European Industries' Energy Efficiency under Different Technological Regimes: The Role of C[O.sub.2] Emissions, Climate, Path Dependence and Energy Mix.

• Author: Stergiou, Eirini

1. INTRODUCTION

   Industrial activities in Europe are in the foreground of the policy agenda imposing a heavy burden on the quality of the environment. Under this perspective, European industries have perceived the huge potential benefits from adopting energy saving and environmental friendly technologies aiming at a lower impact on the environment. A clean and energy saving manufacturing sector has been targeted as a key area for Europe particularly since the European Directive 2009/28/EC. Thus, building up an energy efficient European manufacturing sector can benefit European countries by improving their welfare with greater levels of energy independence and security, achieving the underlying objective of cost minimization and facing successfully the threat of energy rising prices. Additionally, it can evolve into a valuable asset for reducing C[O.sub.2] emissions, fulfilling Kyoto protocol, enhancing industry competitiveness and promoting economic growth through continual innovation. The efforts made by European Commission at this direction result a rather questionable outcome concerning energy efficiency (European Commission, 2011) highlighting the strategic role of its improvement.

   The Paris Agreement (2015) as well as the Climate Action Conference (2018) in Seoul and Katowice Summit (2018), where the importance of a threshold in global temperature was emphasized, are the most recent paradigms of the importance of environmental policies on energy and emissions. With this frame of mind, EU continues its commitment to reduce greenhouse gas emissions, be energy efficient and use energy from renewable sources by 20% in 2020 compared to 1990 levels. In continuance of the Kyoto Protocol (1997), the first international agreement on climate change, and Copenhagen Summit (2009) EU has set a new list of targets that need to be achieved by the end of 2030 and 2050 in order to become a low-carbon economy. More precisely, the Directives "Energy end-use efficiency and energy services" of 2006 (1) and "Energy efficiency directive" of 2012 (2) try to implement several policies for industries, such as financial, informational, legislative, fiscal measures and market-based instrument (EU Emissions Trading). Their main target group are the small-medium industries by diffusing accessible information on energy matters and the large companies by making audits for energy consumption. However, because of the economic-financial crisis (2008-2012) that struck Europe, the results for energy efficient industries and countries were not the expected ones revealing heterogeneous patterns.

   The last few decades the concept of energy efficiency has drawn a lot of attention from policymakers to managers and researchers. However, because of its nature, it is difficult to be calculated. The first researchers assessed it using the Index Decomposition Analysis (IDA) that had taken under consideration economic, structural and energy changes of an economy. As time has gone by, the notion of energy intensity (EI) became identical with the measure of energy efficiency. Nonetheless, this measure serves as a partial energy efficiency due to the absence of non-energy inputs in its estimation. This argument established the necessity for a new universal integrated approach.

   Although the estimation of TFEE is of great interest, the growth and its long-run performance consist equally important issues for investigation. Convergence analysis, a widely used concept in economics, has the advantage of exploring the evolution of cross-industry differences over time in energy efficiency levels. The existence of convergence patterns is substantial for economists and scholars in order to negotiate the multinational climate agreements as effective as possible (Patterson et al., 2014) while some particular industries would need additional attention in global settlements (Liddle, 2010). However, differences on spillovers across industries and countries, technological improvements in the energy production processes and knowledge diffusion, distinct levels of energy mix combustion and the use or lack of energy management systems are some of the causes that could easily lead to divergent paths. Therefore, convergence analysis could provide important information on the design and the evaluation of policies aiming on industrial energy efficiency improvements and the reduction of carbon impact on the environment associated with energy use.

