Assessing the impact of industrial robots on manufacturing energy intensity in 38 countries

Highlights

• We investigate the impact of application of industrial robots on energy intensity.

• We find that industrial robots can significantly improve the manufacturing energy intensity.

• This effect works through technology improvement effect and technological complement effect.

• There exists heterogenous nexus between industrial robots and manufacturing energy intensity.

Abstract

Considering the continuing slowdown of the improvement in energy intensity around the world, it is essential to seek a more effective measure to address the dilemma of energy and sustainable development. To this end, this research attempts to provide fresh insight into the determinants of energy intensity from the perspective of industrial robots and an industry-based view. By applying the dynamic panel GMM estimate methodology to a new data panel that includes 38 countries and 17 manufacturing sectors, this study provides the first comprehensive assessment of the use of industrial robots on manufacturing energy intensity. We found that industrial robots could significantly improve manufacturing energy intensity, and our hypotheses passed a series of robustness tests. Moreover, this improvement effect works through the technology improvement effect and technological complement effect between industrial robots and labor. Finally, we found a heterogeneous nexus exists between industrial robots and manufacturing energy intensity. Specifically, industrial robots can exert influence on non-renewable energy intensity rather than renewable energy intensity. Compared to capital-intensive sectors, we found that the use of industrial robots mainly affected labor-intensive sectors. We also found that Industry 4.0 could promote the improvement effects of industrial robots on manufacturing energy intensity.

Introduction

Energy, one of the most vital factors of production, has become a prominent driving force in economic growth (Apergis et al., 2010; Apergis and Payne, 2009; Mishra et al., 2009; Liu and Lee, 2020; Lee, 2005; Lee et al., 2020). Economic growth is often associated with high energy consumption, and energy intensity may bring about and exacerbate a variety of negative external production impacts, including environmental pollution and global warming, which has become a principal threat to sustainable development (Lee et al., 2021a; Lee et al., 2021b; Schmidt and Sewerin, 2019; Wang and Feng, 2015). Hence, determining the factors affecting energy intensity and improving energy intensity may not only inhibit the increasing trend of global energy demand but also put people in a position to reach climate and energy goals and achieve sustainable economic recovery.

Existing literature in the field of energy economics has investigated and highlighted some factors contributing to energy intensity, such as per capita income (Agovino et al., 2019; Jimenez and Mercado, 2014), technological innovation (Wurlod and Noailly, 2018), urbanization (Farajzadeh and Nematollahi, 2018), trade openness (Pan et al., 2019; Rafiq et al., 2016), financial development (Pan et al., 2019), and energy prices (Tajudeen, 2021; Karimu et al., 2017). Although extensive research has been carried out on factors thought to influence energy intensity, no study has thoroughly investigated the nexus between industrial robots and energy intensity, especially regarding energy intensity in the manufacturing industry from an industry-based perspective.

An industrial robot, defined as an automatically controlled, reprogrammable, and multipurpose machine by the International Federation of Robotics (IFR), is a symbol of digital transformation and artificial intelligence (AI) application in the process of the manufacturing industry (Graetz and Michaels, 2018; Kumaresan and Miyazaki, 1999). Industrial robots have evolved as crucial engines in the fourth industrial revolution (Sherwani et al., 2020). In recent years, as shown in Fig. 1, there has been a marked rise in the number of installations and operational stocks of industrial robots in the manufacturing industry throughout the world. Moreover, there is mounting literature concerning the economic consequences of industrial robots, highlighting their effects on economic growth (Aghion et al., 2019; Berg et al., 2018; Zeira, 1998), productivity (Kromann et al., 2020; Ballestar et al., 2020), labor and wages (Acemoglu and Restrepo, 2020; Acemoglu and Restrepo, 2018), as well as technology improvements (Yun et al., 2016; Jung and Lim, 2020).

However, to date, no studies have explored the effect of industrial robots on energy intensity. Little is known about the nexus between industrial robots and energy intensity, and it is also not clear through which channels industrial robots can affect energy intensity. Intuitively, the impact of industrial robots on energy intensity is ambiguous. In particular, if we assume that total output is constant, installing and using industrial robots will eventually increase electricity consumption during production processes, enhancing energy intensity (defined as the ratio of energy consumption to total output). In contrast, if we relax the assumption above, we can perhaps deduce that industrial robots can increase not only electricity consumption but also total output, as industrial robots are typically associated with higher productivity efficiency; this indicates that the change in energy intensity depends on a comparison of the increase in both energy consumption and output. Therefore, identifying the nexus between industrial robots and energy intensity is an empirically indispensable undertaking. Note that this paper mainly concentrates on energy intensity in manufacturing sectors because manufacturing sectors are the largest users of industrial robots. In addition, our data on industrial robots provides detailed amount information at the level of the manufacturing sector. Indeed, according to the International Energy Agency (IEA), the share of total energy consumption in the manufacturing industry accounted for 37% of global total energy consumption in 2018,¹ making the manufacturing industry one of the primary sources of energy consumption worldwide. Thus, improving manufacturing energy intensity plays a vital role in achieving the eventual goals of reducing energy intensity in full.

To this end, based on a balanced cross-country cross-manufacturing panel covering 38 countries and 17 manufacturing sectors from 2000 to 2014, we apply a two-step generalized method of moments (GMM) estimator system to empirically evaluate the significance of industrial robots on manufacturing energy intensity. Specifically, the study addresses the following issues. First, we try to assess whether and to what extent industrial robots can affect manufacturing energy intensity. Second, this study examines possible channels through which industrial robots impact manufacturing energy intensity. Third, this paper sets out to better understand the nexus between industrial robots and manufacturing energy intensity—that is, to assess the heterogeneous nexus between the two from various perspectives (i.e., species of energy intensity, characteristics of manufacturing sectors, and the Fourth Industrial Revolution [Industry 4.0]).²

In sum, the possible contributions of our study are three-fold. Primarily, this work generates fresh insight into the impact of industrial robots on manufacturing energy intensity and possible routes through which industrial robots affect energy intensity. To the best of our knowledge, no existing study has investigated the effect of industrial robots on manufacturing energy intensity in a cross-country framework. It is noteworthy that our research differs from the extant literature investigating the impact of generalized technological progress on energy intensity (Wurlod and Noailly, 2018; Timma et al., 2016; Moshiri and Duah, 2016; Voigt et al., 2014). We focus on assessing the impact of targeting technological advances (i.e., the application of industrial robots) on energy intensity, allowing us to provide a comprehensive explanation of the impact mechanisms. Second, instead of utilizing cross-country datasets that have been widely applied to the research field of energy intensity, we merge the Industrial Robot database (IRD) released by IFR with the World Input-Output Database (WIOD) to obtain a cross-country cross-manufacturing industry dataset accounting for major economies throughout the world. This provides us with a central opportunity to advance understandings of the nexus between industrial robots and manufacturing energy intensity. Third, in order to reach accurate conclusions, we apply a dynamic panel data estimator—a two-step system GMM estimator—to our dataset instead of using a regular fixed-effect model.³ Compared to other static panel estimators, it not only accounts for the path dependence of energy intensity but also alleviates the concern caused by possible endogeneity through constructing instruments, thus generating more consistent and efficient estimations (Lee et al., 2021b; Lee and Lee, 2019).

The remainder of our paper is organized as follows. Section 2 reviews related literature from two perspectives. Section 3 proposes six research hypotheses. Section 4 presents econometric specifications and data. Section 5 reports empirical results, covering the impact of industrial robots and manufacturing energy intensity, robustness tests, heterogeneous tests, and mechanism tests. Section 6 reviews our main conclusions and proposes some policy recommendations.