The Rebound Effect in Energy-Intensive Industries: A Factor Demand Model with Asymmetric Price Response.

• Author: Dahlqvist, Anna

1. INTRODUCTION

   In this paper the rebound effect of energy efficiency improvements in Swedish energy-intensive industries has been empirically assessed. The challenge of climate change, together with energy security concerns, has spurred an increased societal interest in energy efficiency improvements. A report presented by the UN Climate Panel IPCC in 2018 discusses what is required to limit global warming to 1.5[degrees]C compared to pre-industrial levels, thereby addressing the need for rapid and powerful measures to reduce emissions globally in all sectors of society. In this respect, the IPCC has stated, among other things, that the energy demand in industry must decrease, partially through increased energy efficiency (IPCC, 2018). The EU has committed to reducing its greenhouse gas emissions by at least 40 percent by 2030 compared to 1990 levels. As part of this commitment, the EU presented a package of measures in 2016, "Clean energy for all", in which one ambition is to prioritize improvements in energy efficiency. (1)

   The EU Energy Efficiency Directive has been revised within this package, and now includes updated and extended national energy savings requirements to 2030 (Article 7). The Commission declares that "the most efficient energy is the one which is not consumed--as it results in less energy use." (2) Thus, such an EU ambition addresses the importance of not implementing policy measures resulting in significant rebound effects. The EU energy efficiency target beyond 2020 states that, by 2030, the EU should have achieved an energy efficiency improvement of 32.5 percent compared to forecasts. (3) In Sweden, the government has decided on a target of 50 percent more efficient energy use by 2030 compared to 2005--the target being expressed in terms of primary energy in relation to GDP (Government Bill 2017/18:228).

   In line with this, in January 2017 the Swedish Energy Commission proposed that a specific energy efficiency program should be launched (Government Bill 2017:2). Accordingly, the so-called "Energisteget" (4) was implemented in 2018 with the purpose of encouraging energy efficiency improvements in Swedish industry. (5) Energisteget gives industrial firms the opportunity to seek financial support if they are investing in energy efficiency measures. At the same time, Sweden has the minimum permissible energy tax rate in the EU. The success of such a program in reducing energy consumption primarily depends on two conditions: First, the program must lead to energy efficiency investments that would otherwise have not been implemented (additionality). Second, the resulting improvements in efficiency must not lead to any major rebound effects that represent economic mechanisms that will offset the energy savings that result from energy efficiency improvements (Sorrell, 2014). In the present paper, we focus on the latter condition.

   Even though empirical studies confirm the existence of a rebound effect, and although this effect has caused political concern, it is seldom translated into real political action (Vivanco et al., 2016). Our main purpose is to estimate the rebound effect in the Swedish industry and relate our findings to the use of industrial energy efficiency programs as energy-saving policy measures.

   Mansikkasalo and Soderholm (2013) found that energy efficiency programs in Sweden have foremost attracted firms in energy intensive sectors. Hence, we focus on these sectors. Lundgren et al. (2016) found that there is a considerable potential for energy efficiency improvements in Swedish energy intensive industry. These findings, together with the recently implemented energy efficiency program "Energisteget", make Sweden relevant as a case study. In Sweden, the industry accounted for 38 percent of total final energy use in 2015, of which energy-intensive sectors; pulp and paper, iron, steel and non-ferrous metals, as well as the chemical industry, used around 76 percent (The Swedish Energy Agency, 2017).

   To approximate the magnitude of the rebound effect, we employ a factor demand model to estimate own-price elasticities for various energy inputs in production: fossil fuels, non-fossil fuels and electricity. We also estimate cross-price elasticities between these energy inputs and capital and labor.

