Firm-level Estimates of Fuel Substitution: An Application to Carbon Pricing.

AuthorHyland, Marie
PositionReport - Statistical table

    The ability of firms to substitute between energy inputs is key when considering both the effectiveness and the costliness of climate policy. Stem (2012) highlights that the more difficult it is for firms to substitute between clean and dirty fuels, the more expensive climate-change mitigation will be. The macroeconomic implications of fuel switching are highlighted by Acemoglu et al. (2012), who show that if clean and dirty inputs are substitutable in production, emission-reduction targets can be achieved without sacrificing long-term economic growth. Papageorgiou et al. (2013) note that the innovations in clean technologies needed to achieve significant reductions in C[O.sub.2] emissions will only occur if the right incentives are in place for firms to switch from dirty to clean production. The authors state that this depends on both economic policy and on the structure of the economy. Thus, to get a sense of the extent to which fuel switching is possible, as well as to provide guidance on the most appropriate policy instruments, estimates of fuel substitution at the firm level are crucial.

    In this paper we use six years of recent (2004-2009) panel data from manufacturing firms based in Ireland to analyze firms' ability to switch between different fuels.' We estimate both partial and total fuel substitution elasticities. This allows us to evaluate the likely effectiveness of carbon pricing in achieving a reduction in industrial, energy-related C[O.sub.2] emissions. There are a number of points worth highlighting: First, our estimates of the partial elasticities of substitution indicate that energy inputs are substitutes rather than complements in the production process. Electricity is the least price-responsive energy source. Unlike most other studies, we also calculate total elasticities of substitution. These take into account the feedback effects from individual fuel price changes to aggregate energy use. These estimates indicate a higher sensitivity of electricity demand to changes in its own price, but a weaker sensitivity to changes in the prices of other fuels. Second, we compare the elasticities of substitution estimated at the firm-level to estimates from aggregating our data to NACE letter and NACE 3-digit industry levels.(2) At industry level the estimated elasticities are very different and in particular for oil, the industry-level estimates indicate a much higher degree of substitution between electricity and oil than at the firm level. This confirms the aggregation bias present in studies at industry level. Third, based on the estimated elasticities we simulate the impact of introducing a carbon tax on industrial emissions. The results from the imposition of a domestic carbon tax of [euro]15/tonne CO, would achieve only modest reductions in C[O.sub.2] emissions from oil and gas use in the industrial sector.

    Relative to the literature we provide a study at the firm-level for a relatively recent period. During the period of analysis--2004-2009--there were significant fluctuations in economic activity and sharp changes in energy prices. We, thus, contribute to broaden the rather scarce evidence on interfuel substitution at the firm level. In a meta analysis Stem (2009) notes this lack of evidence and calls for additional research based on micro data and on a much wider variety of countries.(3) In general, studies of interfuel substitution find that electricity is the least price-responsive energy source, this is true for estimations at state, region and firm level. Furthermore, most research suggests that energy inputs are substitutable (as opposed to complementary) in the production process, although the extent of this substitutability varies significantly between studies. Compared to the literature, our estimated elasticities are mid-range relative to the related estimates at the firm level. Our results show that, according to partial price elasticities, electricity is the least price-responsive energy source; a result commonly found in the literature. Crucially, and unlike most other studies, we also estimate total elasticities of substitution. These indicate a higher sensitivity of electricity demand to changes in its own price, but a weaker sensitivity to changes in the prices of other fuels.(4) This implies that incentivizing firms to switch from fossil fuel to electricity usage on a large scale may not be possible using price instruments alone. (5)

    As the Irish manufacturing sector is dominated by high-tech manufacturing and pharmaceutical companies, it is characterized by low-levels of energy intensity unlike many larger countries. (6) Fuel substitution elasticities for Ireland will be of interest to other countries similarly characterized by low levels of energy intensity, for example, according to data from The World Bank (2012), other European countries with low levels of energy use per unit of output include Switzerland, Denmark, Portugal and Cyprus. Furthermore, for countries that are not comparable in terms of the aggregate level of energy intensity of their economies, these results remain of relevance to sectors within their economies that are characterized by lower levels of energy intensity.

