

eho

me…

’

(4)(−

1.1)

[insignif.]

/n.a.

Shen

etal.(20

15)

(1)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,G

raph

ical

enha

ncem

ent]

Mailedho

meen

ergy

repo

rtswithen

ergy

consum

ptionco

mpa

redto

averag

eco

nsum

ptionof

theentireneighb

orho

odan

dof

mosteffi

cien

t20

%withrating

as‘great’,‘goo

d’or

‘morethan

averag

e ’acco

mpa

nied

bysm

iley

face

anden

ergy

conserva

tion

tips.

(1)(−

2.8)

[insignif.]

/n.a.

475

Electricity

US

RCT

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,G

raph

ical

enha

ncem

ent]

Like

(1)bu

tco

mpa

risongrou

pwerestreet

neighb

ors

(2)(−

5.4)**

*/n.a.

(3)Fe

edba

ck,S

ocialCom

parison

[+En

ergy

saving

tips,G

raph

ical

enha

ncem

ent]

Like

(1)bu

tco

mpa

risongrou

pwerene

xt-doo

rne

ighb

ors

(3)(−

2.1)

[insignif.]

/n.a.

(4)Con

trol

grou

pNoinform

ationab

outpa

rticipationin

anexpe

rimen

t

CG/n

.a.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

195

Com

mitmen

tde

vice

san

dgo

alsetting

Becker

(197

8)(1)Goa

lSe

tting

[+En

ergy

saving

tips]

Participan

tswereaske

dto

redu

cetheiren

ergy

consum

ptionby

20%

compa

redto

their

baselin

eusag

e+

they

received

energy

conserva

tion

tips

specificto

theirow

nho

useh

oldap

plianc

es

(1)(−

1.3)

[insignif.]

/n.a.

100

Electricity

US

RCT

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,G

oalSe

tting

[+En

ergy

saving

tips]

Sameinterven

tion

asin

(1)+

they

received

feed

back

abou

ttheiren

ergy

consum

ptionthree

times

aweek

(2)(−

13)

[insignif.]

/n.a.

(3)Goa

lSe

tting

[+En

ergy

saving

tips]

Participan

tswereaske

dto

redu

cetheiren

ergy

consum

ptionby

2%co

mpa

redto

theirba

selin

eusag

e+

they

received

energy

conserva

tion

tips

specificto

theirow

nho

useh

oldap

plianc

es

(3)(+

1.2)

[insignif.]

/n.a.

(4)Fe

edba

ck,G

oalSe

tting

[+En

ergy

saving

tips]

Sameinterven

tion

asin

(3)+

they

received

feed

back

abou

ttheiren

ergy

consum

ptionthree

times

aweek

(4)(−

4.6)

[insignif.]

/n.a.

(5)Con

trol

grou

pPa

rticipan

tsof

theco

ntrolgrou

pkn

ewthey

werepa

rtof

anexpe

rimen

t

(5)CG/n

.a.

Winettet

al.(19

79)

(1)Fe

edba

ck,G

oalSe

tting

[+Com

parisonto

usag

ehistory,

Graph

ical

enha

ncem

ent,En

ergy

saving

tips]

Daily

feed

back

sheetwithinform

ationab

out

electricityusag

e+

compa

risonwithow

nusag

eon

thepreceeding

day+

aha

ppyor

frow

ning

smile

yforade

crease

orincrease

inco

nsum

ption+

usag

ech

ange

compa

redto

baselin

e+

feed

back

whe

ther

they

achiev

eda

goal,w

hich

they

hadsetthem

selves

ina

meeting

before

thestartof

the

expe

rimen

t+en

ergy

conserva

tion

inform

ation

(1)(−

13)***

/n.a.

71Electricity

US

RCT

Long

-run

effects:

(1)

(−11

)[insignif.]

(2)

(−7)

[insignif.]

Effects

arecalculated

relative

tothe

meanco

nsum

ptionof

thetw

oco

ntrol

grou

ps.

Costs:

Feed

back

cond

ition:

–totalexpe

nditurepe

rho

useh

old=

$26

–Sa

ving

spe

rho

useh

old=

$44from

expe

cted

expe

nditures

basedon

the

compa

risongrou

p'susedu

ring

this

samepe

riod

andthemargina

lco

stpe

rkW

h.Se

lf-m

onitoring:

–totalexpe

nditurepe

rho

useh

old=

$22

–Sa

ving

spe

rho

useh

old=

$26

(2)Fe

edba

ck[+

Energy

saving

tips]

Hou

seho

ldsha

dto

read

andrepo

rttheirda

ilyelectricityco

nsum

ption+

received

energy

saving

inform

ation

(2)(−

7)*/n.a.

(3)Con

trol

Group

Hou

seho

ldsag

reed

topa

rticipatein

thestud

y(3)CG/n

.a.

(4)Con

trol

Group

Hou

seho

ldsde

nied

topa

rticipatein

thestud

y(4)CG/n

.a.

Winettet

al.(19

82)

83Electricity

US

RCT

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

196

Stud

y1(duringwinter):

(1)Fe

edba

ck,G

oalSe

tting

[+En

ergy

saving

tips]

Participan

tstook

partin

ameeting

andreceived

inform

ationab

outen

ergy

conserva

tion

+they

hadto

sign

aco

mmitmen

tto

redu

cetheir

energy

consum

ptionby

15%

withinthene

xt35

days

+they

received

written

feed

back

abou

ttheiren

ergy

use+

they

wereprov

ided

witha

tape

which

presen

tedinform

ationon

energy

conserva

tion

inform

ofadiscussion

(1)n.a./

(−13

)***

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Goa

lSe

tting

[+En

ergy

saving

tips]

Participan

tstook

partin

ameeting

andreceived

inform

ationab

outen

ergy

conserva

tion

+they

hadto

sign

aco

mmitmen

tto

redu

cetheir

energy

consum

ptionby

15%

withinthene

xt35

days

+they

received

avide

owhich

presen

tedmod

elho

mes

simila

rto

the

participan

ts´ho

mes

andexplaine

dpo

ssible

way

sto

redu

ceen

ergy

consum

ption.

(2)n.a./

(−11

)***

(3)Fe

edba

ck,G

oalSe

tting

[+En

ergy

saving

tips]

Asin

treatm

ent(2)+

feed

back

(3)n.a./

(−14

)***

(4)Goa

l-Setting

[+En

ergy

saving

tips]

Asin

treatm

ent(1)bu

twitho

utfeed

back

(4)n.a./(−

1)[insignif.]

(5)Con

trol

Group

Participan

tsof

theco

ntrolgrou

pkn

ewthey

werepa

rtof

anexpe

rimen

t

(5)CG/(+

4)

Stud

y2(duringsummer):

(6)Fe

edba

ck,G

oalSe

tting

Asin

treatm

ent(3)bu

twitho

utinform

ationon

energy

conserva

tion

(6)n.a./

(−19

)***

(7)Fe

edba

ck,G

oalSe

tting

[+En

ergy

saving

tips]

Asin

treatm

ent(1)bu

twitho

utdiscussion

tape

(7)n.a./

(−22

)***

(8)Goa

lSe

tting

[+En

ergy

saving

tips]

Asin

treatm

ent(2)

(8)n.a./(−

12)

[insignif.]

(9)Con

trol

Group

Participan

tsof

theco

ntrolgrou

pkn

ewthey

werepa

rtof

anexpe

rimen

t

(9)CG/(−

2)

Katzevan

dJo

hnson

(198

3)(1)Goa

lSe

tting

„Foo

t-in-the

-doo

r-treatm

ent”:H

ouseho

ldswere

aske

dto

fillin

anen

ergy

conserva

tion

question

naire.

Afterwards,the

yweretold

toredu

cetheirelectricityco

nsum

ptionby

10%

withinthene

xttw

oweeks

(1)n.a./(+

11)

[insignif.]

66Electricity

US

RCT

Thesubjects

wererecruitedthroug

ha

door-to-do

orsolic

itationproc

edurein

which

they

wereaske

dforpe

rmission

toread

theirelectricitymetersas

part

ofastud

yon

reside

ntialen

ergy

consum

ption.

