"Show Me the Energy Costs": Short and Long-term Energy Cost Disclosure Effects on Willingness-to-pay for Residential Energy Efficiency.

• Author: Carroll, James

1. INTRODUCTION

   While more energy efficient products generally cost more upfront, their lower running costs can make them better investments over their lifetime. Household failure to make this intertemporal trade-off is known as the energy paradox (Jaffe and Stavins 1994b) and the corresponding underinvestment in energy-saving technologies is known as the energy efficiency gap (Jaffe and Stavins 1994a). Such missed energy-saving opportunities have private and public welfare implications.

   The existence of such investment 'inefficiencies' has been questioned in the literature (Allcott and Greenstone 2012). (1) For example, in relation to property investment, studies show that the future cost savings associated with energy efficiency are fully capitalised into property prices (Myers 2019, Aydin, Brounen, and Kok 2020, Papineau 2017). The size of the energy efficiency gap also clearly depends on the assumed energy savings at the time of investment, and many studies show that savings are below technical expectations (Levinson 2016, de Wilde 2014, Sunikka-Blank and Galvin 2012).

   This paper explores the existence of a specific information problem which may partly drive the energy efficiency gap--that of missing or biased energy cost forecasts at the point of sale. Much policy to date has attempted to eliminate energy consumption information gaps through energy efficiency labelling. For example, the current EU labelling system for appliances contains an energy consumption estimate (kWh/annum) and a colour-coded energy efficiency grading system. Within rental markets, mandatory efficiency labelling helps to alleviate the informational asymmetries and split incentives which underpin the landlord-tenant problem (Davis 2012, Souza 2018, Carroll, Aravena, and Denny 2016, Melvin 2018). For property sales, energy efficiency improvements within a mandatory rating system show clear sales price premiums (Brounen and Kok 2011, Hyland, Lyons, and Lyons 2013, Fuerst et al. 2015, Stanley, Lyons, and Lyons 2016, Jensen, Hansen, and Kragh 2016, Fuerst and Warren-Myers 2018, Frondel, Gerster, and Vance 2020).

   However, converting this (arguably) technical information into an energy cost forecast over the investment duration is not a straightforward exercise for the typical household: prospective buyers must combine consumption estimates with information (and expectations) on energy prices, level of use and duration of investment. Furthermore, in an upgrade situation, households would need to be aware of the energy intensity of existing technologies.

   This paper uses a Discrete Choice Experiment to explore if providing energy cost information in monetary terms could help overcome this 'missing information' problem and would increase willingness to pay for energy efficiency. We also consider if the duration of energy cost forecasts is an important consideration in willingness to pay. While there has been some research conducted to date exploring these questions for appliances and private vehicles, to the best of our knowledge, this paper is the first to examine these issues for properties, a 'product' with relatively high energy consumption.

   Our results indicate that energy cost information increases willingness to pay (WTP) for energy efficiency and that long-term energy cost forecasts (in our case, annual) increase the WTP for energy efficiency whereas the short-term forecast (2 months) has no effect (findings similar to Heinzle (2012). Our findings therefore provide additional evidence that individuals are, to some degree, missing the long-term costs implications of their energy-related investments, and that such information increases the demand for energy efficiency. Thus, contrary to the results of Aydin, Brounen, and Kok (2020), who find that energy efficiency premiums (in property sales) are unaffected by information provision (with and without EPCs), this paper suggests that the framing of such information is key to energy adoption rates. This is also the first paper, to our knowledge, to test the framing of energy efficiency information in the residential property market, a 'product' with relatively high energy consumption.

   The paper is organized as follows: Section 2 describes the literature and the hypotheses to be tested in the paper, Section 3 presents the DCE methodology and our application, Section 4 presents the results and Section 5 concludes.

2. LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT

   This section provides an overview of the salient literature on the energy efficiency gap and develops the hypotheses which will be analysed in this paper. The literature highlights a number of potential explanations for the energy efficiency gap (Allcott and Greenstone 2012), however, here we focus specifically on the explanation relating to the ability of consumers to convert energy consumption information (typically in kWh) into a monetary forecast.

   Heinzle (2012) and Allcott (2011) provide evidence of the high investment transaction costs associated with calculating and comparing models and the resulting errors in forecasts. Studies also show that many households do not understand the components of these calculations, such as typical energy use or the price of a kWh (Sovacool and Blyth 2015, Brounen, Kok, and Quigley 2013). These findings demonstrate that a potentially large proportion of households are unaware of past energy consumption and have difficulty converting technical consumption information into an energy saving forecast required for a rational investment. Furthermore, some buyers may simply be inattentive to energy costs at the point of sale (Allcott and Greenstone 2012). For example, Turrentine and Kurani (2007) show that vehicle owners are unaware of their past fuel expenditures.

   Allcott (2011) presents similar findings and shows that 40% did not consider fuel costs when buying their last vehicle (a further 35% did consider but did not carry out an energy cost calculation). Sallee (2014) suggests that inattention is not necessarily irrational behaviour when there is low variation in energy costs within product groups and high investment transaction costs (rational inattention). Houde (2018) distinguishes between consumer types and their engagement with information complexity (annual appliance energy costs/kWh versus simple binary "Energy Star" ratings). He finds that lower-income groups are more likely to value neither information types, and that higher-income groups are more likely to rely on the continuous (cost/kWh) information.

   So far this literature points to two general observations: first, it is likely that some buyers do not consider energy costs at the point of sale (inattention); second, for those who are attentive, it seems likely that a high proportion will make errors. From a policy perspective, the key question is whether or not this omission or bias is reducing the uptake of more energy efficient technologies. Our first hypothesis is based on these findings:

   H1: Framing product energy consumption in monetary terms increases the demand for energy efficiency, compared to framing consumption in physical units. For those buyers who are currently inattentive to energy costs, it is possible that energy cost labelling would bring the energy efficiency attribute into their consideration set. For those buyers who are attentive to energy costs but make errors in their calculations, energy cost labelling will only increase their demand for energy efficiency if the energy savings (from monetary labels) are higher than expected.

   A number of previous studies (primarily stated preference) have explored the impact of monetary labels. In the US, Newell and Siikamaki (2014) find that the willingness-to-pay (WTP) for efficient water heaters is highest when annual consumption costs are combined with more general informative aids (Energy Star and EU Efficiency Grades). For more efficient lightbulbs, similar effects are observed by Min et al. (2014) and Blasch, Filippini, and Kumar (2017). Andor, Gerster, and Sommer (2017) find that adding annual operating cost information to EU labels (fridges) increases the probability of choosing higher energy efficiency levels.

   The duration of energy cost forecasts (for example, years versus months) may affect the demand for energy efficiency. Heinzle (2012) finds that consumers will pay a higher price premium for energy efficient televisions when ten-year costs are displayed but a lower premium when one-year estimates are displayed (compared to non-monetary energy efficiency labelling). She attributes this result to the 'reversed pennies-a-day' effect (following Gourville (1998)). (2) The importance of longer time horizons is also highlighted by Alberini, Banfi, and Ramseier (2013) who use a conjoint choice experiment to examine attitudes towards energy efficiency retrofits. Those findings motivate our second hypothesis:

   H2: Framing energy cost labels over longer durations increases the demand for energy efficiency, compared to framing costs over short durations. Other studies have compared long-term or 'lifetime' energy cost labelling directly to non-monetary labelling. Kallbekken, Saelen, and Hermansen (2013) find that lifetime labels combined with staff training reduces the mean consumption of tumble dryers in Norway by 4.9%. Using an online field experiment for washing machines, Deutsch (2010) show a small but significant reduction in energy use (0.8%). In the UK, DECC (2014) show similar reductions for washer-dryers (0.7%), but find no effect for washing machines or tumble dryers. (3) We are unaware of prior studies which explore the effects of energy cost labels in the residential property sector, a "product" with a very high share of household energy expenditure.

   We test hypotheses H1 and H2 using a discrete choice experiment (DCE) which explores renters' valuation of a number of housing attributes, including energy efficiency. As our benchmark, we use the current...