Energy Cost Information and Consumer Decisions: Results from a Choice Experiment on Refrigerator Purchases in India.

• Author: Jain, Manisha

1. INTRODUCTION

   Research shows that energy-efficient technologies are cost-effective, yet they face multiple barriers in adoption (Ruderman et al., 1987; Sutherland, 1991), the phenomenon commonly referred to as the 'energy-efficiency gap'. Most studies highlight inadequate information as an important factor contributing to this gap (Jaffe and Stavins, 1994; Brown, 2001; Sanstad and Howarth, 1994). Appliance labelling is being used globally as a policy instrument to provide energy consumption information and facilitate cost-effective decision making. Most countries have adopted comparative categorical labels that divide the products into categories based on energy efficiency. The labels used in the majority of countries give information in physical energy units such as kilowatt-hour (kWh). Few countries where labels have energy use information in monetary units are Japan, USA, Canada, and Jamaica (Harrington and Damnics, 2014).

   Since the adoption of labels, studies have examined their impact on consumers willingness to pay and their effectiveness in design and content. Studies examining the impact of labels have found that consumers value the presence of labels and are willing to pay more for a labelled appliance (Sammer and Wustenhagen, 2006; Galarraga et al., 2011; Shen and Saijo, 2009; Ward et al., 2011; Jain et al., 2018b). Some studies have quantified consumer willingness to pay using market data (Galarraga et al., 2011; Mills and Schleich, 2010) and others have used stated preference surveys (Sammer and Wustenhagen, 2006; Shen and Saijo, 2009; Newell and Siikamaki, 2014; Davis and Metcalf, 2016). Few studies have reported a difference in the impact of labels across appliances (Shen and Saijo, 2009; Jain et al., 2018a). While consumer heterogeneity is an important factor explaining the energy efficiency gap (Gillingham and Palmer, 2014), studies identifying the magnitude and sources of heterogeneity in consumer willingness to pay for labels are limited (Jain et al., 2018a). Researchers analyze the observations on consumer choices from choice experiments using discrete choice models such as multinomial logit, probit or mixed logit models. Mixed logit models allow the model parameters to vary in the sample under an assumed distribution (Revelt and Train, 1998); however, their application in consumer response to labels has been limited (Ward et al., 2011; Jain et al., 2018a).

   Research on the effectiveness of labels is limited to US and European markets. Since the US Energy Guide label gives operating cost information, researchers have focused on the impact of alternative label formats, such as a simplified label (Newell and Siikamaki, 2014) or state-specific label (Davis and Metcalf, 2016). In the purchase of electric-water heaters Newell and Siikamaki (2014) found that simple information on the economic value of energy saving is the most crucial element guiding cost-efficient decisions. In air-conditioners buying decisions, Davis and Metcalf (2016) found that state-specific labels lead to better choices. There are several studies on the impact of energy cost information on the European Union (EU) label. These studies look at the impact of annual or lifetime energy cost information either on the choice of an energy-efficient appliance using choice experiments (Bull, 2012; Heinzle, 2012; Andor et al., 2020) or change in mean energy consumption of chosen appliance using actual/expected purchase data (Deutsch, 2010; Kallbekken et al., 2013; DECC and the Behavioural Insights Team, 2014; Stadelmann and Schubert, 2018). Blasch et al. (2017) studied the impact of operating cost information on the probability that consumers undertake cost-effectiveness analysis. The findings of these studies are summarised in Table A.1 in Appendix A.

   The difference in the impact is found across appliances within studies (Kallbekken et al., 2013; DECC and the Behavioural Insights Team, 2014; Stadelmann and Schubert, 2018), and for the same appliance across studies, e.g.tumble dryer (DECC and the Behavioural Insights Team, 2014; Stadelmann and Schubert, 2018). The findings on the effect of displaying energy cost on refrigerators are becoming less divergent with growing research. While Kallbekken et al. (2013) did not find any effect; two recent studies have found a positive impact of displaying energy cost on a refrigerator label, such as an increase in the choice of energy-efficient refrigerator (Andor et al., 2020) and increase in the probability that consumers perform cost-effectiveness analysis (Blasch et al., 2017). Blasch et al. (2017) found that energy consumption in monetary terms increases the probability that a consumer makes a calculation and identifies the refrigerators with the lowest lifetime cost. However, since the authors did not observe consumer choices, they did not estimate the magnitude of willingness to pay for the energy-efficient refrigerator. On the other hand, Andor et al. (2020) observed consumer choice in stated preference survey and analyzed the data using logit and probit models. The authors report the direction of change in consumer choice due to annual operating cost information but do not estimate the change in the magnitude and distribution of willingness to pay for higher efficiency class.