Long

-run

effects:

(2)Goa

lSe

tting

Hou

seho

ldswereaske

dto

redu

cetheir

(2)n.a./(+

12)

[insignif.]

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

197

(+1)**

(−2)**

*(−

5)**

(+5)

(alleff

ects

compa

redto

baselin

epe

riod

;sign

ificanc

eco

mpa

redto

controlgrou

p)Costs:

n.a.

electricityco

nsum

ptionby

10%

withinthene

xttw

oweeks

(3)Con

trol

Group

(3)n.a./(+

7)[insignif.]

(4)Con

trol

Group

Hou

seho

ldsag

reed

toha

vetheirelectricity

metersread

bytheexpe

rimen

ter

(4)CG/0

Katzevan

dJo

hnson

(198

4)(1)Goa

lSe

tting

[+En

ergy

saving

tips]

Hou

seho

ldswereaske

dto

redu

cetheiren

ergy

consum

ptionby

15%

withinthene

xttw

oweeks

+received

autility

guideforho

useh

old

energy

conserva

tion

(1)(−

6.1)

[insignif.]

/(−

13.4)

[insignif.]

90Electricity

US

RCT

Long

-run

effects:

(1)

(−2.8)/(−14

.3)

(2)

(−1.3)/(−12

.8)

(3)

(+0.2)/(−11

.3)

(4)

(−10

.7)/(−

22.2)

(5)

(−1.5)/(−13

.0)

(6)

CG/(−11

.5)

Alleff

ects

areinsign

if.

Costs:

n.a.

(2)Goa

lSe

tting

Hou

seho

ldswereaske

dto

redu

cetheiren

ergy

consum

ptionby

15%

withinthene

xttw

oweeks

+ha

dto

fillin

anen

ergy

conserva

tion

survey

(2)(+

1.4)

[insignif.]

/(−

5.9)

[insignif.]

(3)Con

trol

Group

[+Fina

ncialincentive,

Energy

saving

tips]

Hou

seho

ldsreceived

afina

ncialreward

depe

ndingon

theam

ount

ofelectricitysave

d+

autility

guideforho

useh

olden

ergy

conserva

tion

(3)(+

2.0)

[insignif.]

/(−

5.3)

[insignif.]

(4)Goa

lSe

tting

[+Fina

ncialincentive,

Energy

saving

tips]

Hou

seho

ldsreceived

aco

mbina

tion

ofthe

interven

tion

sin

(2)an

d(3)

(4)(−

3.7)

[insignif.]

/(−

11.0)

[insignif.]

(5)Con

trol

Group

Hou

seho

ldswereaske

dto

fillin

ashorten

ergy

conserva

tion

survey

(5)(+

7.5)

[insignif.]

/(+

0.2)

[insignif.]

(6)Con

trol

Group

Hou

seho

ldsag

reed

toha

vetheirelectricity

metersread

bytheexpe

rimen

ters

(6)CG/(−

7.3)

[insignif.]

vanHou

welinge

nan

dva

nRaa

ij(198

9)(1)Fe

edba

ck,G

oalSe

tting

[+En

ergy

saving

tips]

SmartM

eter

andIH

Dswereinstalledan

dho

useh

olds

wereaske

dto

mon

thly

jotdo

wn

theirpreferredan

dactual

consum

ptionof

gas.

Externally

impo

sedgo

al:1

0%co

nsum

ption

redu

ctionin

compa

risonto

preced

ingye

ar.

Add

itiona

llytips

tosave

energy

.

(1)n.a./

(−12

.3)***

325

Gas

NL

RCT

Long

-run

effects:(on

eye

arafteren

dof

theinterven

tion

)(1)

insign

if./(−

2.1)*

(2)

insign

if./(−

3.2)**

(3)

n.a./(−1.5)

(insignif.)

(4)

n.a./(+

1.4)

(insignif.)

(5)

n.a./(−2.2)*

(6)

n.a./(−2.9)*

Costs:

n.a.

(2)Fe

edba

ck[+

Energy

saving

tips]

Mon

thly

feed

back

(not

stated

throug

hwhich

cana

l)an

dtips

tosave

energy

(2)n.a./

(−7.7)**

*

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

198

(3)Fe

edba

ck[+

Energy

saving

tips]

Hou

seho

ldswereaske

dto

docu

men

ttheir

mon

thly

gasco

nsum

ptionon

observation

sheets,a

dditiona

llytips

tosave

energy

(3)n.a./

(−5.1)**

*

(4)Con

trol

Group

[+En

ergy

saving

tips]

Prov

isionof

tips

tosave

energy

(4)n.a./

(−4.3)**

(5)Con

trol

Group

Noinform

ationab

outpa

rticipationin

expe

rimen

t

(5)n.a./(−

0.3)

[insignif.]

(6)Con

trol

Group

Con

tacted

househ

olds

didno

twan

tto

participatein

theexpe

rimen

t

(6)CG/(−

0.2)

[insignif.]

McC

alleyan

dMidde

n(200

2)La

boratory

stud

y,in

which

thepa

rticipan

tsha

dto

setup

washing

machine

son

aco

mpu

ter:the

firstsixwashing

cycles

foran

evalua

tion

ofthe

„baseco

nsum

ption“

,which

thepa

rticipan

tsreceived

feed

back

abou

t.Th

efollo

wing

interven

tion

swererand

omly

assign

ed,

afterw

ards

anothe

r20

washing

cycles.

(1)CG/n

odifferen

ceto

(2)

(2)n.a./(−

9.6)

[insignif.]

(3)

(−12

.3**

*)/

(−21

.9)***

(4)(−

9.9)**

/(−

19.5)**

100

Electricity

NL

Labo

ratory

Expe

rimen

tLo

ng-run

effects:

n.a.

Costs:

n.a.

(1)Con

trol

Group

NoFe

edba

ck,n

ogo

al(2)Fe

edba

ckFe

edba

ck,n

ogo

al(3)Fe

edba

ck,G

oalSe

tting,

Com

mitmen

tFe

edba

ck,s

elf-im

posedgo

al(O

ptions:0

%,5

%,

10%,1

5%,20%

)(4)Fe

edba

ck,G

oalSe

tting

Feed

back,e

xterna

llyim

posedgo

al(20%

)Abrah

amse

etal.(20

07)

(1)Fe

edba

ck,G

oalSe

tting

Externally

setg

oalo

fene

rgyco

nserva

tion

(5%),

onlin

efeed

back

abou

tpe

rson

alco

nsum

ption

aftertw

oan

dfive

mon

ths

(1)+

(2)

sign

ificant

differen

ceto

control

grou

p***

/(−

8.3)

(3)CG/(+

0.4)

[insignif.]

189

Electricity,

Gas

(and

gasolin

e)

NL

RCT

Effects

referto

direct

energy

consum

ption,

notto

theas

well

measuredindirect

consum

ption,

becausetheindirect

consum

ption

strong

lyva

ried

andwas

notseen

asrobu

stby

theau

thors.

Forthean

alysis

ofthedirect

energy

consum

ptionbo

thtreatm

entgrou

psweretake

ntoge

ther,

astherewereno

sign

ificant

differen

ces

betw

eenthem

.

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Fe

edba

ck,G

oalSe

tting

[+Fram

ing(soc

ialno

rm)]

Like

(1)plus

amutua

lgoa

l(alltog

ethe

rsaving

5%)an

don

linefeed

back

abou

tthegrou

p's

consum

ptionaftertw

oan

dfive

mon

ths

(3)Con

trol

Group

Twicefilling

outaqu

estion

naire,

noothe

rinterven

tion

Harding

andHsiaw

(201

4)Offer

forcu

stom

ersto

participatein

prog

ram

tosave

electricity.

Ran

geof

goalsto

setfor

oneself

betw

een0%

,0–1

5%,1

5–50

%,o

ver50

%(1)(−

1.5)

[insignif.]

/n.a.

(2)(−

11.0)**/

n.a.

