تاثیر افزایش بهره وری در استفاده صنعتی از انرژی: تجزیه و تحلیل تعادل عمومی قابل محاسبه برای انگلستان

The conventional wisdom is that improving energy efficiency will lower energy use. However, there is an extensive debate in the energy economics/policy literature concerning “rebound” effects. These occur because an improvement in energy efficiency produces a fall in the effective price of energy services. The response of the economic system to this price fall at least partially offsets the expected beneficial impact of the energy efficiency gain. In this paper we use an economy–energy–environment computable general equilibrium (CGE) model for the UK to measure the impact of a 5% across the board improvement in the efficiency of energy use in all production sectors. We identify rebound effects of the order of 30–50%, but no backfire (no increase in energy use). However, these results are sensitive to the assumed structure of the labour market, key production elasticities, the time period under consideration and the mechanism through which increased government revenues are recycled back to the economy.

Governments and environmental pressure groups across the world are advocating energy efficiency programs for either energy security or environmental reasons (Cabinet Office, 2001, Carbon Trust, 2003, Department of Environment Food and Rural Affairs, 2005, European Environment Agency, 1999, Nordic Council of Ministers, 1999 and Schutz and Welfens, 2000). Whilst the conventional wisdom is that improving energy efficiency will lower energy use, there is an extensive debate in the energy economics/policy literature on the actual impact of such improvements. This debate focuses on the notion of “rebound” effects, according to which the expected beneficial impacts on energy intensities are partially offset as a consequence of the economic system's responses to the fall in the effective price of energy services that accompany the improvement in energy efficiency. The “Khazzoom–Brookes postulate” (Saunders, 1992) asserts an extreme form of this: that improvements in energy efficiency can actually increase the demand for energy, a phenomenon initially identified by Jevons (1865) and now known as “backfire”. There is general agreement that some degree of rebound is to be expected, so that if, for example, a 5% improvement in energy efficiency in a particular use will generate energy savings of 3%, rebound would be 40%.1 Of course, the key question is the pragmatic one: how big is this rebound effect likely to be? Empirical work has concentrated on measuring rebound effects in consumer services (Dufournaud et al., 1994, Greening et al., 2000, Small and Van Dender, 2005 and Zein-Elabdin, 1997). Moreover, existing studies generally focus on the “direct” rebound effects. This restricts the analysis solely to the energy requirements to provide the consumer services to which the efficiency improvement directly applies. Less frequently studied are the “indirect” and “economy-wide” effects that are associated with the relative price, output and income effects that will affect the consumption and production in other energy using industries. This decomposition of rebound into direct, indirect and economy-wide effects is first made by Greening et al. (2000), who also point to a shortage of empirical studies on the “non-direct” rebound effects. A recent UK House of Lords (2005, p. 29) report sums up the present position as follows: Absolute reductions in energy consumption are thus possible at the microeconomic level. However, this does not mean that an analogy can be made with macroeconomic effects. Apart from anything else, the substitution effects observable at the macroeconomic level cannot be replicated by households, where demand for a range of goods is relatively inelastic… a business on the other hand, could respond to cheaper energy by deliberately increasing consumption — using a more energy intensive process, which would allow savings to be made elsewhere, for instance in manpower. The House of Lords report seems to be making two points here. First, that energy savings in production sectors are likely to have stronger indirect and economy-wide impacts than energy saving in consumption activities. Second, that energy substitution possibilities might be substantially greater in production than consumption. In this paper we wish to tackle the question: how large are the rebound effects likely to be for general improvements in energy efficiency in production activities in a developed economy? Specifically, does an increase in the efficiency by which energy is used in industrial production processes raise or lower the consumption of energy by industry? Further, to what extent does this response differ across industries? The method that we adopt is to undertake simulations with an economy–energy–environment computable general equilibrium (CGE) model for the UK. Using CGE modeling to analyse this problem has both strengths and weaknesses. A key strength is that CGE models have a strong grounding in conventional economic theory. They give an appropriate treatment of the supply-side changes resulting from supply-side policies and allow the net impacts of energy policy to be considered against a clear “counter-factual”. CGE models are also able to deal numerically with the simultaneity prevalent where major economic changes occur and to identify the orders of magnitude, not only the direction, of the resulting economic effects. However, the approach does also have weaknesses. One is the data required to operationalise the model: a set of multi-sectoral accounts is needed, together with a large number of behavioural and technical parameters. Further, Sorrell et al. (2004) argue that for the specific case of energy efficiency, the conventional neoclassical relationships which CGE models typically use might fail to capture some of the significant barriers to the penetration of new technologies. Finally, the particular assumptions made about the CGE model closure make comparisons across models difficult. For example, changing the assumption about the operation of the labour market can generate significant variation in the energy use results. Our key conclusion is that for our central case simulation, a general, across the board, improvement in efficiency in energy use in UK production sectors has rebound effect of the order of 55% in the short run and 30% in the long run, but no backfire (no increase in energy use). Sensitivity analysis suggests that this central case result is particularly sensitive to the imposed elasticities of substitution in the production hierarchy. However, the results also vary significantly with the nature of the assumptions made concerning the labour market, the way in which increased government revenues are recycled back into the economy and the time period under consideration. In Section 2 we briefly summarise previous theoretical discussions and sketch our own analysis of the likely system-wide ramifications of a stimulus to industrial energy efficiency. We conclude, as have many others (e.g., Saunders, 2000b), that the extent of rebound and backfire is an empirical issue. In Section 3 we describe our economy–energy–environment computable general equilibrium (CGE) model of the UK, UKENVI. In Section 4 we present the results of simulating an across the board stimulus to energy efficiency in production sectors and in Section 5 we discuss the results of our sensitivity analysis. Section 6 outlines the strengths and weaknesses of the CGE modeling approach. Section 7 concludes and offers some recommendations for future research.

The simulations reported in this paper suggest that for the UK, we expect a general, across the board, improvement in efficiency in energy use in production to have significant rebound effects but no backfire. Short-run rebound effects are above 50%, whilst the long-run values are around 30%. Increases in energy efficiency will reduce energy use, but not by the full proportionate amount. Moreover, these impacts will vary across energy types. Most of the rebound effect is captured within the demand for energy as an intermediate input. There are increases in the demand for energy driven by the accompanying expansion in final demand. However, in this particular case these increases are small. We test these results to see how sensitive they are to changes in key parameter values and model closure rules. This analysis backs up the generally held view that the energy use results are heavily dependent on the substitution possibilities in production in general, and the value taken by the elasticity of substitution between the energy and non-energy inputs in particular. However, note that we found that these results are also sensitive to assumptions made about the nature of the labour market, the conceptual time period under consideration and the recycling of government revenue saving. Our analysis shows that it is quantitatively important how we treat any changes in government revenues that result from changes in economic activity and employment that accompany improvements in energy efficiency. We have enforced a government budget constraint in which the additional government revenue is recycled through two alternative channels. Recycling this revenue through raising government expenditure delivers a significantly smaller economic impact than recycling it through reducing the average rate of income tax. With the income tax adjustment, the demand side is affected by increased household consumption, whilst the supply side is simultaneously stimulated via a lower nominal wage being required to generate any given real take-home wage. This makes substitution towards labour more attractive, and production more competitive, boosting employment, GDP and increasing long-run rebound to 40%. This work could usefully be extended in a number of ways. First, we have dealt here solely with improvements in energy efficiency across production sectors, and have excluded efficiency gains in household energy use, where a significant portion of energy policy is directed. Second, analysis could be focused on specific energy efficiency improvements in individual sectors, rather than general, across the board improvements applying to all production sectors.