    As has been pointed out by Solow (1987) in relation to the estimation of interfactor elasticities, using aggregate data will likely result in aggregation bias as changes in the combination of inputs used will be confounded with changes in the product mix. Solow shows how the existence of general equilibrium effects (whereby energy price shocks alter consumer demand patterns) may lead to an illusion of factor substitution, even when substitution is not technologically possible, and thus illustrates that estimates of substitution based on aggregate data can be unreliable. In the case of fuel substitution, an increase in the price of a particular fuel may change the relative prices of finished goods, causing changes in consumers' demand patterns. Studies based on aggregate data may infer from this that substitution has taken place, even in cases where substitution between fuel types is not technologically feasible, leading to elasticities that are biased upwards. Our results confirm this prediction for certain fuels. Moreover, studies based on aggregate data ignore the fact that, in some industries, substitution between fuel types may not be technologically possible.

    Finally, we demonstrate how these elasticities can be used to evaluate, ex-ante, the likely effectiveness of certain environmental policies. We do so by simulating the impact of a carbon tax on industrial emissions, a topic of growing international interest as more and more countries introduce some form of carbon pricing.' The results from simulating the introduction of a domestic carbon tax of [euro]15/tonne C[O.sub.2] suggest that such a policy would achieve only modest reductions in C[O.sub.2] emissions from oil and gas use in the industrial sector. In terms of policy, this highlights that trying to compensate for the currently poor functioning of the European Union Emissions Trading Scheme (EU ETS), which has been characterized by low carbon prices, by imposing a higher, domestic carbon tax on the firms that are not subject to the EU ETS is unlikely to be successful. Emissions from power generation are regulated under the EU ETS and, therefore, in Ireland the price of electricity does not increase when a domestic carbon tax is imposed as sectors covered by the ETS are exempt. If the carbon tax were also applied to electricity, (8) the emissions reduction would be considerably larger due to, firstly, the importance of electricity as an energy source in the Irish manufacturing sector; secondly, the high carbon intensity of grid-supplied electricity in Ireland; and, thirdly, the high total elasticity for electricity demand. Furthermore, as electricity is substitutable with oil and gas for the majority of firms in our sample, part of the decrease in emissions from oil and gas is offset by an increase in emissions from electricity. As a rise in the ETS price would affect all firms (because all firms use electricity), the same reduction in emissions could be achieved by a considerably smaller increase in the ETS price.

    The remainder of this paper is structured as follows: Section 2 discusses the related literature on fuel substitution. Section 3 outlines the methodology used. Section 4 gives an overview of the data. Section 5 presents the results. Finally, in Section 6 we draw some concluding remarks.


    Seminal papers in the literature on fuel substitution include those of Fuss (1977) and Pindyck (1979), both of whom pioneered the use of two-stage production models to estimate own and cross-price elasticities of demand for fuels. While Fuss (1977) and Pindyck (1979) use very different datasets (Fuss looks at the Canadian manufacturing sector, whereas Pindyck's analysis covers the industrial sector often OECD countries), both studies conclude that the price elasticity of demand for electricity is low.

    Since these seminal works, many authors have estimated fuel substitution elasticities using a variety of datasets and methodologies, and a heterogeneous pattern of results has emerged. Numerous authors have estimated elasticities based on industry or country level data for the US. For example, Jones (1995) employs a dynamic linear logit model and finds that electricity is the least elastic input, and that the relationship between most fuel inputs is one of weak substitutability. In contrast, Taheri and Stevenson (2002), find that light oil and coal are the least elastic inputs. More recently estimates from Serletis et al. (2010b) suggest that oil is least responsive to changes in its own price. They find that all fuels are substitutable, with the exception of gas and coal.

    Examples of other studies based on data from a single country include Floros and Vlachou (2005) and Steinbuks (2012)...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT