Explaining electricity demand and the role of energy and investment literacy on end-use efficiency of Swiss households

Highlights

• The level of transient and persistent energy efficiency of Swiss households is estimated.

• We apply the newly developed generalized true random effects model (GTREM).

• The analysis is based on (unbalanced) panel data covering 1,994 households over 5 years.

• Significant transient (11%) and persistent (22%) inefficiencies are found.

• High energy and investment literacy is associated with lower electricity consumption.

Abstract

This paper estimates the level of transient and persistent efficiency in the use of electricity in Swiss households using the newly developed generalized true random effects model (GTREM). An unbalanced panel dataset of 1994 Swiss households from 2010 to 2014 collected via a household survey is used to estimate an electricity demand frontier function. We further investigate whether energy and investment literacy have an influence on the household electricity consumption. The results show significant inefficiencies in the use of electricity among Swiss households, both transient (11%) and persistent (22%). We note that the high persistent inefficiency is indicative of structural problems faced by households and systematic behavioral shortcomings in residential electricity consumption. These results indicate a considerable potential for electricity savings and thus reaching the reduction targets defined by the Swiss federal council as part of the Energy Strategy 2050, wherein end-use efficiency improvement is one of the main pillars. The results support a positive role of energy and, in particular, investment literacy in reducing household electricity consumption. Policies targeting an improvement of these attributes could help to enhance efficiency in the use of energy within households.

Introduction

In Switzerland, electricity is primarily produced by hydropower plants (60%) and nuclear power plants (40%). In 2011, after the Fukushima Daiichi nuclear accident, the Swiss federal council decided to abandon nuclear energy. For this reason, the Swiss federal council developed a new energy policy concept, called Energy Strategy 2050. One important goal of this strategy is to reduce electricity consumption by improving the level of efficiency in the use of electricity and to increase the share of electricity produced with new renewable sources of energy such as wind and solar. The efficiency improvement and the development of new renewable sources should, therefore, allow substituting the amount of electricity produced by nuclear power plants. In this context, the residential sector is characterized by great potential for energy efficiency gains and could make an important contribution to a reduction of total end-use electricity consumption.¹

Against this background, it is important for policy makers to have information on the potential for electricity savings in the residential sector. Moreover, it is important to know which are the determinants that influence the level of efficiency in the use of electricity. A low level of efficiency, as discussed in Filippini and Hunt (2015), may be due to the fact that households do not adopt and use energy efficient appliances or do not use their appliances in an optimal way. For instance, a household might postpone substituting an old and inefficient refrigerator that consumes a lot of electricity, or does not use a cooling system or washing machine in the most efficient way.

The determinants of residential energy efficiency have been widely covered in the economic literature Gillingham et al., 2009, Allcott and Greenstone, 2012, Feb, Frederiks et al., 2015. The potential explanations for an inefficient use of appliances on the one hand and for an under-investment in energy-efficient household appliances on the other can be attributed to either market failures or behavioral failures (Broberg and Kazukauskas, 2015). Market failures that prevent investments in energy-efficient appliances can take the form of information problems (e.g., lack of information and information asymmetries), misplaced incentives and principal-agent problems such as the landlord-tenant problem. But even if these market failures could be overcome, several behavioral failures such as bounded rationality, loss aversion, status-quo bias, risk aversion or inattentiveness² potentially reduce the level of efficiency in a household's energy use. All these behavioral failures tend to prevent households from identifying the appliances that minimize lifetime costs or from using the appliances in an efficient way. However, as shown by Blasch et al. (in press), households that are scoring high with respect to investment and energy literacy seem to be less prone to boundedly rational behavior.

To our knowledge, relatively few studies have looked into the relationship between energy and investment literacy and residential energy efficiency (for an example, see Brounen et al. (2013)). Investment literacy can be defined as the ability to perform an investment analysis and to calculate the lifetime cost of an appliance or energy-efficient renovation. Energy literacy can be defined as an individual's cognitive, affective and behavioral abilities with respect to energy-related choices. According to DeWaters and Powers (2011), energy literacy comprises an individual's or household's (1) knowledge about energy production and consumption and its impact on the environment and society; (2) attitudes and values towards energy conservation; and (3) corresponding behavior. In this paper, we therefore put particular emphasis on examining the influence of energy literacy, investment literacy and energy-saving behavior on a household's level of efficiency in the use of electricity.³

Hence, in this paper, we provide an answer to the following questions: Which are the factors that influence the electricity demand at the household level? What is the level of efficiency in the use of electricity of Swiss households? How large are the potentials for energy savings in the residential sector for a given level of energy services? Does a household's level of energy and investment literacy influence its level of efficiency in the use of electricity?

