Cost-effective pollution-abatement in an input-output model

Production activities exert a negative impact on the environment through emissions of pollutants. Environmental policies aim at the implementation of abatement technologies to reduce these emissions. Cost-effectiveness analysis is often applied as a tool to find abatement strategies that yield a certain emission reduction at minimum direct abatement cost to the producer concerned. The strength of this tool is the possibility to compare various abatement technologies at a rather detailed level. However, in a cost-effectiveness analysis the influence of the abatement costs on the rest of the economy is neglected. The eventual effect of the increase in production cost on the price of intermediates and final demand is not taken into account. In this paper we present the first steps towards an input-output model that will give in to these objections. A standard environmental input-output model is extended in such a way that it is possible to evaluate the cost and emission reduction of various abatement strategies. The model is formulated as an optimization model, minimizing total (economy-wide) production cost. The model results in a list of specific end-of-pipe abatement technologies that have to be applied in various sectors in order to achieve the national or sector-specific emission reduction targets for one or more pollutants at least cost for the economy as a whole. Moreover, under the assumption that the abatement costs can be passed on to the consumers (and neglecting the possible consequences for competitiveness on the international market), the model will show the price change of final demand commodities. This will help policymakers to evaluate the eventual distribution of the abatement costs over society. Introduction The Input–Output framework has been developed for the quantitative description of system interdependencies. The framework can be used to model interdependencies in both the money economy and in the physical world. It therefore proofed to be useful for the description of interdependencies between the economic system and the ecological system. Examples of early applications for environmental modelling are the works of Cumberland (1966), Daly (1968), Ayres and Kneese (1969), Leontief (1970) and Victor (1972). The Input-Output framework has been used in many studies to analyse the environmental impact of technological changes and changes in final demand since these early studies. To analyse the impact of prices, levies or for cost-efficiency studies the Input-Output frameworks proofed to be less suitable. If the impact of price-changes (caused by taxes or levies) is limited to a change in final demand volume or composition everything is still fine (Symons et al., 1994; Cornwell and Creedy, 1996). But if the impact also includes changes in producers behaviour the Input-Output 01 June 2007 DRAFT version – not to be quoted Pag. 2 van 17 system runs into problems. There is only a limited amount of studies on the interdependency between technological choices and prices within the Input-Output literature; the study by Duchin and Lange (1992) is one of them. Moreover, the extended input-output framework seems hardly fit to study the impact of emission taxes on the reduction of emissions (Idenburg and Steenge, 1991). Many environmental-economic models have been constructed for the purpose of evaluating the economic impacts of environmental policies. On the one hand there are top-down models, that evaluate the system from aggregate economic variables. On the other hand there are bottom-up models, that consider technological options or project-specific policies (Markandya et al., 2001). Topdown models assess macro-economic impacts of environmental policies, but disregard the specific abatement technologies that have to be implemented. Bottom-up models focus on specific abatement options, but cannot deal with the indirect economic effects induced (Dellink, 2003). For environmental policymakers rather detailed information on abatement options is important. Therefore, bottom-up models often are used in the process of environmental policymaking. An example of a bottom-up model that played an important role in the negotiations in Europe that led to emission reduction targets for air pollution is the RAINS model (Alcamo et al., 1990; Amann et al., 1999). This model includes detailed data on the cost and effect of abatement options for emissions of air pollutants from various activities. The model allows to evaluate what abatement options have to be implemented in order to achieve specific targets with respect to air pollution at least cost. Brink (2003) and Klaassen et al. (2004) present extensions of this modelling approach to include both air pollutants and greenhouse gases. These models can provide detailed data on the direct cost of abatement, but they do not allow to analyse the impact of abatement on the economy and to evaluate the consequences of abatement for the prices of commodities. Therefore, in this paper we propose a way to integrate the bottom-up, cost-minimizing approach and the environmental Input-Output framework. The proposed model allows to link a rather detailed descriptions of an economic system by an InputOutput model, to rather detailed information on abatement options. In the paper we first present an evaluation of the use of the environmental Input-Output framework and various shortcomings for environmental policy analysis, particularly with respect to air pollution issues. Then a description and the equations are presented of an extension of the standard environmental Input-Output framework, that makes it more suitable for environmental policy evaluation. Finally, we present some results of a numerical application of the proposed model for the Dutch economy, evaluating abatement of CO2 and NOx emissions from the production sectors. Environmental policy in IO models Since Leontief's (1970) well known extension of input-output analysis to include environmental issues, an increasing amount of literature has occurred on environmental input-output models. These models are based on a standard input-output model augmented by (i) pollution, generated by industries as a by-product to their normal economic production, and (ii) pollution-abatement, i.e. activities by 01 June 2007 DRAFT version – not to be quoted Pag. 3 van 17 purification sectors, eliminating the pollution produced by the conventional industries (e.g. Lowe, 1979; Idenburg and Steenge, 1991; Lager, 1998; Luptacik and Böhm, 1999; Nakamura and Kondo, 2002). Following the reformulation of the Leontief environmental input-output (EIO) model as suggested by various authors (e.g. Lowe, 1979; Idenburg and Steenge, 1991; Luptacik and Böhm, 1999), the