   In this study, we examine consumer decisions in Indian refrigerator market to confirm if operating cost information on labels facilitates cost-effectiveness analysis and, if it does, estimate consumer willingness to pay for a higher efficiency class. The research on the effectiveness of appliance labelling is growing but mainly in the developed economies. This study provides evidence of the effectiveness of appliance labelling in an emerging market. As discussed, the recent research on the impact of operating cost information on refrigerators shows a positive effect on consumer willingness to pay for energy-efficient products. Still, it does not estimate the magnitude of the change and how it compares with the associated savings. Studies using discrete choice experiments to determine the effect of monetary cost information have not considered statistical and design efficiency in the construction of hypothetical choice situations. The strength of choice models to identify the distribution of consumer willingness to pay has not been utilized so far. This study aims to expand the growing knowledge on the effect of monetary cost information on labels. Following Street and Burgess (2007); Huber and Zwerina (1996) we construct choice-sets ensuring orthogonality, level balance, minimal overlap and utility balance. We specify mixed-logit model and allow the model parameters to vary and correlate (Revelt and Train, 1998). This study provides robust estimates of the impact of annual energy cost information in the direction, magnitude and distribution of consumer willingness to pay for higher efficiency category.

   In a stated preference survey administered in a metropolitan city in India in May 2015, we interviewed households on their choice of refrigerators from alternatives differing in their attributes such as capital cost, brand and star rating. We divided the sample between the control and experiment group such that the respondents in the experiment group got the energy cost of the refrigerator alongside the star rating label. Confirming the recent findings in the literature, we show that unlike the existing labels, the labels with annual operating cost information facilitate cost-effectiveness analysis and allows the consumer to choose a cost-effective refrigerator. The consumer willingness to pay for the most energy-efficient category in the experiment group is estimated to be US$200 at 95% confidence interval of 36 104-296. To place these values in perspective, the mean price of the refrigerators considered in the study is US$285, and the associated lifetime savings in the operating cost due to increase in energy efficiency for a range of assumed discount rate (1-10%) is US$102-66. The share of the sample placing a higher value on the highest energy efficiency category as compared to a lower category increased from 54% in the control group to 76% in the experiment group.

   We have organized this paper into five sections. In section 2, we explain the labelling scheme in India with a focus on the design and content of the labels on different appliances. We describe the research design, including the model specification and the design of the discrete choice experiment in section 3. In section 4, we discuss the results explaining the estimation strategy and goodness of fit, model estimates and summarize key findings. In section 5, we present conclusions showing how we place our study in the existing literature and our relative contribution. We also discuss the limitations and policy suggestions.

2. INDIAN ENERGY LABEL

   India launched the appliance standards and labelling (S&L) program in 2006. It is implemented by the Bureau of Energy Efficiency (BEE) under the provisions of the Energy Conservation Act 2001. The program has a comparative five star-labelling system where 1-star is the least energy-efficient, and 5-star is the most energy-efficient. The number of stars gives information on the relative energy consumption of appliances. The program was initially voluntary and is gradually being made mandatory. The program was first made mandatory for wall-mounted air conditioners and frost-free refrigerators in 2010. There are ten types of equipment under the mandatory scheme and thirteen under the voluntary programme. The star rating on different types of equipment is given based on various technical specifications relating to energy consumption. For example, the star rating on refrigerators is given based on a range of electricity consumption calculated using total adjusted volume, constant multiplier (kWh/litre/year) and constant fixed allowance (kWh/year). The maximum allowed electricity consumption...