(3)(−

1.0)

12,451

Electricity

US

Match

ing

Hou

seho

ldswithop

timisticgo

als

(15–

50%)save

quitealotshortlyafter

startof

theprog

ram.T

hiseff

ectwan

esaftertw

oto

threemon

ths,

presum

ably

becausetheco

nsum

ersrealized

that

they

wou

ldno

tbe

able

toreachtheir

(1)Fe

edba

ck,G

oalSe

tting,

Com

mitmen

t[+

Com

parisonto

usag

ehistory]

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

199

[insignif.]

/n.a.

(4)0/

n.a.

(5)CG/n

.a.

very

optimisticgo

als.

Long

-run

effects

(after

18mon

ths)

for(2):Sign

ificant

effects

with

sign

ificanc

eleve

lof

95%;n

ospecific

statem

ents

abou

tsize

Costs:

n.a.

Self-set

goal:0

%,d

aily

access

toweb

site

with

mon

thly

bills

ofco

nsum

ption

(2)Fe

edba

ck,G

oalSe

tting,

Com

mitmen

t[+

Com

parisonto

usag

ehistory]

Self-set

goal:0

–15%

,daily

access

toweb

site

withmon

thly

bills

ofco

nsum

ption

(3)Fe

edba

ck,G

oalSe

tting,

Com

mitmen

t[+

Com

parisonto

usag

ehistory]

Self-set

goal:1

5–50

%,d

aily

access

toweb

site

withmon

thly

bills

ofco

nsum

ption

(4)Fe

edba

ck,G

oalSe

tting,

Com

mitmen

t[+

Com

parisonto

usag

ehistory]

Self-set

goal:>

50%,d

aily

access

toweb

site

withmon

thly

bills

ofco

nsum

ption

(5)Con

trol

Group

Noinform

ationab

outpa

rticipationin

expe

rimen

tJaeg

eran

dSchu

ltz

(201

7)(1)So

cial

Com

parison

[+En

ergy

saving

tips]

Duringadrou

ght,reside

ntsreceived

doorha

ngerswiththesocial

norm

:“Ove

r80

%of

househ

olds

inyo

urco

mmun

ityuseeffi

cien

tland

scap

eirriga

tion

tech

niqu

es”plus

water

conserva

tion

inform

ation

(1)n.a.

5,36

3Water

US

RCT

Long

-run

effects:

(fou

rmon

thspo

stinterven

tion

)(1)

n.a.

(2)

(−8.00

)***

(3)

n.a.

(4)

(−3.2)

[insignif.]

(5)

CG

(6)

CG

Costs:

n.a.

(2)So

cial

Com

parison,

Com

mitmen

t[+

Energy

saving

tips]

Like

(1)plus

commitmen

t“I,r

esiden

tof

____

committo

efficien

tlywateringmyland

scap

e.___(sign.)”

(2)for

compliers

(−3.53

)***

/n.a.

(3)Fram

ing(w

arning

)[+

Energy

saving

tips]

Doo

rhan

gerinclud

ingthewarning

“Duringthe

drou

ght,please

keep

inmindthat

using

excessivean

dun

necessaryam

ountsof

water

canresultin

pena

lties,

includ

ingfine

sof

upto

$500

,water

serviceinterrup

tion

,an

d/or

prosecution.” ,

addition

ally:water

conserva

tion

inform

ation.

(3)n.a.

(4)Com

mitmen

t[+

Energy

saving

tips,F

raming(w

arning

)]Like

(3)plus

theco

mmitmen

t‘I,

reside

ntof

____

committo

efficien

tlywateringmyland

scap

e.___(sign.)’

(4)for

compliers:

(−5.56

)***

/n.a.

(5)Con

trol

Group

[+En

ergy

saving

tips]

Doo

rhan

gerwithinform

ationan

dsometips

for

water

conserva

tion

beha

vior

(5)CG/n

.a.

(6)Con

trol

Group

Noinform

ationco

ntrol.

(6)CG/n

.a.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

200

Labe

ling

Verplan

kenan

dWeenig

(199

3)La

boratory

expe

rimen

t:Cho

icebe

tweenfour

refrigerators

120

Electricity

(ind

irect)

NL

Cho

ice-

expe

rimen

twith

rand

omassign

men

tto

expe

rimen

tal

grou

ps

Effect=

shareof

grou

pthat

prefersan

efficien

trefrigerator

toan

ineffi

cien

ton

ein

%

Long

-run

effects:

n.a.

Costs:

n.a.

(1)Con

trol

Group

Inform

ationon

consum

ptionin

kWh,

notime

pressure

(1)27

(2)La

belin

g[+

grap

hicalen

hanc

emen

t]Ann

ualen

ergy

consum

ptionas

grap

hicallabe

l(m

easuredin

localcu

rren

cy),no

timepressure

(2)50

*

(3)Con

trol

grou

pInform

ationon

consum

ptionin

kWh,

time

pressure

(5min

forde

cision

)

(3)33

(4)La

belin

g[+

grap

hicalen

hanc

emen

t]Ann

ualen

ergy

consum

ptionas

grap

hicallabe

l,timepressure

(4)23

Kurzet

al.(20

05)

(1)La

belin

gA

series

oflabe

lswas

placed

ondifferen

tap

plianc

esin

theho

useh

old:

refrigerators,

air

cond

itione

rs,s

howers,

washing

machine

s,clothe

sdrye

rs,d

ishw

ashe

rs,toilets

andou

tdoo

rtaps.T

helabe

lsprov

ided

inform

ationab

outthe

water

anden

ergy

-con

sumptionleve

lsof

the

labe

ledap

plianc

es

(1)n.a./from

insign

if.to

(−23

)**

166

Electricity

andWater

AU

RCT

Thestud

yrepo

rtstheeff

ects

forseve

nweeks

sepa

rately,witho

utdo

cumen

ting

anav

erag

etreatm

ent

effect.Th

erefore,

werepo

rtthe

interval

ofthedo

cumen

tedeff

ects.

Long

-run

effects:

n.a.

Costs:

n.a.

(2)Con

trol

Group

Hou

seho

ldswereprov

ided

withthesame

inform

ationas

intreatm

ent(1)bu

tin

theform

ofinform

ationleafl

etsinsteadof

labe

ls

(2)n.a./

insign

if.

(3)Fe

edba

ck,S

ocialCom

parison

[+Graph

ical

enha

ncem

ent]

Hou

seho

ldsreceived

e-mails

withgrap

hical

feed

back

ontheirleve

lsof

water

anden

ergy

consum

ptionan

daco

mpa

risonto

othe

rho

useh

olds

ofsimila

rsize

who

participated

inthestud

y

(3)n.a./

insign

if.

Brou

nenan

dKok

(201

1)La

belin

gAna

lysisof

thee ff

ectof

theEU

labe

lfor

energy

efficien

cyon

thesalespriceof

houses

witha

„green

“lab

el(A

,Bor

C)

(+3.7)**

*31

,993

Electricity

Water

Gas

(ind

irect)

NL

Observa

tion

stud

y,co

ntrolle

dforself-selection

withtw

o-step

Heckm

an-

proc

edure

Effect=

averag

emark-up

in%

onho

uses

with"green

"labe

lrelative

toco

mpa

rableho

uses

Long

-run

effects:

n.a.

Costs:

n.a.

Heinz

le(201

2)Ex

perimen

t1:

Participan

tsareaske

dto

evalua

tetheen

ergy

saving

potentialof

differen

tTV

s.Th

eprov

ided

inform

ationva

ries:

Electricity

(ind

irect)

DE

Cho

ice-

Expe

rimen

tswithrand

omassign

men

tto

expe

rimen

tal

(1)18

.6/6

6.3/

15.1

(1)–(3)25

2Eff

ects

(1)–(3)=

Prop

ortion

ofpa

rticipan

tswho

correctlyestimatethe

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

201

grou

pswithina

survey

(1)La

belin

gCon

sumptioninform

ationin

Wattan

dcu

rren

tpriceforelectricity

energy

saving

potentialin

%/w

hoov

erestimate/who

unde

restim

ate

(2)La

belin

gCon

sumptioninform

ationin

Watt

(2)19

.8/6

4.0/

16.3

(3)La

belin

gActua

lan

nual

operatingco

stsof

thede

vices

(3)92

.5/3

.8/

3.8**

*(significant

differen

ceto

theothe

rtreatm

ent

grou

ps)

Expe

rimen

t2:

Cho

icebe

tweentw

oTV

swithdiffering

inform

ation:

(4)Con

trol

Group

Con

sumptioninform

ationin

Watt(60W

vs.

