Urban Residential Energy Demand and Rebound Effect in China: A Stochastic Energy Demand Frontier Approach.

• Author: Du, Kerui

1. INTRODUCTION

   Climate change has been one of the most urgent issues across the world, and it is widely accepted that the surging fossil energy consumption during the past three hundred years is the main cause (IPCC, 2014). In particular, China has been the most active economy all over the world, accompanied by booming energy demands that make the country become the greatest primary energy consumer as well as the largest carbon dioxide emitter (IEA, 2017). Referring to the China Statistical Yearbook 2016, China's average growth rate of total energy use is 5.14% from 2006 to 2015, while the rate in the urban residential sector is 6.19%.' Given the fast-increasing energy demand, the share of urban residential energy use in total energy use rose from 9.69% to 11.65% during 2006-2015. With the growing income of China's urban residents, their energy consumption and the corresponding share are expected to increase continuously. Therefore, the urban residential energy demand has become a crucial driving force of China's rising energy use, and hence it must be under control to meet China's energy-policy targets.

   To take responsibility for tackling climate change, China has made ambitious plans for energy demand control and energy conversion, including the control over residential energy consumption (Fransen et al., 2015). According to those plans, one of the fundamental ideas is to improve energy use efficiency through different measures. For example, the end-use energy efficiency may account for over 30% of total [CO.sub.2] reduction in China from the basic scenario to the 'sustainable development scenario' during 2010-2040, according to calculations by the International Energy Agency (IEA, 2017). However, the existence of the energy rebound effect might make energy-use reduction plans in the residential sector full of uncertainty. To be specific, energy efficiency improvements may induce relative changes in the prices of energy service to other goods. They may loosen consumers' budgets, which in turn can, directly and indirectly, increase energy use to some extent. As a result, the actual energy saving may be smaller than the potential energy saving of energy efficiency gains (Greening et al., 2000; Saunders, 2000). In this sense, the degree to which China's climate-change-mitigation plans can be achieved depends on the scientific estimation and effective mitigation of the energy rebound effect.

   The literature on the energy rebound effect is emerging. Many studies have explored the topics on the methodological definition of the energy rebound effect (Saunders, 2000; Sorrell, 2007; Saunders, 2008), microeconomic-level measurement (Sorrell and Dimitropoulos, 2008; Stapleton et al., 2016), economy-wide measurement (Shao et al., 2014; Brockway et al., 2017; Zhang and Lin Lawell, 2017), etc. The studies on sector-level rebound effects are relatively less and mostly focused on industrial sectors (Li and Lin, 2015; Li and Lin, 2017; Zhang et al., 2017; Liu et al. 2019; Li et al. 2019). For instance, Li and Lin (2015) made a detailed measurement of the energy rebound effect in 21 industrial sectors in China over 2006-2010. Liu et al. (2019) compared the direct rebound effect between China's heavy and light industries. Regarding the energy demand of China's residential sector, only a few studies took account of the energy rebound effect (Lin and Liu, 2013). For example, Lin et al. (2013) found that the energy rebound effect of the Chinese residential sector was approximately 22% over 1987-2008 by using a simulation approach. Lin and Liu (2013) focused on energy pricing reform in China and found that electricity tariff adjustment can reduce the energy rebound effect in the residential sector to a large extent. As argued by Orea et al. (2015), neglecting the energy rebound effect may cause bias when exploring the issue related to residential energy demand. Recently, some micro-level studies on Chinese households' energy use have argued that the magnitude of residential energy rebound in China is relatively large (Wang et al., 2014; Wang et al., 2016). For example, Liu et al. (2016) developed a life-cycle energy rebound effect model. They used a survey data set of 400 households in China to find that the life-cycle rebound effect in the Chinese residents' air-conditioner consumption was 67%.

