توسعه یک مدل تعادل عمومی قابل محاسبه جهانی همراه با فن آوری های استفاده نهایی انرژی

A global computable general equilibrium (CGE) model integrating detailed energy end-use technologies is developed in this paper. The paper (1) presents how energy end-use technologies are treated within the model and (2) analyzes the characteristics of the model’s behavior. Energy service demand and end-use technologies are explicitly considered, and the share of technologies is determined by a discrete probabilistic function, namely a Logit function, to meet the energy service demand. Coupling with detailed technology information enables the CGE model to have more realistic representation in the energy consumption. The proposed model in this paper is compared with the aggregated traditional model under the same assumptions in scenarios with and without mitigation roughly consistent with the two degree climate mitigation target. Although the results of aggregated energy supply and greenhouse gas emissions are similar, there are three main differences between the aggregated and the detailed technologies models. First, GDP losses in mitigation scenarios are lower in the detailed technology model (2.8% in 2050) as compared with the aggregated model (3.2%). Second, price elasticity and autonomous energy efficiency improvement are heterogeneous across regions and sectors in the detailed technology model, whereas the traditional aggregated model generally utilizes a single value for each of these variables. Third, the magnitude of emissions reduction and factors (energy intensity and carbon factor reduction) related to climate mitigation also varies among sectors in the detailed technology model. The household sector in the detailed technology model has a relatively higher reduction for both energy intensity and the carbon factor.

Integrated Assessment Models (IAM) are widely used in climate mitigation analysis. For example, the following models are all well-known IAMs: AIM/CGE [1], GCAM [2], IMAGE [3], MESSAGE [4], and ReMIND [5] and [6]. These models more or less couple economy, energy, GHG emissions, agriculture, land use, and climate components. If the scope is broadened to not only this type of large-scale IAM but also to simple energy models, there are many more examples. One is the Asian Modeling Exercise [7], which compiles the results of 23 models. Although there is no clear definition, energy models are generally classified into two types depending on how they represent energy technologies: the so-called bottom-up (BU) type model, which has a detailed representation of energy technologies, and the top-down (TD) type model, which uses either a production function or price elasticity to represent aggregated energy technologies. Furthermore, there are two classes of TD type models depending on the extent to which goods are dealt with. One is the computable general equilibrium (CGE) model, which covers all goods and services transactions, whereas the other is known as the Partial Equilibrium (PE) model, which treats specific goods (e.g., energy goods). AIM/CGE, EPPA [8], and IMACRIM-R [9] are classified as CGEs, and GCAM 1 and TIMER (which is a part of a module within IMAGE) [10] are PEs. There are many studies using CGE models if the models are not limited to IAMs (e.g. [11], [12] and [13]). The advantages of BU models are the disadvantages of TD models and vice versa. TD models are easily able to represent the heterogeneity of energy technology selection in non-linear functions, but the representation of energy technology is aggregated and it is difficult to assess whether the solutions are feasible from a technological point of view. In addition, TD CGE models deal with all of the transactions for goods and production factors. They can assess the responses of the macro-economy and the prices of all goods to interventions, such as a carbon emissions constraint. The BU model can simulate more realistic technological descriptions of energy and assess the technological feasibility of climate mitigation targets. BU models, however, are usually linear to minimize total costs and the cheapest single technologies are chosen without extra constraints even if actual consumer behavior would not be as extreme.2 Several studies have tried to complement each type model’s advantages and disadvantages. MESSAGE did this with MESSAGE-MACRO [14], and CIMS [15] maintains overall consistency by exchanging information with a macroeconomic module. Previous studies using the CGE model have not fully integrated the BU structure within it. CGE is generally a large-scale model, and two methods have been used to account for both TD and BU. One is to prepare a BU model outside of the CGE model and exchange information with the BU model more than one time. Drouet et al. [16] and EPPA [17] input the outcomes of a BU model into energy consumption functions in the household and the transport sectors, respectively. IMACRIM-R exchanges the output information with a BU model iteratively for all sectors. Another way to integrate TD and BU models is to deal with detailed energy technologies within the TD model but focus only on a specific sector. Many studies have disaggregated electricity sectors [18], [19] and [20]. However, most such studies have treated only specific sectors and have not expanded to other sectors. Moreover, exchanging information with a BU model does not guarantee a consistent solution with a convergence.3 If the first method can be expanded to other sectors, the results would be quite informative because both the economic and technological sides will be represented consistently. Furthermore, it could potentially improve the representation of reality as well as the reliability of the CGE model analysis. In this context, a CGE model integrating detailed technological information not only for the electricity sector but also for energy end-use sectors is proposed in this study with two objectives: (1) to demonstrate how to integrate detailed BU information within a CGE model and (2) to understand the characteristics of the model behavior. Section 2 presents the model structure for both types of models. In one, energy is represented by a traditional aggregated function, and in the other, energy is represented by detailed BU information. In Section 3, the scenario framework and assumptions are explained necessary to determine the characteristics of the proposed model. To test the model differences, two scenarios with and without climate change mitigation are implemented. Section 4 presents the model results, focusing on how the results differ for the two types of model. In Section 5, the discussions are made on the interpretation and the implications of the results, and limitations of this type of modeling. Finally, concluding remarks are shown in Section 6.

This paper proposes a CGE model that is coupled with detailed energy end-use technologies. Two things are demonstrated (1) the methodology by which energy end-use technologies are treated within the CGE model and (2) the characteristics of the model behavior. Energy service demand and end-use technologies were explicitly differentiated, and the share of technologies was determined by a discrete probabilistic function (the Logit function) to meet the energy service demand. The model was compared with the production function under the same assumption scenarios. The proposed model has several unique characteristics although overall energy supply and GHG emissions impressions of the results are similar to those presented in traditional models. First, macroeconomic losses caused by mitigation are slightly less than those shown in previous models. Second, values of energy price elasticity and AEEI rates derived from the results were heterogeneous across regions and sectors. Third, the ways in which the transport and household sectors reduced CO2 emissions were found to be different from aggregated models. The proposed model showed high levels of energy efficiency improvement and carbon factor reduction, particularly in the household sector as a response to climate mitigation. Although the proposed model within this paper is integrating several methods which have been already seen in earlier studies, applying to global model along with this aggregation level is the first trial in this research field. In that sense, we think that the methodology is original. However, this model still has several limitations so far as discussed previously. Therefore, the future works are expected to direct overcoming these points. The first direction would be improving the method how the share parameter in the Logit function is determined for the future. In this paper, it was simply assumed, but this parameter represents many aspects associated with the consumer behavior and intangible cost, for example externality of the air pollutants caused by combustion of fuels. Thus, quantifying these elements by using historical statistics would be one of the interesting directions. The other one is taking into account infrastructure. The infrastructure is the quite important factor related to the modal shift. If the cost of infrastructure is taken into account along with technological device cost, this model would be able to address appropriately the modal shift which is supposed to be one of the non-technological mitigation measures. Although many issues remain, we believe this model has significant potential to be useful in making policy decisions as compared to previous CGE models. In that context, the model presented in this paper could represent a useful direction in the further development of CGE models.