Residential energy consumption (REC), the second largest energy consumer in China, is responsible for about 12.8% of total energy demand with an average annual increase of 7.24% from 2000 to 2017 (Fan et al., 2020). With both direct and indirect CO₂ emissions taken into consideration, the residential sector, the major contributor to global warming, has contributed more emissions than the industrial sector (Qi and Li, 2020). Thus, it is of great significance for China to conduct energy conservation and carbon abatement in the residential sector (Zhang H et al., 2020). More recently, China’s structure of energy consumption has largely shifted toward electricity and away from direct burning of coal (Xie et al., 2020). The electricity consumption in the dwellings has seen even faster growth, with an average growth rate of 11.49% from 1990 to 2017 (NBS, 2019), so there is enormous untapped potential for energy saving in the residential electricity consumption (Zhang M et al., 2020). In this context, residential electricity saving in China is becoming a priority. Understanding the policy measures to reduce residential electricity consumption is an essential part of mitigating global warming, making it a prime target for intervention (Davis and Metcalf, 2016). One such intervention is the information feedback, which provides households with the amount of their electricity consumption and monetary costs over a specified time period to promote energy conservation (Ito, 2014). Information feedback addresses problems of incomplete information in markets, which is at the crux of the energy efficiency gap, thus helping the residents overcome cognitive or behavioral biases and make better decisions of electricity consumption (Ratner et al., 2008). This method has been implemented and proven to be successful for promoting behavior change from individual to group levels in various fields, such as public health and education (Aydin et al., 2018).

In the field of electricity, bounded rationality constrains effective planning for electricity consumption, because households have limited information on electricity consumption due to separation of consumption and payment (Gilbert and Zivin, 2014), along with complicated electricity tariff structures (Bushnell and Mansur, 2005), making it difficult for households to understand actual electricity use. Studies have found that only 56% of the households understand their monthly energy consumption (Brounen et al., 2013) and only about 27% of the residents have knowledge about the average price of electricity (Blasch et al., 2017). Therefore, provision of more information feedback will increase awareness and transparency of electricity use. Among feedback-based intervention, electricity bills are considered a low-cost strategy for motivating household electricity conservation and has received considerable research attention (Henryson et al., 2000). A second resource for electricity feedback is from smart metering technology, which helps to remove informational constraints on residential consumption due to advances in energy infrastructure (Carroll et al., 2014). All that feedback information could motivate people to increase awareness of and knowledge about energy consumption levels, and to develop new energy-efficient habits and adapt to real-time pricing more effectively.

Policymakers are increasingly looking to information feedback for guidance. EU Directive (2006/32/EC) on Energy End-Use Efficiency and Energy Services (Energy Services Directive), dating from April 2006, requires the Member States to provide information feedback, including metering, frequent informative billing, historic and normative comparisons, contact information and advice for improving energy end-use efficiency (EUR, 2006). Denmark sets forth a legal obligation to offer informative electricity bills presenting historic and normative comparisons along with environmental impact. Sweden is the first country in Europe to have started a large-scale deployment of smart meters and enacted statutes to encourage provision of monthly billings by smart meters for all customers from July 2009 (Fischer, 2008). Likewise, the United States government allocated over $4 billion to install smart meters in 2009, encouraging utilities to supply residents with access to their energy information (Chopra, 2011). Aware of the significance of information feedback for residential electricity conservation, China has implemented a multitude of policies and programs to increase information feedback capacity, such as deployment of in-house display and smart meters. However, these initiatives are often designed without accurate resident’s requirement in mind (Du et al., 2017).

In this paper, we have employed the data of majority of Chinese provinces from the large-scale fieldwork through the Chinese General Social Survey (CGSS) launched jointly by Renmin University and the Hong Kong University of Science and Technology in 2018 to analyze how the information feedback can influence electricity consumption in the residential sector. The CGSS is the first national representative survey project run by academic institution in China mainland for researchers and scientists to study the behavior of Chinese households (Li et al., 2019). In this survey, the information feedback is made available by sending emails to households about their electricity bills or made the consumption shown on the dashboard by using the smart meters. The findings have indicated that providing feedback by sending emails contributes to a substantial reduction in residential electricity consumption, but the deployment of smart meters has failed to achieve the similar result as electricity bills. These results have indicated that information feedback via electricity bills can be used as a strategy to address behavioral barriers and ultimately conserve electricity. Advanced meters alone are not sufficient to change household energy consumption practices in China. The real-time display of electricity expenditure can lead to energy-conservative behavior in households through the learning channel (Lynham et al., 2016). As such, further training about those infrastructures are needed to make consumers gain adequate information in an instantaneous and easily accessible way. Understanding the underlying mechanism of electricity information feedback thus helps to identify challenges and opportunities and offer concrete suggestions for China's future energy policy.

As the effectiveness of information feedback is receiving increasingly more attention, the research has shifted from the laboratory to field experiments (Levitt and List, 2009). However, we still have insufficient knowledge of the effectiveness of information feedback on residential energy consumption (Bernedo et al., 2014), especially in the context of China. In this paper, we evaluate the effects of information feedback via both electricity bills and smart meters using a large-scale fieldwork and a large sample in China. The study uses the propensity score matching to focus on the effectiveness of feedback intervention by collecting the data of households from the Chinese General Social Survey (CGSS).

Much literature has discussed over the effectiveness of information feedback on electricity consumption, but with a variety of conclusions by applying the different feedback mechanisms. A set of meta-analysis results indicate that information feedback is effective, and can encourage favorable changes in energy consumption behavior (Delmas et al., 2013; Karlin et al., 2015; Nemati and Penn, 2020). According to some information feedback studies, feedback initiatives could lower residential electricity consumption ranging from 1.1% to 20% (Fischer, 2008; Vine et al., 2013).

Accurate and frequent billing, as interactive feedback service, will give the households a much better sense of the electricity expenditure at different times of a year. Pon (2017) found that billing reports contributed to reducing residential electricity use. Carroll et al. (2014) demonstrated that households could curtail the electricity consumption by 8.7% and 5.4% when receiving monthly billing and bi-monthly billing respectively. The Electric Power Research Institute (2009) concluded that monthly feedback and daily/weekly feedback could promote energy saving by 9% and 8% respectively (Neenan et al., 2009). Gleerup et al. (2010) showed that providing information via messages and emails led to averagely 3% of reduction in total annual electricity consumption. Other studies, however, have found smaller effects or no effects. Andor et al. (2020) found that the electricity feedback based on home energy report could cut down electricity consumption only by 0.7%. Restrepo and Morales-Pinzón (2020) and Khanna et al. (2016) demonstrated that electricity consumption seems unaffected by the delivery of feedback information due to the fact that consumers do not know which activity generates the waste of electricity.

Smart meters could help residents to monitor and better understand their electricity consumption. Gans et al. (2013) and Schleich et al. (2017) indicated that the installation of a smart metering could help reduce electricity consumption by 11–17% and 5% per household respectively. Aydin et al. (2018) showed that feedback initiatives through the use of in-home displays could save electricity by around 20% on average. In China, Zhang et al. (2016) tested the effectiveness of real-time in-home displays on residential electricity consumption, and showed that the in-home displays reduced monthly electricity consumption by around 9.1% in Shanghai City (one of the big and highly developed cities in China). However, Matsukawa (2018) found that feedback from in-home displays only reduced electricity on the part of energy-using households rather than energy-saving households. Hargreaves et al. (2013) found that smart meters did not necessarily bring electricity conservation in the United Kingdom, which is close to the estimates documented in the study by Nilsson et al. (2014) and Khanna et al. (2016) who also found that provision of information feedback via smart meters had little significant impact on electricity consumption on the side of residents. The researchers have ascribed this result to residents’ lack of ability to comprehend and utilize the information from advanced intelligent devices.

The effectiveness of information feedback via electricity bills and smart meters in cutting down energy consumption varies widely across studies. Given the variety in reported results, it is essential to identify what types of information feedback can have a positive impact on electricity conservation and to understand the mechanisms behind households’ behavior in China. In particular, so far, few studies have identified the causal effects between information-based initiatives and residential electricity consumption at a national level covering both urban and rural residents in China, which is critical to policy intervention. This study tries to fill in this gap.

In light of the previous literature, this paper studies the effects of information feedback via emails and smart meters on households’ electricity consumption. Therefore, following hypotheses are put forward based on the arguments presented above.

Hypothesis 1: Feedback information could reduce electricity consumption through sending emails of electricity bills.

Information feedback promotes public awareness and knowledge, which makes residents change their consumption behavior and promoting energy efficiency (Oltra et al., 2013). These changes in household behavior contribute to a reduction in electricity expenditure. Numerous studies have concluded that the provision of electricity bills affects residential electricity consumption (Darby, 2006; Gleerup et al., 2010; Carroll et al., 2014; Pon, 2017). Thus we hypothesize that households that receive electricity bills tend to change their electricity-use behavior, resulting in lower electricity consumption.

Hypothesis 2: The installation of smart meters is conducive to the reduction of electricity consumption.

Smart meters could help households to better monitor their electricity consumption. Studies have examined whether and how smart meters influence electricity usage, but with different conclusions (Gans et al., 2013; Hargreaves et al., 2013; Nilsson et al., 2014; Schleich et al., 2017; Aydin et al., 2018; Matsukawa 2018). We assume that the Chinese households with smart meters are likely to consume less electricity.

Several matching algorithms are offered to further estimate the effects of information feedback on electricity consumption for robustness check of the regression results, as shown in Table 8. This is the estimated results of the average treatment effects on treated (ATT) by the single and four nearest neighbor methods with caliper (0.01) along with the kernel estimator and bootstrapped standard errors based on 500 replications of the data. All those matching estimators have provided similar results.

The study results of ATT show that electricity bills have significant influence on the reduction of electricity consumption for all matching algorithms. Households that receive electricity bills have their electricity consumption decreased between 15% and 23%. The significant effect of electricity bills on electricity consumption is in accordance with Vine et al. (2013) and Fischer (2008), who argued that information feedback could cut down consumption by up to 20%. The electricity bills provide the residents with receipts of printed bills, including electricity consumption quantity and expenditure. This kind of information feedback may inform households of their monthly consumption, so that they can compare it with historical records. However, the installation of smart meters increases electricity consumption. This is also in conformity with Nilsson et al. (2014) and Hargreaves et al. (2013) whose studies proved that the installation of smart meters did not necessarily reduce electricity consumption. A possible explanation is that the residents have limited access to electricity consumption information in spite of the installation of smart meters. In this survey, only 2% of the households’ smart meters are located inside the house and 16% of those households who have deployed smart meters do not know their electricity bills. Most smart meter screens in China only show the accumulated electricity consumption in kWh, so some households may not fully understand the meaning of the display. In addition, households have limited access to historical electricity consumption records via smart meters. By comparison, Zhang et al. (2016) showed that the in-home displays reduced monthly electricity consumption by around 9.1% in Shanghai City due to the fact that the study focused on real-time in-home displays, which could make households check bills online and compare history electricity usage and seek advices about potential energy saving. In this case, the residents are presented the electricity consumption data visually. Another reason why installation of smart meters increases electricity consumption smart is that the installation of smart meters is usually associated with high-income urban residents. Table 4 shows that the urban residents have higher potential to install smart meters. Those residents may have little motivation to conserve electricity because electricity expenditure only takes up a small proportion of their household income. Besides, a boomerang effect of smart meters will occur among the households whose electricity expenditure is relatively modest before the experiment (Matsukawa, 2018). Providing the households with smart meters would increase their electricity consumption by informing them of excessive electricity saving. Fischer (2008) also illustrated that the households with extremely low usage may increase their usage overall when provided with more information. Therefore, further studies are needed to investigate how smart meters could be wisely used to conserve electricity consumption.