   Moreover, there are some methodological issues related to the studies on China's residential energy rebound effect. First, the energy rebound effect fundamentally stems from energy efficiency improvements. Still, the existing studies on China's residential energy efficiency seldom account for climate factors, such as temperature changes, which can affect residents' use of space heating devices or space cooling devices. Second, the previous measurement of the energy rebound effect is generally based on several strong assumptions that may cause some estimation biases. For instance, a widely-used estimation approach of the energy rebound effect is to calculate the elasticity of energy demand with respect to energy price, which imposes a strict assumption of the symmetric relationship between energy-price elasticity and energy-efficiency elasticity. However, whether energy price would symmetrically change as energy efficiency improves is a pending theoretical issue to be carefully examined (Sorrell, 2007; Sorrell and Dimitropoulos, 2008). Besides, some recent studies provide methodological innovations by estimating the energy rebound effect with two steps, i.e., firstly estimating energy efficiency and then calculating the energy-efficiency elasticity of energy demand (Adetutu et al., 2016). Such a two-step process typically needs careful control over potential biases in each step, for accurate energy-rebound estimation.

   Under such a background, this study uses the stochastic energy demand frontier model to investigate the degrees and determinants of China's urban residential energy demand and energy rebound effect for the first time, based on a panel-data set of 30 provincial-level regions over the period of 2000-2014. Compared with the previous studies, the main contributions of this paper are two-fold. First, this paper employs a newly-developed method that can avoid potential biases in the existing rebound-effect studies. On the one hand, we adopt a direct estimation approach of the energy rebound effect, i.e., the stochastic energy demand frontier approach, which can simultaneously estimate the energy efficiency and energy rebound effect by one step. Thus, we can avoid the restrictions in the methods using price elasticity or using the proxy of energy efficiency change to improve the estimation accuracy of the rebound effect. On the other hand, we examine the influencing factors of energy demand in China's urban residential energy demand and identify the determinant factors of the rebound effect.

   Second, based on the pioneering contribution of Orea et al. (2015) in the methodology, we contribute to the existing literature by providing a detailed picture of the energy rebound effect of China's residential sector at the provincial level for the first time. As we have known, China is a large country with evident regional variations in many aspects, such as geographical conditions and resource endowments (Du et al., 2014). In particular, there are uneven development levels of regional economies. It is natural to be questioned whether and what extent there are variations of the energy rebound effect in the urban residential sector among China's different regions. To our best knowledge, almost no studies are devoted to this specific issue. It is, therefore, the purpose of this study to fill such a gap. Also, this study is expected to provide some precise policy reference for effectively mitigating the energy rebound effect and achieving energy-saving targets in China and even other developing countries.

   The rest of this paper is structured as follows. In the next section, we describe the methodology and used data. Then, Section 3 presents and discusses the estimation results of the stochastic energy demand frontier model, energy efficiency, and energy rebound effect in the Chinese residential sector. In the last section, we draw the main conclusions and raise some policy implications.

2. METHODOLOGY AND DATA

   In this paper, we use a stochastic energy demand frontier model that was firstly proposed by Filippini and Hunt (2012) and adapted by Orea et al. (2015) for directly estimating the energy rebound effect. We then discuss the econometric specification of the model based on the available data.

   1. Definition of the Energy Rebound Effect

      Among various approaches to estimate the energy rebound effect, Saunders (2000) proposed a direct measurement and defined the energy rebound effect (R) as follows:

      [mathematical expression not reproducible] (i)

      where [[epsilon].sub.E] is the elasticity of energy demand with respect to energy efficiency improvement and can be written as follows:

      [mathematical expression not reproducible] (2)

      where q represents actual energy demand, and E is the energy efficiency level. For example, if the energy demand reduces by 0.5%, given a 1% improvement of energy efficiency, the[[epsilon].sub.E]equals to -0.5, and the rebound effect is 50%. This means that the rebound effect offsets 50% of potential energy savings. Generally, there are five classes of the energy rebound effect according to the value of R (Saunders, 2008; Wei, 2010): (1) Backfire effect, that is, if R>1, the energy efficiency improvement causes more energy demand than before; (2) Full rebound, that is, if R=1, energy demand does not change after energy efficiency improves; (3) Partial rebound, that is, if 0

   2. Stochastic Energy Demand Frontier Model without the Rebound Effect

      As shown in Eq. (2), the direct measurement of the rebound effect requires the observation of energy efficiency that is typically unknown in reality. Filippini and Hunt...