225W)

(4)48

1.22

(4)–(6)20

8Eff

ects

(4)–(6)=

Med

ianwillingn

ess

topa

yforamoreeffi

cien

tTV

inEu

ro(actua

lsaving

potential:48

0Eu

ro).

Long

-run

effects:

n.a.

Costs:

n.a.

(5)La

belin

gAbsoluteop

eratingco

stsfortheco

urse

often

years

(180

€vs.6

60€)

(5)64

1.96

**

(6)La

belin

gAnn

ualo

perating

costs(18€

/yearvs.6

6€/y

ear)

(6)35

3.97

**

Hou

de(201

4)Natural

Expe

rimen

t1:

Electricity

(ind

irek

t)US

Twona

tural

expe

rimen

tsEff

ects

(1)+

(2)=

Ave

rage

willingn

essto

payfortheEn

ergy

Star

Labe

lin

$/shareof

totalpricefor

refrigerator

in%

Long

-run

effects:

n.a.

Costs:

Theop

portun

ityco

stsof

impe

rfect

inform

ationam

ount

toab

out$1

5pe

rrefrigerator

sold.E

xtrapo

latedon

the

who

lemarke

tforrefrigeratorsthat

mak

esasum

of$1

35mil/

year.

According

totheau

thor

that

istw

iceas

muc

has

thean

nual

costsforthe

Energy

Star

Labe

lprog

ram.

(1)La

belin

gTh

etigh

tening

ofthecriteria

forthereceiptof

theEn

ergy

Star

Labe

lin20

08in

theUSallowed

toob

servethewillingn

essto

payforthesame

mod

elsof

refrigeratorsin

thesamestorewith

andwitho

utthelabe

l(beforean

dafterthe

chan

geof

regu

lation

)

(1)(+

19)**/

1.5

Natural

Expe

rimen

t2:

(2)La

belin

gIn

Janu

ary20

10itwas

publishe

dthat

some

refrigeratorsreceived

theEn

ergy

Star

Labe

l,althou

ghthey

didno

tactua

llyfulfillthecriteria

(inc

orrect

applianc

eof

thetest

proc

edure).

Becausethefollo

wingwithd

rawal

ofthelabe

lco

uldno

tbean

ticipa

tedby

either

theprod

ucers

orthecu

stom

ers,

thelabe

l'sva

luecanbe

determ

ined

bythewillingn

essto

payforthese

mod

elsof

refrigeratorsbe

fore

andafterJanu

ary

2010

.

(2)(+

89.6)

[insignif.]

/7.07

(2)18

4,64

5

New

ellan

dSiikam

äki

(201

4)Su

rvey

(online)

abou

tsixdifferen

thy

pothetical

purcha

sede

cision

sforwarm

water

proc

essors.

Therearethreebo

ilers

toch

oose

from

.The

boile

rsdiffer

inpricean

deach

boile

rislabe

led

withadifferen

tve

rsionof

theEn

ergy

Star

1,18

4Electricity

andGas

(ind

irect)

US

Cho

ice-

expe

rimen

twith

rand

omassign

men

tto

expe

rimen

tal

grou

ps

Effects=

relative

willingn

essto

pay

foren

ergy

-efficien

tdev

ices:a

WTP

of1

stan

dsforco

stminim

izingbe

havior,

whe

reach

ange

insalespricean

da

chan

gein

theop

eratingco

stsare

equa

llyweigh

ted.

Ava

luelower

than

1

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

202

Labe

l,which

differ

inthekind

ofinform

ation

that

isprov

ided

:show

san

unde

restim

ationof

saving

sthroug

hen

ergy

efficien

cy,w

hile

ava

lue>

1show

san

overestimation.

Thesign

ificanc

eis

measuredas

the

differen

ceto

1.

Long

-run

effects:

n.a.

Costs:

n.a.

(1)La

belin

gOnlythesimpleco

nsum

ptioninform

ationis

prov

ided

(1)0.7*

*

(2)La

belin

gLike

(1)+

relative

consum

ptionco

stsin

compa

risonto

differen

tde

vices

(2)0.68

**

(3)La

belin

gLike

(2)+

Inform

ationab

outtheCO2

emission

s

(3)0.93

[insignif.]

(4)La

belin

gLike

(2)+

specificinform

ationon

energy

consum

ption

(4)1.02

[insignif.]

(5)La

belin

g[+

grap

hicalen

hanc

emen

t]Wie

(4)+

awarde

dwithan

"Ene

rgy-Star-Lab

el“

(5)1.36

**

(6)La

belin

g[+

grap

hicalen

hanc

emen

t]Like

(1)+

Inform

ationab

outen

ergy

consum

ptionin

relation

toothe

rde

vicesin

overseeablenu

ances(A

-Grank

ing:

„EU-Style“)

(6)1.34

**

Öland

eran

dTh

øgersen

(201

4)Cho

icebe

tweenfour

TVde

viceswithdifferen

tlabe

ls;q

uestionwhich

device

thecu

stom

erwou

ldbu

y

151

Electricity

(ind

irect)

DK

Cho

ice-

expe

rimen

twith

rand

omassign

men

tto

expe

rimen

tal

grou

ps

Long

-run

effects:

n.a.

Costs:

n.a.

(1)La

belin

gLa

belswithen

ergy

efficien

cygrad

esfrom

A-G

(1)+

(2)Th

escalefrom

A–

Gmorethan

doub

ledthe

prob

ability

toch

oose

anen

ergy

-effi

cien

tde

vice

inco

mpa

rison

tothescale

from

A+

++

−D

(2)La

belin

gLa

bels

withen

ergy

e fficien

cygrad

esfrom

A+

++

-D

Allc

ottan

dTa

ubinsky

(201

5)Ex

perimen

t1:

Cho

icebe

tweenan

energy

-sav

ingbu

lban

dthreeco

nven

tion

albu

lbswithco

mpa

rable

power.

(1)+

(2)

1,53

3Electricity

(ind

irect)

US

Expe

rimen

t1:

„Artefactual

field

expe

rimen

t“:

Cho

ice-

Expe

rimen

twith

rand

omassign

men

tto

expe

rimen

tal

grou

ps.O

neof

30ch

oices

(ran

domly

selected

)lead

sto

Forexpe

rimen

t1seve

raltreatm

ent

grou

psweresummarized

forthe

analysis.

Effects

(1)=

Willingn

essto

payin

Dollarforan

energy

-sav

ingbu

lbin

compa

risonto

aco

nven

tion

albu

lban

din

compa

risonto

theco

ntrolg

roup

(2)

(1)La

belin

gInform

ationab

outco

nsum

ption,

cost

saving

withen

ergy

-sav

ingbu

lbs,do

wnsides

ofen

ergy

-saving

bulbs(+

Inform

ationlik

ein

(2))

(1)

(+2.30

)***

/n.a.

(2)Con

trol

Group

Inform

ationab

outam

ount

ofen

ergy

-sav

ing

bulbssold

andsalesde

velopm

entin

thepa

st.

(2)CG

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

203

actual

buying

decision

.Ex

perimen

t2:

Customersin

abigelectrical

storewereaske

dab

outtheirco

nsum

ptionbe

havior

andwith

theirindividu

alinform

ationaco

mpa

risonof

theen

ergy

consum

ptionof

energy

-sav

ingbu

lbs

andco

nven

tion

albu

lbswas

prov

ided

viaiPad

(3)La

belin

g[+

Fina

ncialincentive]

Ann

ualan

dtotalen

ergy

costs+

discou

ntco

upon

(10%

onallbu

lbs)

(3)(−

2.2)

[insignif.]