model can be written as:       =             − − − − 0 2 1 22 12 12 11 y x x A I A A A I α α (1) where A11 is the square matrix of conventional input-output coefficients; A12 is the coefficient matrix of economic inputs per unit level of abatement activities; A21 is the matrix showing environmental pollution per unit of production by the conventional sectors; A22 the matrix showing pollution generated as a by-product of abatement activities; x1 is the vector of production levels of the conventional sectors; x2 shows the levels of abatement activities; y is the vector of final demand for conventional goods; α is a diagonal matrix with the percentage of the pollution which has to be eliminated. Leontief's extended system as represented by equation (1) has become an important framework for addressing economy-environment relationships. The approach is, however, characterized by a number of assumptions that cause some problems with the implementation of the model for environmental policy analysis. These have been pointed out and dealt with in several studies. In the following, we discuss three of them, that are relevant for the analysis of cost-efficient environmental policies with respect to air pollution. First, pollution is supposed to be eliminated once it is released into the environment (surface water, atmosphere, etc.). Although this might be the case for certain types of pollution (like waste for example), for most gaseous substances (like greenhouse gases and air pollutants), once released into the atmosphere it is hardly possible to eliminate them (Lager, 1998). Instead, pollution has to be reduced at the source through the use of less polluting alternative production technologies. This can be achieved by substitution of the conventional production technology by less polluting production technologies or by applying add-on abatement technologies to the conventional production technologies. This has two important implications: (i) abatement activities (and their cost and effect) are directly related to the pollution at the various specific sources, and (ii) different substitution and add-on technologies will be available for each of the various sources, which implies that the cost of reduction and the reduction potentials are sector-specific. This brings us to the second problem, viz. in the standard EIO model a choice among alternative production and abatement techniques is not allowed. It is assumed that there is exactly one production process for each good and exactly one abatement activity for each type of pollutant. However, in reality, several types of abatement methods will be available at different costs. Moreover, sometimes it 01 June 2007 DRAFT version – not to be quoted Pag. 4 van 17 is possible to apply multiple abatement methods together to a single pollutant and also to treat multiple pollutants simultaneously in a single abatement process. This gives problems with the traditional way input-output models are solved, because choice of technique implies that there are more processes than products and matrices will be rectangular. Lowe (1979) and Luptacik and Böhm (1999) give a reformulation of the Leontief environmental input-output system which allows for choice of technique, by formulating the input-output model as an optimization model. Nakamura and Kondo (2002) present a generalization of the Leontief environmental input-output model that "can deal with an arbitrary combination of treatment methods applied to an arbitrary combination of waste type provided that the combinations are technically feasible" (Nakamura and Kondo, 2002). In their model, the allocation of waste to the various treatment methods is exogenous to the model. Finally, in the standard EIO model, it is assumed that the degree of abatement (i.e. the proportions of pollutants eliminated, represented by α in (1)) is exogenous to the model. Moreover, the proportional emission reduction is the same for each sector. With abatement taking place once pollutants are released into the environment this might be the right way to do, because the abatement cost for a unit of pollution are the same, regardless the source of pollution. The approach implies that the cost of abatement is spread over the sectors according to their relative contribution to total pollution. In the context of sector-specific abatement (at varying cost) this will not result in an efficient use of scarce resources to reduce environmental pollution. In fact, it reflects the instrument of environmental policy called command and control, prescribing the same abatement technology for each sector. Standard environmental economic theory shows that this will be suboptimal from a welfare maximizing perspective. With economic policy instruments, like tax and subsidy schemes and tradable permit systems, the sectoral degrees of abatement are determined by the market. This results in an efficient (i.e. minimizing the cost for the economy as a whole) way to reduce environmental pollution (see also Lager, 1998). Lager (1998) presents a model to find activity levels of processes and abatement techniques and a set of prices such that a given final demand for commodities can be met without violating the environmental constraints at minimum cost. Environmental IO model for cost-effectiveness analysis To be able to analyze cost-effective reduction of emissions of greenhouse gases and air pollutants we propose an extension of the EIO model that will meet the above mentioned problems. The various IO models including the environment (starting with Leontief (1970) and further developed by others (including Lowe, 1979; Luptacik and Böhm, 1999; Nakamura and Kondo, 2002)) serve as a starting point. Most important differences with the existing models are (i) the direct link between abatement options and the specific sector to which abatement options can be applied; (ii) endogenous determination of the abatement strategy (i.e. the various abatement options that will be applied in order to achieve the given emission targets; and (iii) inclusion of the price model to analyze the impact of abatement on prices of goods and services. The model is formulated as an optimization problem, 01 June 2007 DRAFT version – not to be quoted Pag. 5 van 17 minimizing the total cost of production (i.e. the gross national product (GNP) at factor cost) at which the society satisfies final demand and the environmental targets that may exist. Base model First a standard environmental IO model is described, which will be extended subsequently. In order to be able to include cost-effectiveness analysis (i.e. cost minimization) into the model, the model is formulated as an optimization model (see also Lowe, 1979; Luptacik and Böhm, 1999): minimize ( ) 1 1 1 x v x V ′ = (value added) (1)