To answer these questions, it is important to remember that a household's energy demand is not a direct demand for energy or electricity, but rather a derived demand for the production of energy services such as warm food, clean clothes and lit rooms. Therefore, behind electricity demand there is a production function. A reduction in energy consumption for the production of a given level of energy services can be achieved either by improving the level of efficiency in the use of inputs (i.e. in the use of appliances), or by adopting a new energy-saving technology (i.e. purchase of new appliances, investments in energy-saving renovations), or both. Technological change can induce a reduction of energy consumption for a given level of energy services, provided that the inputs are used in an efficient way, i.e. given that the households are productively efficient. The total reduction in residential energy consumption is therefore a result of the interplay of technological change and a household's behavior.⁴

The level of energy efficiency of households can be measured with a bottom-up approach, by making an on-site efficiency analysis of buildings. However, with such an economic-engineering approach, the behavioral aspects in energy use are often not accounted for. In addition, this approach is not based on the microeconomics of production. In this paper, we therefore estimate a household's level of energy efficiency with econometric methods, accounting for total electricity consumption and factors such as the size and characteristics of the dwelling, household composition and other socioeconomic attributes, level of energy services consumed, energy literacy, investment literacy and energy-saving behavior. With this approach a broader and more adequate bench-marking of Swiss households with respect to their electricity consumption can be performed.

The existing literature on the measurement of the level of energy efficiency in the residential sector using an economic approach is relatively sparse. While the Stochastic Frontier Analysis (SFA) has been used with aggregated energy data (e.g., Filippini and Hunt, 2012, Filippini et al., 2014, we use dis-aggregated data since residential consumers are typically very heterogeneous and household level data can add more detail to the knowledge of consumer response. Weyman-Jones et al. (2015) are one of the first to estimate energy efficiency using SFA with dis-aggregated household survey data. They estimate an energy input demand frontier function, originally proposed by Filippini and Hunt (2011), using a cross-sectional household dataset from a survey in Portugal. However, the model used by Weyman-Jones et al. (2015) is relatively simple with only a few explanatory variables. In contrast, Boogen (2017) uses a much richer model using not only the information on appliance stock but also on the amount of energy services consumed to estimate the technical efficiency of a set of Swiss households using a sub-vector distance function. However, as Boogen (2017) uses a cross-sectional dataset, the unobserved heterogeneity cannot be accounted for. Moreover, only the level of technical efficiency is estimated. Alberini and Filippini (2015) employ an energy demand frontier approach similar to Weyman-Jones et al. (2015) using a large panel dataset from US households to estimate the level of energy efficiency. By using panel data they are able to distinguish and estimate the level of persistent and transient energy efficiency.⁵ The limitation of Alberini and Filippini (2015) is that the amount of energy services consumed by a household was not included as an explanatory variable.

In this paper, we follow the energy demand frontier approach using an unbalanced panel dataset of 1994 Swiss households from 2010 to 2014. Moreover, using an approach proposed by Coelli et al. (1999), we will also measure the level of efficiency by comparing the electricity consumption of all households to the optimal level obtained from an energy input demand frontier function associated with a high level of investment literacy.

The contribution of this paper is twofold – firstly, we estimate the persistent and transient efficiency in electricity consumption of a large sample of Swiss households and demonstrate an application of the newly developed GTREM model Colombi et al., 2014, Filippini and Greene, 2016 that estimates both types of efficiency conveniently by a simulated maximum likelihood approach. We benefit from a unique panel dataset covering a five-year period collected via a household survey conducted in 2015. The dataset includes information on the level of energy services, which is usually not measured as it can be difficult to collect this information.⁶ Information on the level of energy services is a critical issue when using SFA (Filippini et al., 2014). Finally, to our knowledge, this paper is the first to provide a systematic analysis of the impact of both energy and investment literacy on the total electricity consumption of households while controlling for the effects of the general level of education of the household members. Our results can therefore provide new insights into the interrelations between the literacy and education variables and their role for transient and persistent efficiency in residential electricity consumption.

The rest of the paper is organized as follows. Section 2 discusses the role of energy literacy and investment literacy for energy efficiency. Section 3 presents an econometric model of residential electricity demand using dis-aggregated household data and discusses the empirical specifications for estimating the level of efficiency in the use of electricity. Section 4 describes the household survey data and the variables used in the model. Section 5 presents the results and Section 6 concludes.