(3)–(6)

1,08

7Ex

perimen

t2:

RCT

Effects

(3)–(6)=

Prob

ability

ofthe

purcha

seof

anen

ergy

-sav

ingbu

lbin

compa

risonto

conv

ention

albu

lbsafter

theinform

ationtreatm

entan

din

compa

risonto

theco

ntrolgrou

pin

percen

tage

points.

Long

-run

effects:

n.a.

Costs:

n.a.

(4)La

belin

g[+

Fina

ncialincentive]

Ann

ualan

dtotalen

ergy

costs+

discou

ntco

upon

(10%

onallbu

lbs,

30%

onen

ergy

-saving

bulbs)

(4)(+

11.0)

[insignif.]

(5)Con

trol

Group

[+Fina

ncialincentive]

Nointerview/information,

discou

ntco

upon

(10%

onallb

ulbs,3

0%on

energy

-sav

ingbu

lbs)

(5)+

(7.8)*

(6)Con

trol

Group

[+Fina

ncialincentive]

Nointerview/information,

discou

ntco

upon

(10%

onallbu

lbs)

(6)CG;

Prob

ability

for

purcha

seof

anen

ergy

-sav

ing

bulb:3

8%Waech

teret

al.(20

15)

Expe

rimen

t1:

Ontheba

sisof

ascalefrom

0(not

efficien

t)to

100(veryeffi

cien

t)thepa

rticipan

tsha

dto

estimatetheen

ergy

consum

ptionof

aTV

.Fou

rdifferen

tve

rsions

ofalabe

l(ran

domly

assign

ed)prov

ided

inform

ationon

thegrad

eof

energy

efficien

cy(A

–high

,B–low)an

dthe

electricityco

nsum

ption(high,

low).

Expe

rimen

t1:

Expe

rimen

t1:

166

Electricity

(ind

irect)

Online-

expe

rimen

t,which

was

cond

ucted

in Switzerlan

d

Expe

rimen

t1:

Cho

ice-

Expe

rimen

twith

rand

omassign

men

tto

expe

rimen

tal

grou

ps

Dep

ende

ntva

riab

les:

Expe

rimen

t1:

Estimationof

theen

ergy

e fficien

cyof

theTV

devices

(1)La

belin

gInform

ationab

outthegrad

eof

energy

efficien

cyof

theTV

devices

(1)Th

ehigh

erthegrad

eof

energy

efficien

cy,the

lower

the

estimates

for

theelectricity

consum

ption

(significant**

*)(2)Con

trol

Group

Inform

ationab

outtheelectricityco

nsum

ption

oftheTV

devices

(2)Noeff

ecto

nestimates

ofelectricity

consum

ption

Expe

rimen

t2:

Ontheba

sisof

twode

viceswithdifferen

tlabe

ls,p

articipa

ntsha

dto

decide

which

device

they

wou

ldreco

mmen

dto

anen

ergy

Expe

rimen

t2:

Expe

rimen

t2:

305

Expe

rimen

t2:

nogrou

psEx

perimen

t2:

Shareof

participan

tswho

wrong

lyreco

mmen

dedthede

vice

withthe

high

eractual

energy

consum

ptionin

%

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

204

conserving

person

.The

device

withthehigh

ergrad

eof

energy

efficien

cythereb

yfeatured

the

high

eractual

energy

consum

ption.

(3)La

belin

gDev

ice=

TV(1)44

.6

(4)La

belin

gDev

ice=

Freezer

(2)72

.8

Expe

rimen

t3:

Ontheba

sisof

twodifferen

tlabe

lspa

rticipan

tsha

dto

evalua

tetheen

ergy

consum

ptionof

afreezerin

compa

risonto

areferenc

erefrigerator,w

hereas

thelabe

lsdiffered

only

inthegrad

eof

energy

efficien

cy,b

utno

tinactual

energy

consum

ption.

Expe

rimen

t3:

Expe

rimen

t3:

166

Expe

rimen

t3:

Cho

ice-

Expe

rimen

twith

rand

omassign

men

tto

expe

rimen

tal

grou

ps

Expe

rimen

t3:

Estimationof

theen

ergy

consum

ption

ofthefreezerco

mpa

redto

the

referenc

erefrigerator

onascalefrom

1to

100

Long

-run

effects:

n.a.

Costs:

n.a.

(5)La

belin

gLa

bel:“G

rade

ofen

ergy

efficien

cy:A

++

+”

(1)67

.72*

*

(6)Con

trol

Group

“Grade

ofen

ergy

efficien

cy:A

+”

(2)77

.07(C

G)

Jensen

etal.(20

16)

Natural

expe

rimen

tthat

exam

ines

theeff

ectof

theEU

-req

uiremen

tto

includ

ean

energy

performan

cerating

from

Ato

Gin

thereal

estate

adve

rtisem

ent,which

wen

tinto

effectin

2010

.The

docu

men

tedeff

ectsizesreferto

the

chan

gein

prop

erty

salespricebe

fore

andafter

2010

whe

nco

mpa

redto

acertainreferenc

egrou

p.Mod

el1:

catego

rizesgo

odan

dba

drating

sin

twoexclusivegrou

ps(1)La

belin

ggo

oden

ergy

performan

cerating

ofbu

ildings

(Ato

C)

(1)from

(+2.4)**

*to

(+10

.1)***

/n.a.

117,48

3DK

FELo

ng-run

effects:

n.a.

Costs:

n.a.

(2)Con

trol

Group

energy

performan

cerating

ofbu

ildings

Dto

G(2)CG

Mod

el2:

compa

reseach

energy

rating

tothe

referenc

erating

G(3)La

belin

gEn

ergy

performan

cerating

Aor

B(4)La

belin

gEn

ergy

performan

cerating

C(5)Con

trol

Energy

performan

cerating

D(6)La

belin

gEn

ergy

performan

cerating

F(7)La

belin

gEn

ergy

performan

cerating

G

(1)from

(+6.6)**

*to

(+6.2)**

*/n.a.

(2)from

(+0.2)

to(+

5.1)**

*/n.a.

(3)CG

(4)from

(−1.5)**

*to

(−5.4)**

*/n.a.

(5)from

(−9.3)**

*to

(−24

.3)***

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

205

App

endixB

Theke

ywordsearch

inEc

onLitc

omprised

thecatego

ries

“Title”,“A

bstract”,a

nd“Sub

jects(SU)”

forbo

thjourna

larticlesan

dworking

pape

rs.InScienc

eDirect,wesearch

edwithke

ywords

inthe

catego

ry“A

bstract,Title,

Key

words”an

dinclud

edarticles

from

thedisciplin

es‘Eco

nomics,

Econ

ometrics,Fina

nce’,‘Psycho

logy

’,‘Soc

ialScienc

es’,‘Env

iron

men

t’,an

d“E

nergy”.

TableB1

“Listof

keyw

ords”.

Interven

tion

Key

words

Social compa

rison

“Soc

ialno

rms”

OR“soc

iallearning

”OR“soc

ialmod

eling”

OR“soc

ialinflue

nce”

OR“p

eerco

mpa

rison”

OR“p

eerinform

ation”

OR“com

parative

energy

inform

ation”

OR

“feedb

ack”

AND

“ene

rgyco

nserva

tion

”OR“ene

rgyco

nsum

ption”

OR“ene

rgyuse”

OR“ene

rgyusag

e”OR“ene

rgyde

man

d*”OR“ene

rgysaving

”OR“electricity

conserva

tion

”OR“electricity

consum

ption”

OR“electricity

use”

OR“electricity

usag

e”OR“electricity

deman

d*”OR“electricity

saving

”OR“gas

conserva

tion

”OR“gas

consum

ption”

OR“gas

use”

OR“gas

usag

e”OR“gas

deman

d*”OR“gas

saving

”OR“w

ater

conserva

tion

”OR“w

ater

consum

ption”

OR“w

ater

use”

OR“w

ater

usag

e”OR“w

ater

deman

d*”OR“w

ater

saving

”OR

“con

servationbe

havior”