   There are few studies that estimate industry rebound effects. Our study contributes to the literature by: (i) relating the rebound effect to the use of energy efficiency programs as energy-saving policy measures; (ii) adding to the evidence of whether a sector-specific direct rebound effect is present in each of the energy-intensive industries: pulp and paper, iron and steel, chemical and mining, for which we have a unique and detailed firm level data set, covering the period from 2001-2012. Since the rebound effect is closely linked to a decline in the effective price of an energy source, our research also contributes by (iii) employing an approach that allow for an energy demand that responds differently to increasing and decreasing energy prices. Finally, (iv) our study adds to previous work by addressing the fact that the rebound effect caused by energy efficiency improvements ought to be studied in a broader perspective than is usually the case. More efficient use of energy can be the result of both technological developments and improvements in technical efficiency. The latter has been proposed and studied by Orea et al. (2015) and Amjadi et al. (2018) applying US and Swedish data, respectively. The overall impact of technological developments and improved technical efficiency can collectively provide a different and more complete picture of the rebound effect than the image given by these two separate factors individually. The analysis conducted in this paper focuses on the rebound effect that stems from technological developments only. This means that, to a certain extent, it complements Amjadi et al. (2018).

   The rest of the paper is organized as follows: In the next section, energy efficiency programs are briefly discussed. In section 3 the rebound effect is defined and the literature on the subject is reviewed. In Section 4 the factor demand model is outlined theoretically and empirically. Theoretically, the model is based on profit maximizing firms. Empirically, the profit function is specified as a quadratic function in a system with input demand functions. Finally, substitution elasticities are specified. Data is presented in Section 5 and the results are outlined in Section 6. Section 7 concludes the paper and presents some policy implications.

2. VOLUNTARY PROGRAMS

   The Swedish energy efficiency goal is expressed as an intensity target (primary energy divided by GDP). However, the intensity and efficiency concept differ. If the ambition is to reduce energy intensity, policies could aim to achieve a rapid GDP development in relation to energy use. In this case, the ambition is not to reduce total energy use. Nonetheless, a major part of Swedish energy policy focuses on energy-saving measures, thus implying a political will to reduce total energy consumption.

   However, Swedish energy-intensive industries face considerable energy tax credits compared to other sectors such as the residential sector. For instance, the tax paid on electricity amounts to 0.5 Euro/MWh, which is the minimum permissible energy tax rate approved by the EU Energy Tax Directive. One political argument is that this is necessary to prevent domestic firms from moving their production abroad (Governmental Bill 2015:87). Thus, industry policies are instead designed to encourage energy efficiency improvements through different investment support schemes, etc. For example, voluntary agreements that focus on industrial process energy use have been implemented. Such agreements not only exist in Sweden but also in several other European countries including Belgium, Denmark, Estonia, Ireland, Slovenia and the Netherlands.

   In Sweden, the Program for Improving Energy Efficiency in Energy Intensive Industries (PFE) was launched in January 2005 and lasted until 2012. (6) PFE was a five-year program and firms that participated were fully exempted from energy taxation on electricity. In return, the firms committed to introduce an energy management system and assess the potential to improve energy efficiency. They also had to implement the improvements revealed by the assessment before the program expired (The Swedish Energy Agency, 2012).

   Exempting energy-intensive industries from energy taxation was not compatible with EU state aid rules. Thus, Sweden has launched "Energisteget", a successor to the previous PFE program. This is a new voluntary scheme in Sweden and, similarly to the PFE, its aim is for firms to implement energy efficiency measures. According to Swedish law, large firms are required to assess their process or plant energy use (Swedish Code of Statutes 2014:266). Instead of tax exemption, firms may seek investment aid (a lump sum subsidy) to implement energy efficiency improving investments identified in their mandatory energy assessment.

   The effectiveness of voluntary programs such as the PFE has been questioned. According to the Swedish National Audit Office (2013), it is doubtful whether the PFE led to significant energy efficiency improvements. Instead, changes in energy-relevant practices were mainly caused by rising energy prices during the program period (Mansikkasalo and Soderholm, 2013). In line with this, Dahlqvist and Soderholm (2019) suggest that the energy price constitutes the most important determinant of inter-firm differences in energy intensities and that neither so-called informational nor organizational failures are main drivers of energy efficiency improvements. Instead, firms in which energy (and hence also the cost of energy) constitutes an important input in production have incentives to become "energy aware". Still, firms may apply "rules of thumb" which, in some respects...