The current paper has looked into the effects of information feedback on electricity consumption in the residential sector with the micro-level data from the Chinese General Social Survey (CGSS) by using a propensity score matching (PSM) method. The study shows that the feedback in the form of electricity bills proves effective in encouraging the residents to make efficient use of electricity, which is consistent with our hypothesis. This is because in the absence of feedback, the households lack information on their electricity consumption and are unaware of the potential for further electricity conservation (Tiefenbeck et al., 2018).

Providing feedback via electricity bills can inform the residents of actual electricity consumption and costs frequently enough to motivate them to control their own electricity consumption, thus resulting in adjustments and changes of households’ behavior and improving energy efficiency in the residential sector (Gans et al., 2013). Besides, historical information feedback serves as a self-comparison that can drive the residents by establishing personal norms (Promann and Brunswicker, 2017), which could lead to 10% reduction in electricity consumption (Wood and Newborough, 2007). Therefore, the households should be provided with the amount of electricity compared to a previous period or some pre-set standards. In addition, evidence has shown that peers’ or neighbors’ behavior significantly influences the residents’ behavior change and proves the effectiveness of social norms in response to information feedback (Allcott, 2011). Information feedback via social comparison can cut down energy consumption of private households, ranging from 1.2% to 30% (Andor and Fels, 2018). Those households, who receive both individual and comparative information feedback, are successful in curtailing energy consumption in both the short run and long run, especially when targeting high consumers of energy due to a higher energy-saving potential (McCalley and Midden, 2002).

Concerning frequency of electricity bills, Abrahamse et al. (2005) concludes that energy saving potential of feedback relies heavily on its frequency through the systematic analysis. Frequent information feedback has proven to be a successful policy intervention for promoting energy conservation. Therefore, the persistence of feedback should be taken into consideration when planning this type of intervention because the effects of feedback often disappear when it is removed (Ehrhardt-Martinezet al., 2010). Hence, from a policy perspective, the information provided to the households via electricity bills should include knowledge and tips regarding energy conservation, a comparison of usage history and energy consumption of neighbors, pro-environmental behaviors and their potential impact on the environment and climate change. Besides, the information feedback should be offered more frequently to remain the effects of feedback.

As for smart meters, our studies show that the installation of smart meters fails to exert positive effects on electricity conservation, which contradicts the hypothesis. The differences of the studies’ duration, sample size, country, methodologies applied among researches may make studies come to different conclusions. This finding is consistent with the results drawn by Hargreaves et al. (2013) and Khanna et al. (2016). The previous studies prove that the feedback via smart meters is closely associated with positive environmental attitudes and understanding of information that smart meters provide (Darby, 2006). However, a majority of the residents in China lack information about smart meters and thus cannot convert the information into action to conserve electricity. Therefore, the government and public utility should provide trustworthy advice and technical training to increase consumers’ understanding of smart meters to make the conservation possible.

Although the smart meters boast a relatively high penetration in China, making up more than 50% of the market, most resident users of those smart meters have limited access to the information as meters are installed outside the house. Smart meters for the households only perform functions of measuring usage data and fail to deliver adequate information to the residents (Du et al., 2017). Thus, smart meters limit the capability to inform the residents as it relies on the residents’ self-initiation with low residents’ engagement in energy conservation. Therefore, from a policy perspective, smart meters should bring all this information onto residents’ computers and mobile phones inside home in a form that is accessible, clear and easy to understand, facilitating the residents to curb their electricity consumption. Besides, as learning plays a prominent role in enhancing energy conservation in the residential sector (Lynham et al., 2016), the policy-makers should provide training and educational outreach programs about the usage of smart meters. With the help of additional training and education, smart meters help the households to access to the information about usage in real time and keep informed of the use of specific appliances where they can conserve electricity most effectively. Provision of appliance-specific feedback by using smart meters keeps the households better informed of the potential ways through which they could reduce electricity consumption, in particular, from high energy-use electrical devices. Overall, instantaneous direct feedback via smart meters and frequent, accurate electricity billing are needed as a basis for sustained electricity demand conservation, which provides the guidance for future policy formation.