Com

mitmen

t“P

re-com

mitmen

t”OR“p

reco

mmitmen

t”OR“p

ledg

e”OR“b

ehav

ioralco

ntract”OR“com

mitmen

tco

ntract”OR“com

mitmen

tde

vices”

OR“com

mitmen

tap

proa

ch*”

OR

“persona

lco

mmitmen

t”OR“p

ublic

commitmen

t”OR“self-co

ntrol”

OR“self-regu

lation

”AND

‘ene

rgyco

nserva

tion

’OR‘ene

rgyco

nsum

ption’

OR‘ene

rgyuse’

OR‘ene

rgyusag

e’OR‘ene

rgyde

man

d*’O

R‘ene

rgysaving

’OR‘electricity

conserva

tion

’OR‘electricity

consum

ption’

OR‘electricity

use’

OR‘electricity

usag

e’OR‘electricity

deman

d*’O

R‘electricity

saving

’OR‘gas

conserva

tion

’OR‘gas

consum

ption’

OR‘gas

use’

OR‘gas

usag

e’OR‘gas

deman

d*’O

R‘gas

saving

’OR‘w

ater

conserva

tion

’ OR‘w

ater

consum

ption’

OR‘w

ater

use’

OR‘w

ater

usag

e’OR‘w

ater

deman

d*’O

R‘w

ater

saving

’OR‘con

servation

beha

vior’

Goa

l-setting

‘Goa

lsetting’

AND

‘ene

rgyco

nserva

tion

’OR‘ene

rgyco

nsum

ption’

OR‘ene

rgyuse’

OR‘ene

rgyusag

e’OR‘ene

rgyde

man

d*’O

R‘ene

rgysaving

’OR‘electricity

conserva

tion

’OR‘electricity

consum

ption’

OR‘electricity

use’

OR‘electricity

usag

e’OR‘electricity

deman

d*’O

R‘electricity

saving

’OR‘gas

conserva

tion

’OR‘gas

consum

ption’

OR‘gas

use’

OR‘gas

usag

e’OR‘gas

deman

d*’O

R‘gas

saving

’OR‘w

ater

conserva

tion

’OR‘w

ater

consum

ption’

OR‘w

ater

use’

OR‘w

ater

usag

e’OR‘w

ater

deman

d*’O

R‘w

ater

saving

’OR‘con

servation

beha

vior’

Labe

ling

“Ene

rgylabe

ling”

OR“ene

rgylabe

lling

”OR“informationlabe

l*”OR“ene

rgyinform

ation”

OR“ene

rgylabe

l”OR“informationacqu

isition”

OR“informationdisclosure”OR

“env

iron

men

talcertification

”AND

‘Ene

rgyco

nserva

tion

’OR‘ene

rgyco

nsum

ption’

OR‘ene

rgyuse’

OR‘ene

rgyusag

e’OR‘ene

rgyde

man

d*’O

R‘ene

rgysaving

’OR‘electricity

conserva

tion

’OR‘electricity

consum

ption’

OR‘electricity

use’

OR‘electricity

usag

e’OR‘electricity

deman

d*’O

R‘electricity

saving

’OR‘gas

conserva

tion

’OR‘gas

consum

ption’

OR‘gas

use’

OR‘gas

usag

e’OR‘gas

deman

d*’O

R‘gas

saving

’OR‘w

ater

conserva

tion

’OR‘w

ater

consum

ption’

OR‘w

ater

use’

OR‘w

ater

usag

e’OR‘w

ater

deman

d*’O

R‘w

ater

saving

’OR‘con

servation

beha

vior’

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

206

Ana

lysisrefers

tothis

stan

dardized

inclusionde

cision

form

inCitav

i.Ifon

eof

thecriteria

was

code

dwith“n

o”,the

stud

ywas

exclud

edfrom

thesystem

atic

review

.

TableB2

“Inc

lusion

decision

form

”.

Autho

r(year)

Text

Title

Text

Nam

eof

code

rTe

xtStud

yinclud

esap

pliedresearch

(not

just

theo

reticalmod

els)

Options:y

es/n

o/discuss

Stud

yinclud

esat

leaston

eof

thech

osen

interven

tion

sOptions:F

eedb

ack/

Social

Com

parison/

Com

mitmen

t/Goa

l-Setting

/Lab

eling/

discuss

Stud

yrefers

toat

leaston

eof

theou

tcom

eva

riab

les

Options:g

as/w

ater/electricity/d

iscu

ssStud

ytargetsprivateho

useh

olds

orindividu

alsin

privateho

useh

olds

Options:y

es/n

o/discuss

Stud

ywas

carriedou

tin

ade

velope

dco

untry

Options:y

es/n

o/discuss

Inclusionde

cision

Options:inc

lude

/exclude

/relev

antmeta-stud

y/discuss

TableB3

“Cod

ingsheet“.

Stud

yAutho

rs(year)

Text

Type

ofinterven

tion

e.g.

1 1+

21+

2+

40

Ope

nco

de0=

Con

trol

grou

p1=

Feed

back

2=

Social

compa

rison

3=

Com

mitmen

t4=

Goa

l-setting

5=

Labe

ling

1+

2=

Feed

b.+

Social

Com

parison

3+

4=

Com

mitmen

t+Goa

l-Se

tting

Interven

tion

design

e.g.

Onetim

elette

r,external

goal

(5%)

Onetim

elette

r,self-setgoal

Onetim

elette

r,external

goal,c

ompa

risonwith

anothertreatm

entgrou

p

Text

Com

bina

tion

withad

dition

alinterven

tion

se.g.

Non

eEn

ergy

saving

tips

Non

e

Text

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

207

Effect(in%)in

compa

risonto:c

ontrol

grou

p/ba

selin

eSign

ificanc

eleve

l:**

*:p<

.01

**:p

<.05

*:p<

.10

(insignif.):n

otsign

ificant

e.g.

(−2.2)**

/(−

5.4)**

(−2.7)*/(−

10.0)*

(−0.5)

[insignif.]

/(−1.9)*

KG/(−1.1)

[insignif.]

Num

ber

Ne.g.

34,000

Num

ber

Dep

ende

ntva

riab

lee.g.

2Ope

nCod

e1=

Electricity

2=

Water

3=

Gas

ForLa

belin

g:1A

=Electricity(ind

irectly)

2A=

Water

(ind

irectly)

3A=

Gas

(ind

irectly)

Cou

ntry

e.g.

US

Cou

ntry

code

Metho

dof

causal

analysis

e.g.

1Ope

nco

de1=

RCT

2=

Match

edCom

parison

3=

RDD

4=

Diff-in-Diff

5=

FE6=

Other

Rem

arks

e.g.

(1)

Limita

tions

(2)

Cost–benefitan

alysis

(3)

Long-run

effect+

follo

w-upperiod

afterinterventio

n

Text

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

208

References

Abrahamse, W., Steg, L., Vlek, C., Rothengatter, T., 2005. A review of intervention studiesaimed at household energy conservation. J. Environ. Psychol. 25 (3), 273–291.

Abrahamse, W., Steg, L., Vlek, C., Rothengatter, T., 2007. The effect of tailored in-formation, goal setting, and tailored feedback on household energy use, energy-re-lated behaviors, and behavioral antecedents. J. Environ. Psychol. 27 (4), 265–276.http://dx.doi.org/10.1016/j.jenvp.2007.08.002.

Alberts, G., Gurguc, Z., Koutroumpis, P., Martin, R., Muuls, M., Napp, T., 2016.Competition and norms: a self-defeating combination? Energ Policy 96, 504–523.

Allcott, H., 2011. Social norms and energy conservation. J. Public Econ. 95 (9–10),1082–1095. http://dx.doi.org/10.1016/j.jpubeco.2011.03.003.

Allcott, H., 2015. Site selection bias in program evaluation. Q. J. Econ. 130 (3),1117–1165.

Allcott, H., 2016. Paternalism and energy efficiency: an overview. Annu. Rev. Econ. 8,145–176.

Allcott, H., Kessler, J.B., 2015. The Welfare Effects of Nudges: A Case Study of Energy UseSocial Comparisons (NBER Working Paper No. 21671). National Bureau of EconomicResearch.

