Download the full study report
Oct. 2008, pdf, 0.8 MB
This study by the Sectoral e-Business Watch explores links between ICT diffusion and energy consumption in different sectors at the aggregate level. There is not much economic research which uses quantitative analysis to determine the relationships between ICT usage and energy use. Qualitative studies have typically focused on the energy efficiency potentials. In contrast to these existing and mainly analytical-descriptive studies, this study aims to empirically test several hypotheses derived from economic theory. The motivation for undertaking this kind of analysis is twofold:
The role of ICT and e-business in shaping energy needs and energy consumer behaviour has significantly increased. ICT and e-business can help to reduce energy consumption and thus costs by reorganising production processes in a more efficient manner, but it can also lead to additional demand for energy due to new products and services provided and the energy consumption of the ICT capital stock itself. Hence, the overall impact of ICT on energy consumption is ambiguous, and depends on the relative magnitude of two countervailing forces:
Furthermore, there might also be some substitution of ICT and energy for labour and other production input factors, so that it seems useful to look at the relative impacts of the various input factors.
Empirically, a certain decoupling of GDP and energy use has been observed. In the U.S., for instance, GDP and energy consumption grew on average by 3.2% and 2.4% annually in the “pre-Internet era” (1992-1996) and by 4% and 1% in the “Internet era” (1996-2000). Note, however, that this observed overall decrease in energy intensity, measured as the ratio between energy consumption and production, may not be the case in every single sector of the economy. Moreover, the ICT sectors in particular seem to be less energy intensive than the overall economy (U.S. figures for 1996: 4.4% vs. 0.8%). In contrast to energy intensity, the intensity of electricity use is rising in many countries. It is thus interesting to study this potential causality between the diffusion of ICT capital goods and the observed decrease in energy (or electricity) intensity of production.
Methodologically, however, the evaluation of the causality is more complicated, due to the manifold consequences of ICT diffusion on the structures of the economy and society. Romm (2000) suggests to distinguish two types of energy gains related to the diffusion of ICT capital: (1) efficiency gains, for instance due to improved management of an assembly line, and (2) structural gains, for instance due to lowered individual transport needs because of increased Internet shopping. While appealing at first sight, these two kinds of gains, and especially structural gains, are very difficult to quantify empirically.
This report surveys some of the relevant literature on ICT and energy consumption, and provides a description of the research objectives and hypotheses followed, the methodologies and data used, and the results obtained from this initial study. Specifically, it contains the synopsis of three case studies conducted on the role of ICT to reduce energy consumption, and the results obtained from the econometric analysis.
Econometric studies focusing on the links (and causality) between the diffusion of ICT capital and energy consumption (or energy intensity of production, respectively) are still scarce, and complement (typically case-based) expert analysis and microeconomic studies. Our empirical econometric research for selected countries and industry sectors indicates that ICT, at the aggregate level, may not necessarily reduce energy (electricity) intensity, let alone absolute levels of energy (electricity) consumption, and that generalisations have to be made with care. In this respect, we show that an analysis on a still more disaggregate level, where communications devices and computers and software are analysed separately, can yield important additional insights. The last-mentioned disaggregation, however, is only feasible for value terms, due to current data restrictions.
While it is premature to come up with far-reaching conclusions and policy implications from this initial research, we have gained the following empirical results:
Results based on the CFP model
Chemicals industry: Communications technology has a positive impact on electricity efficiency (0.4); We find some (weak) evidence for electricity-augmenting technical change (0.03); transport equipment has a negative influence on electricity efficiency (0.5); computers and software are found to have an insignificant influence on electricity efficiency (0.13-0.15).
Metals industry: Communications technology has a positive impact on electricity efficiency (0.11-0.13); computers and software exert a negative influence on electricity efficiency (0.4-0.5), which is in line with what Collard, Fève and Portier (CFP) (2005), have found for the French services sector; (weak) evidence for electricity-augmenting technical change (0.01);
Transport industry: We find an electricity-augmenting impact of technical change (0.01); communications technology has a positive impact on electricity efficiency (0.25); transport equipment has a negative influence on electricity efficiency (0.40).
Results based on the Cobb-Douglas model
Chemicals industry (volumes): The largest impact on gross output in the chemicals industry is exerted by material inputs (0.39), followed by energy (0.19) and service (0.08) inputs. ICT capital (0.04) and technical change (0.02) have the weakest effect on output, while non-ICT capital is statistically insignificant.
Metals industry (volumes): In the metals industry material inputs (0.57) exceeds all other input factors by far. The next largest effect is exerted by service (0.12) and energy (0.11) inputs, followed by non-ICT (0.06) and ICT (0.02) capital.
Transport industry (volumes): Service inputs (0.52) have the decidedly greatest effect among all input factors in the transport industry. Material inputs and ICT capital (0.06) are followed by non-ICT capital (0.02), while energy inputs and technical change do not have a statistically significant impact on gross output.
Chemicals industry (values): The two largest effects on gross output in the chemical industry are exerted by material (0.37) and service (0.30) inputs. Non-ICT capital (0.16) and energy inputs (0.06) are followed by computers & software (0.05) and communication devices (0.01).
Metals industry (values): Material inputs (0.57) have the largest impact on gross output in the metals industry by far. The next largest effect is exerted by service (0.15) and energy (0.03) inputs. Computers & software (0.02), communication devices (0.01) and technical change (0.003) have the weakest influence on the sector’s gross output.
Transport industry (values): The largest impact on gross output in the transport industry is exerted by service inputs (0.38), followed by material (0.13), non-ICT capital (0.10) and energy inputs (0.04). Computers & software and technical change (0.01) again have the weakest effect on output.
As complementary evidence at the firm level, three case studies were conducted on how companies use ICT-based tools to save energy and thus energy costs. The cases are intended to support the understanding of aggregate results, but also to indicate whether there could be an untapped potential which is not yet reflected in the aggregate picture at the sectoral level.
Erdemir is Turkey’s largest iron and steel producer, and accounts for 1.7% of Turkey’s entire energy consumption. The company used IT applications to bring together all its control systems under one switch, and to provide an on-screen Plant Information System as a means of monitoring energy consumption. This has resulted in energy savings of up to 5%, and an early-warning system for any anomalies (e.g. pressurised air leakages) within the production system.
Irish food producer Jacob Fruitfield has introduced an Energy Monitoring System to tackle deficiencies identified in an energy audit. The system has helped to reduce gas consumption by 9%, provided better understanding of consumption patterns, and has instilled greater energy awareness among staff.
Coop, Switzerland’s second-largest retailer has implemented an Energy Management System in an attempt to reduce electricity and heat consumption, and to meet its commitments under national climate policy. The system combines data collection from its 950 food retail stores with a comprehensive building management system, which makes sure that target values for temperature and consumption of fuel, electricity and water are met. It also oversees the recovery of energy from the cooling system, which has reduced heat energy demand by some 60%.
It is of great policy relevance to better understand whether and under what particular circumstances the promotion of ICT and e-business can actually help to reduce energy consumption by increasing energy efficiency. In the literature, the majority of the research undertaken so far is based on case studies and qualitative analysis and has focused on the residential sector (e.g. PCs) and service sectors (e.g. data centres), rather than on the industrial sectors, as it is done in the sectoral econometric studies presented in this report.
Due to the heterogeneity of industrial structures and ICT diffusion patterns across different industries, such an analysis is likely most useful at the sectoral level. Moreover, it seems useful to also look at the role of other input factors of production than energy, such as material input, service input, and non-ICT capital stock, for assessing their impact on production output.
The main results from this study can be summarised as follows