   To fill this gap in literature this study performs the following exploratory research. Firstly, we deploy a fully nonparametric approach to perform benchmarking on TFEE scores across fourteen European industries over twenty-seven countries using Data Envelopment Analysis (DEA) and Directional Distance Function (DDF) approaches for the 1995-2011 period under a metatechnology framework. The incorporation of the metatechnology concept allows us to reveal possible differences on energy efficiency measures under a Frontier (F) (country) and a Metafrontier (MF) (European) technology framework and investigate technological heterogeneity. In addition, DEA and DDF methodologies are used to estimate differences arising from desirable-undesirable outcomes on TFEE estimates in this context. Secondly, we utilize panel tobit models to investigate possible factors that are likely to increase energy efficiency with respect to the different technological regimes and methodologies adopted. More precisely, we focus on energy and emission intensities, energy mix, technological and climatic characteristics. Finally, we perform [beta]- and [sigma]-convergence analysis for our TFEE estimates.

   The obtained results show that countries with small-scale economies are unexpectedly energy efficient. Moreover, when undesirable outputs are included in our analysis, scores change dramatically for the majority of industries. On the other hand, the omission of technological heterogeneity that incurs in Europe and the consideration that industries compete each other solely on the boundaries of a national level, lead easily to overestimated energy efficiency scores. In this case, several industries that are efficient in their country, become inefficient when they are compared with others outside of their economy. Furthermore, regarding energy efficiency's influencing factors, path dependence phenomena and climatic characteristics are in presence while the energy mix that utilizes each industry initiates linear and nonlinear relationships. Finally, testing for energy efficiency convergence patterns we found evidence of beta and sigma convergence which states that backward industries have been catching up with leading industries in terms of energy efficiency.

   The paper is structured as follows: The next Section reports the literature review, highlighting open issues, while Section three develops the methodological approach adopted in this study. Section four presents the data including a description of energy and C[O.sub.2] emissions across European countries. In Section five the results and discussion are given while Section six concludes and suggests for future research.

2. REVIEW OF LITERATURE

   Energy efficiency is a multifaceted concept with a vital role for industries' energy strategy. From the seminal paper of Berndt and Wood (1975), who engaged with the notion of energy and its substitution elasticity relation with capital and labor, to the different divisions of energy efficiency (3) (Patternson, 1996), several authors have tried to measure energy efficiency using different methodologies. The first attempts to measure energy efficiency were focused on the decomposition of energy consumption, through a bottom up framework, using Index Decomposition Analysis (IDA) (4) and on energy intensity, defined as the amount of energy used per unit of output-activity (Kounetas et al., 2012). However, IDA measures have a number of shortcomings which have become increasingly evident, as more sophisticated parametric and non-parametric models are brought to bear on distributional questions (Shorrocks, 2013). On the other hand, as energy intensity does not seem to take into consideration non-energy inputs in the production process, its establishment as a "partial" measure of energy efficiency is inevitable

   Bearing the above-mentioned shortcomings some authors have paid attention on the production theory framework (i.e. Filippini and Hunt, 2012). The basic idea is based on the estimation (using parametric or nonparametric approaches) (5) of a best practice frontier for the use of energy. Thus, energy efficiency can be easily computed as the difference of the actual and the predicted energy use by incorporating the TFEE index (Hu and Wang, 2006). In literature, relating to SFA, the derivation of the Shepard distance function allows for the estimation of shadow prices (Choi et al., 2012; Lin and Du, 2013; Llorca et al., 2017), the curvature of the sustainability along the frontier (Fare et al., 2005) and the derivation of "underlying energy efficiency" (Filippini and Hunt, 2012).

   Comparing the two approaches we can argue that the majority of the relevant studies on energy efficiency follows the nonparametric approach aiming on the TFEE calculation. A significant number of empirical studies is mainly concentrated on China (Wei et al., 2012; Wang et al., 2013; Shao et al., 2019) and India (Paul and Bhattacharya, 2004; Worrell et al., 2009), either by focusing on provinces or an industrial and firm level. On the other hand, the last few decades the concept of energy efficiency has amused progressively researchers to conduct, although on a minor extent, similar studies in Organisation for Economic Co-operation and Development (OECD), Asia-Pacific Economic Cooperation (APEC) countries and Europe (e.g. Hu and Kao, 2007; Zhou and Ang, 2008; Voigt et al., 2014). Notwithstanding, a divergence of various methods is occurred when the measure of energy efficiency is computed. Some...