Allcott, H., Mullainathan, S., 2010. Behavior and energy policy. Proc. Am. Assoc. Adv. Sci.327 (5970), 1204–1205. http://dx.doi.org/10.1126/science.1180775.

Allcott, H., Rogers, T., 2014. The short-run and long-run effects of behavioral interven-tions: experimental evidence from energy conservation. Am. Econ. Rev. 104 (10),3003–3037. http://dx.doi.org/10.1257/aer.104.10.3003.

Allcott, H., Taubinsky, D., 2015. Evaluating behaviorally motivated policy: experimentalevidence from the lightbulb market. Am. Econ. Rev. 105 (8), 2501–2538. http://dx.doi.org/10.1257/aer.20131564.

Anderson, K., Song, K., Lee, S.H., Krupka, E., Lee, H., Park, M., 2017. Longitudinal ana-lysis of normative energy use feedback on dormitory occupants. Appl. Energy 189,623–639. http://dx.doi.org/10.1016/j.apenergy.2016.12.086.

Andor, M., Fels, K., 2017. Energiesparen durch verhaltensökonomisch motivierteMaßnahmen?. acatech Schriftenreihe Energiesysteme der Zukunft.

Andor, M., Gerster, A., Peters, J., Schmidt, C.M., 2017a. Social norms and energy con-servation beyond the US, (Ruhr Economic Papers #714).

Andor, M., Gerster, A., Sommer, S., 2017b. Consumer Inattention, Heuristic Thinking andthe Role of Energy Labels, (Ruhr Economic Papers #671).

Angrist, J.D., Pischke, J.-S., 2009. Mostly Harmless Econometrics: An empiricists'Companion. Princeton University Press.

Ayres, I., Raseman, S., Shih, A., 2013. Evidence from two large field experiments that peercomparison feedback can reduce residential energy usage. J. Law Econ. Org. 29 (5),992–1022. http://dx.doi.org/10.1093/jleo/ews020.

Bandura, A., 1986. Social Foundations of Thought and Action: A Social Cognitive Theory.Prentice Hall.

Becker, L.J., 1978. Joint effect of feedback and goal setting on performance: a field studyof residential energy conservation. J. Appl. Psychol. 63 (4), 428–433. http://dx.doi.org/10.1037/0021-9010.63.4.428.

Benartzi, S., Beshears, J., Milkman, K.L., Sunstein, C.R., Thaler, R.H., Shankar, M.,Tucker-Ray, W., Congdon, W.J., Galing, S., 2017. Should governments invest more innudging? Psychol. Sci. 28 (8), 1041–1055. http://dx.doi.org/10.1177/0956797617702501.

Brandon, A., Ferraro, P.J., List, J.A., Metcalfe, R.D., Price, M.K., Rundhammer, F., 2017.Do The Effects of Social Nudges Persist? Theory and Evidence from 38 Natural FieldExperiments. (NBER Working Paper No. 23277). National Bureau of EconomicResearch.

Brounen, D., Kok, N., 2011. On the economics of energy labels in the housing market. J.Environ. Econ. Manag. 62 (2), 166–179. http://dx.doi.org/10.1016/j.jeem.2010.11.006.

Campbell Collaboration, 2014. Campbell Systematic Reviews: Policies and Guidelines,Campbell Systematic Reviews 2014: Supplement 1.

Chetty, R., 2015. Behavioral economics and public policy: a pragmatic perspective. Am.Econ. Rev. 105 (5), 1–33.

Costa, D.L., Kahn, M.E., 2013. Energy conservation 'nudges' and environmentalistideology: evidence from a randomized residential electricity field experiment. J. Eur.Econ. Assoc. 11 (3), 680–702. http://dx.doi.org/10.1111/jeea.12011.

Damgaard, M.T., Gravert, C., 2018. The hidden costs of nudging: Experimental evidencefrom reminders in fundraising. J. Public Econ. 157, 15–26.

Delmas, M.A., Lessem, N., 2014. Saving power to conserve your reputation? The effec-tiveness of private versus public information. J. Environ. Econ. Manag. 67 (3),353–370. http://dx.doi.org/10.1016/j.jeem.2013.12.009.

Delmas, M.A., Fischlein, M., Asensio, O.I., 2013. Information strategies and energy con-servation behavior: a meta-analysis of experimental studies from 1975 to 2012. EnergPolicy 61, 729–739. http://dx.doi.org/10.1016/j.enpol.2013.05.109.

Dolan, P., Metcalfe, R., 2013. Neighbors, Knowledge, and Nuggets: Two Natural FieldExperiments on the Role of Incentives on Energy Conservation (Economics WorkingPaper No. 2589269). Institute for Research, Becker Friedman. http://dx.doi.org/10.2139/ssrn.2589269.

Ferraro, P.J., Price, M.K., 2013. Using nonpecuniary strategies to influence behavior:evidence from a large-scale field experiment. Rev. Econ. Stat. 95 (1), 64–73. http://dx.doi.org/10.1162/REST_a_00344.

Ferraro, P.J., Miranda, J.J., Price, M.K., 2011. The persistence of treatment effects withnorm-based policy instruments: evidence from a randomized environmental policyexperiment. Am. Econ. Rev. 101 (3), 318–322. http://dx.doi.org/10.1257/aer.101.3.318.

Frondel, M., Vance, C., 2013. Energy efficiency: don't belittle the rebound effect. Nature494(430-430) http://dx.doi.org/10.1038/494430c.

Haakana, M., Sillanpaeae, L., Talsi, M., 1997. The effect of feedback and focused adviceon household energy consumption. Paper Presented at the Summer Study of theEuropean Council for an Energy Efficient Economy.

Hahn, R., Metcalfe, R.D., Novgorodsky, D., Price, M.K., 2016. The Behavioralist as PolicyDesigner: The Need to Test Multiple Treatments to Meet Multiple Targets(Experimental Economics Center Working Paper Series No. 2016-05.). ExperimentalEconomics Center, Andrew Young School of Policy Studies, Georgia State University.

Harding, M., Hsiaw, A., 2014. Goal setting and energy conservation. J. Econ. Behav.Organ. 107, 209–227. http://dx.doi.org/10.1016/j.jebo.2014.04.012.

Heinzle, S.L., 2012. Disclosure of energy operating cost information: a silver bullet forovercoming the energy-efficiency gap? J. Consum. Policy 35 (1), 43–64. http://dx.doi.org/10.1007/s10603-012-9189-6.

Houde, S., 2014. How Consumers Respond to Environmental Certification and the Valueof Energy Information (NBER Working Paper No. 20019). National Bureau ofEconomic Researchhttp://dx.doi.org/10.3386/w20019.

van Houwelingen, J.H., van Raaij, W.F., 1989. The effect of goal-setting and daily elec-tronic feedback on in-home energy use. J. Consum. Res. 16 (1), 98–105. http://dx.doi.org/10.1086/209197.

Imbens, G.W., Wooldridge, J.M., 2009. Recent developments in the econometrics ofprogram evaluation. J. Econ. Lit. 47 (1), 5–86. http://dx.doi.org/10.1257/jel.47.1.5.

Ito, K., 2015. Asymmetric incentives in subsidies: evidence from a large-scale electricityrebate program. Am. Econ. J. Econ. Pol. 7 (3), 209–237.

Jaeger, C.M., Schultz, P.W., 2017. Coupling social norms and commitments: testing theunderdetected nature of social influence. J. Environ. Psychol. 51, 199–208. http://dx.doi.org/10.1016/j.jenvp.2017.03.015.

Jensen, O.M., Hansen, A.R., Kragh, J., 2016. Market response to the public display ofenergy performance rating at property sales. Energ Policy 93, 229–235.

Kahneman, D., 2003. Maps of bounded rationality: psychology for behavioural eco-nomics. Am. Econ. Rev. 93 (5), 1449–1475.

Kahneman, D., Tversky, A., 2013. Prospect theory: an analysis of decision under risk. In:Handbook of the Fundamentals of Financial Decision Making: Part I, pp. 99–127.

Kantola, S.J., Syme, G.J., Campbell, N.A., 1984. Cognitive dissonance and energy con-servation. J. Appl. Psychol. 69 (3), 416–421. http://dx.doi.org/10.1037/0021-9010.69.3.416.

Karlin, B., Ford, R., Zinger, J., 2015. The effects of feedback on energy conservation: ameta-analysis. Psychol. Bull. 141, 1205–1247.

Katzev, R.D., Johnson, T.R., 1983. A social-psychological analysis of residential electricityconsumption: the impact of minimal justification techniques. J. Econ. Psychol. 3(3–4), 268–284. http://dx.doi.org/10.1016/0167-4870(83)90006-5.

Katzev, R.D., Johnson, T.R., 1984. Comparing the effects of monetary incentives and foot-in-the-door strategies in promoting residential electricity conservation. J. Appl. Soc.Psychol. 14 (1), 12–27. http://dx.doi.org/10.1111/j.1559-1816.1984.tb02217.x.

Komatsu, H., Nishio, K., 2015. An experimental study on motivational change for elec-tricity conservation by normative messages. Appl. Energy 158, 35–43. http://dx.doi.org/10.1016/j.apenergy.2015.08.029.

Kőszegi, B., Szeidl, A., 2012. A model of focusing in economic choice. Q. J. Econ. 128 (1),53–104.

Kurz, T., Donaghue, N., Walker, I., 2005. Utilizing a social-ecological framework topromote water and energy conservation: a field experiment. J. Appl. Soc. Psychol. 35(6), 1281–1300. http://dx.doi.org/10.1111/j.1559-1816.2005.tb02171.x.

Laibson, D., 1997. Golden eggs and hyperbolic discounting. Q. J. Econ. 112 (2), 443–478.Lokhorst, A.M., Werner, C., Staats, H., van Dijk, E., Gale, J.E., 2013. Commitment and

behavior change: a meta-analysis and critical review of commitment-making strate-gies in environmental research. Environ. Behav. 45 (1), 3–34.

McCalley, L.T., Midden, C.J.H., 2002. Energy conservation through product-integratedfeedback: the roles of goal-setting and social orientation. J. Econ. Psychol. 23 (5),589–603. http://dx.doi.org/10.1016/S0167-4870(02)00119-8.

Mizobuchi, K., Takeuchi, K., 2013. The influences of financial and non-financial factorson energy-saving behaviour: a field experiment in Japan. Energ Policy 63, 775–787.http://dx.doi.org/10.1016/j.enpol.2013.08.064.

Newell, R.G., Siikamäki, J.V., 2014. Nudging energy efficiency behavior: the role of in-formation labels. J. Assoc. Environ. Resour. Econ. 1 (4), 555–598. http://dx.doi.org/10.3386/w19224.

Nisan, M., Horenczyk, G., 1990. Moral balance: the effect of prior behaviour on decisionin moral conflict. Br. J. Soc. Psychol. 29 (1), 29–42.

Nobelprize.org (2017). Richard H. Thaler - facts. Nobelprize.org. Nobel Media AB 2014.Web. 14 Nov 2017. Link: http://www.nobelprize.org/nobel_prizes/economic-sciences/laureates/2017/thaler-facts.html.

O'Donoghue, T., Rabin, M., 1999. Doing it now or later. Am. Econ. Rev. 89 (1), 103–124.O'Donoghue, T., Rabin, M., 2008. Procrastination on long-term projects. J. Econ. Behav.

Organ. 66 (2), 161–175.Ölander, F., Thogerson, J., 2014. Informing versus nudging in environmental policy. J.

Consum. Policy 37, 341–356. http://dx.doi.org/10.1007/s10603-014-9256-2.Peschiera, G., Taylor, J.E., 2012. The impact of peer network position on electricity

consumption in building occupant networks utilizing energy feedback systems.Energ. Buildings 42 (8), 1329–1336. http://dx.doi.org/10.1016/j.enbuild.2012.03.011.

Peschiera, G., Taylor, J.E., Siegel, J.A., 2010. Response–relapse patterns of building oc-cupant electricity consumption following exposure to personal, contextualized andoccupant peer network utilization data. Energ. Buildings 42 (8), 1329–1336. http://dx.doi.org/10.1016/j.enbuild.2010.03.001.

Reisch, L., Zhao, M., 2017. Behavioural economics, consumer behaviour and consumerpolicy: state of the art. Behavioural Public Policy 1 (2), 190–206. http://dx.doi.org/10.1017/bpp.2017.1.

Schubert, R., Stadelmann, M., 2015. Energy-using durables – why consumers refrain fromeconomically optimal choices. Front. Energy Res. 3 (7), 1–13. http://dx.doi.org/10.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

209

3389/fenrg.2015.00007.Schultz, P.W., Nolan, J.M., Cialdini, R.B., Goldstein, N.J., Griskevicius, V., 2007. The

constructive, destructive, and reconstructive power of social norms. Psychol. Sci. 18(5), 429–434. http://dx.doi.org/10.1111/j.1467-9280.2007.01917.x.

Schultz, P.W., Estrada, M., Schmitt, J., Sokoloski, R., Silva-Send, N., 2015. Using in-homedisplays to provide smart meter feedback about household electricity consumption: arandomized control trial comparing kilowatts, cost, and social norms. Energy 90 (1),351–358. http://dx.doi.org/10.1016/j.energy.2015.06.130.

Seyranian, V., Sinatra, G.M., Polikoff, M.S., 2015. Comparing communication strategiesfor reducing residential water consumption. J. Environ. Psychol. 41, 81–90. http://dx.doi.org/10.1016/j.jenvp.2014.11.009.

Shen, M., Cui, Q., Fu, L., 2015. Personality Traits and Energy Conservation. Energy Policy85, 322–334. http://dx.doi.org/10.1016/j.enpol.2015.05.025.

Staats, H., Harland, P., Wilke, H.A.M., 2004. Effecting durable change: a team approachto improve environmental behavior in the household. Environ. Behav. 36, 341–367.http://dx.doi.org/10.1177/001391650326016.

Thaler, R.H., Sunstein, C.R., 2008. Nudge: Improving Decisions about Health, Wealth, andHappiness. Yale University Press.

Tiefenbeck, V., Staake, T., Roth, K., Sachs, O., 2013. For better or for worse? Empirical

evidence of moral licensing in a behavioral energy conservation campaign. EnergPolicy 57, 160–171. http://dx.doi.org/10.1016/j.enpol.2013.01.021.

Tiefenbeck, V., Goette, L., Degen, K., Tasic, V., Fleisch, E., Lalive, R., Staake, T., 2016.Overcoming salience bias: how real-time feedback fosters resource conservation.Manag. Sci. https://pubsonline.informs.org/doi/pdf/10.1287/mnsc.2016.2646.

Verplanken, B., Weenig, M.W.H., 1993. Graphical energy labels and consumers' decisionsabout home appliances: a process tracing approach. J. Econ. Psychol. 14 (4),739–752. http://dx.doi.org/10.1016/0167-4870(93)90019-H.

Waechter, S., Sütterlin, B., Siegrist, M., 2015. The misleading effect of energy efficiencyinformation on perceived energy friendliness of electric goods. J. Clean. Prod. 93,193–202. http://dx.doi.org/10.1016/j.jclepro.2015.01.011.

Winett, R.A., Neale, R.A., Grier, H.C., 1979. Effects of self-monitoring and feedback onresidential electricity consumption. J. Appl. Behav. Anal. 12 (2), 173–184. http://dx.doi.org/10.1901/jaba.1979.12-173.

Winett, R.A., Hatcher, J.W., Fort, T.R., Leckliter, I.N., Love, S.Q., Riley, A.W., Fishback,J.F., 1982. The effects of videotape modeling and daily feedback on residentialelectricity conservation home temperature and humidity, perceived comfort andclothing worn: winter and summer. J. Appl. Behav. Anal. 15 (3), 381–402. http://dx.doi.org/10.1901/jaba.1982.15-381.

M.A. Andor, K.M. Fels Ecological Economics 148 (2018) 178–210

210