مجتمع انرژی زیستی در مدل های تعادل عمومی قابل محاسبه - بررسی

In the past years biofuels have received increased attention since they were believed to contribute to rural development, energy security and to fight global warming. It became clear, though, that bioenergy cannot be evaluated independently of the rest of the economy and that national and international feedback effects are important. Computable general equilibrium (CGE) models have been widely employed in order to study the effects of international climate policies. The main characteristic of these models is their encompassing scope: Global models cover the whole world economy disaggregated into regions and countries as well as diverse sectors of economic activity. Such a modelling framework unveils direct and indirect feedback effects of certain policies or shocks across sectors and countries. CGE models are thus well suited for the study of bioenergy/biofuel policies. One can currently find various approaches in the literature of incorporating bioenergy into a CGE framework. This paper gives an overview of existing approaches, critically assesses their respective power and discusses the advantages of CGE models compared to partial equilibrium models. Grouping different approaches into categories and highlighting their advantages and disadvantages is important for giving a structure to this rather recent and rapidly growing research area and to provide a guidepost for future work.

In the context of energy security and climate protection, bioenergy is ascribed high importance. Especially biofuels have received increased attention since they are able to replace fossil energy in the transport sector. As the transport sector is contributing an increasing share to global carbon emissions they are seen as an important mitigation option, also because other renewable energy sources usually only replace fossil fuels in the electricity (wind, hydro, photovoltaics) or in the heat sector (wood pellets, geothermal energy, solar thermal energy). Currently, only Brazil is able to produce bioethanol from sugar cane at sufficiently low costs to be competitive with conventional fuels. Nevertheless and for reasons just explained, bioenergy and biofuels are part of climate and energy policy packages in several countries and supported by quotas, tax exemptions or direct production subsidies. This has resulted in growing production and consumption of biofuels worldwide. Plans to further increase the use of bioenergy are on the table. In Europe, the directive on the promotion of the use of energy from renewable sources (henceforth RES directive) targets a 20% share of renewables — including bioenergy — in total energy use in 2020 and additionally imposes a 10% minimum share of renewable energy in transport (cf. European Union, 2009). Without the widespread availability of alternative renewable transport fuels, these will primarily be biofuels. A crucial element of the RES directive is that both domestically produced and imported biofuels need to meet sustainability criteria. Furthermore, the binding character of the 10% quota is subject to biofuels being produced sustainably and second-generation biofuels becoming commercially available. In 2007, the EU share of biofuels in total fuel consumption was 2.6% corresponding to an estimated combined EU ethanol and biodiesel consumption of 9.9 billion litres. Thus big efforts have to be undertaken in order to reach the 10% target in 2020, which would approximately correspond to a biofuel use of 22.8 billion litres as projected by the European Commission.1 The 2007 US Energy Independence and Security Act (EISA) stipulates that by 2022, 36 billion gallons (ca. 136 billion litres) out of total transportation fuels used shall be biofuels. Out of that, 15 billion shall be conventional biofuels, which will mostly be corn-based ethanol. This requires a substantial increase from a 2008 ethanol production of 9 billion gallons. The remaining 21 billion shall be advanced biofuels, including biodiesel as well as cellulosic fuels.2 US ethanol production has even overtaken Brazilian ethanol production recently. Other countries also pursue their own policies in promoting the use and production of biofuels, among them China and India.3 All these developments indicate a strong rise in biofuel production over the next years. The governmental support for bioenergy has been heavily criticized especially in the context of rapidly rising food prices in 2007/2008. A heated ‘food vs. fuel’ debate has emerged that reflects the fear that enhanced biofuel production may lead to enormous land use competition that would drive up agricultural product and food prices. It is therefore vital to get a better understanding of the economy-wide impacts of enhanced bioenergy production and especially its impact on land use competition and on agricultural and ultimately food prices. There are thus two essential dimensions that the study of bioenergy has to take into account: (1) Biofuels should be studied from an international perspective given worldwide biofuel support policies and the likely reliance on imports for fulfilling mandatory biofuel quotas. (2) One has to analyse economy-wide effects as suggested by the impacts of biofuel production on the agricultural and food sectors but also on other sectors of the economy, e.g. on the energy sector to name only one. Computable general equilibrium (CGE) models have been widely employed in order to study the effects of international climate policies (see e.g. Springer, 2003). The main characteristic of these models is their encompassing scope: Global models cover the whole world economy disaggregated into regions and countries as well as diverse sectors of economic activity. Such a modelling framework unveils direct and indirect feedback effects of certain policies or shocks across sectors and countries. CGE models are thus well suited for the study of bioenergy policies. This paper focuses on different approaches to include bioenergy into CGE models, the advantages and disadvantages of these approaches and implications for future modelling work. In this course we also try to compare and explain major results of different models. Included in this paper are all major multi-region CGE models we are aware of that include bioenergy. These are in particular the models USAGE (Dixon et al., 2007), a GTAP-E version modified at LEI Institute (Banse et al., 2008), WorldScan (Boeters et al., 2008), DART (Kretschmer et al, 2008), EPPA (Reilly and Paltsev, 2007, Gurgel et al., 2007 and Melillo et al., 2009) and augmented versions of GTAP (Birur et al., 2008, Hertel et al., 2008 and Keeney and Hertel, 2009). Besides CGE models there are of course other types of models suitable for studying bioenergy. Indeed, many studies to-date have used partial equilibrium (PE) models (see Gerber et al., 2008) that mostly focus on the agricultural sector. It is beyond the scope of this paper and also not the aim to provide a detailed survey of modelling bioenergy in PE models. A report that goes into this direction and includes a number of PE models is by Pérez Domínguez and Müller (2008) while Gerber et al. (2008) compare the results of different models with respect to effects of biofuel policies on food prices. Yet, to make the strengths and weaknesses and the limitations of the CGE approach more transparent we also include a section on PE models and their analysis of bioenergy policies. The data base of CGE models are so-called social accounting matrices (SAMs). A SAM is a balanced matrix that summarizes all economic transactions taking place between different actors of the economy in a given period, e.g. one year. Economic transactions are represented in value terms and the SAM is balanced in the sense that the value of, for instance, a production sector's output equals the value of its inputs, although SAMs can be much more detailed than that including taxes, subsidies, transfer payments etc. It is assumed that a SAM for a certain year represents an equilibrium of the economy and the model is calibrated in such a way that the SAM is a result of the optimizing behaviour of firms and consumers in the model. The Global Trade Analysis Project (GTAP) provides every few years new consistent international SAMs that are used by basically all global CGE models. The most recent data base GTAP7 was published in October 2008 (Narayanan & Walmsley, 2008) and is based on input–output and trade data for the year 2004. Many models still run on GTAP6 with 2001 as the base year (Dimaranan, 2006). The problem is that there was only little production of bioenergy until recently and that the SAMs used for the calibration of existing models thus give little information on the production and trade patterns of bioenergy that begin to emerge today. Fig. 1 shows the development of biofuel production in the major producing countries and nicely illustrates the fact that biofuel production in the USA and the EU only really took off after 2001. Brazil on the other hand already had an important ethanol industry much earlier. In addition, even if some production and trade existed in a certain base year, it is not shown explicitly in the SAMs, but aggregated e.g. to total fuel use. Furthermore, current bioenergy production is mainly the result of a variety of different governmental support measures that are neither — at least not explicitly — included in the SAMs yet. Future production and trade patterns are likely to look very different from today's patterns and depend crucially on policy assumptions. Thus, there is a general lack of consistent production and trade data for bioenergy and biofuels. Full-size image (23 K) Fig. 1. Historical production of ethanol in USA and Brazil and of total biofuels in EU27. Sources: RFA, 2009, UNICA, 2009 and Biofuels Platform, 2009. Figure options Hence, the general challenge in modelling bioenergy is that, on the one side, bioenergy is not a production sector that is included in the base year SAMs of CGE models, so that it cannot be calibrated in the usual way. On the other side, it is also not a pure future technology but one has to account for the production and trade patterns that exist today as a result of governmental support. One can currently find various approaches in the literature to overcome these difficulties and to incorporate bioenergy into a CGE framework. This paper intends to give an overview of existing approaches and to critically assess their respective power. Grouping different approaches into categories and highlighting their advantages and disadvantages is important for giving a structure to this rather recent and rapidly growing research area and to provide a guidepost for future work. The paper is organized as follows: We first compare the advantages and disadvantages of general and partial equilibrium models in Section 2 and discuss general modelling issues in the context of biofuels in Section 3. Section 4 describes the first type of modelling approach that we distinguish, the implicit approach, which is a rather ad-hoc approach that avoids an explicit modelling of bioenergy production technologies but instead prescribes the amount of biomass necessary for achieving a certain production level. Section 5 deals with a second category of models that include biofuel production with the help of so-called latent technologies. These are production technologies that are existent but not active in the base year of the model and that can become active at a later stage or in counterfactual scenarios. Section 6 outlines the third approach that intends to actually disaggregate bioenergy production sectors directly from a social accounting matrix (SAM), the underlying data structure of CGE models. Section 7 summarizes results with respect to agricultural price effects across studies and Section 8 concludes.

The intention of this paper was to highlight various techniques of introducing bioenergy technologies into CGE modelling frameworks. We classified the various approaches into three broad categories, each characterized by its particular advantages and disadvantages as summarized in Table 2. Table 2. Three approaches of modelling bioenergy. Approach Advantages Disadvantages Studies Implicit approach Elegant approach avoiding a breaking up of the original model structure No explicit bioenergy production sector Dixon, Osborne and Rimmer (2007) → No commodity “biofuel” Banse et al. (2008) → Trade in biofuels cannot be modelled Latent technologies More realistic representation of bioenergy production processes by including separate sectors Projections based on limited time series of biofuel production and trade data or even on pure assumptions Boeters et al. (2008) Allows for including trade in biofuels Complex procedure, increase in computational burden Kretschmer et al. (2008) Allows for including new developments (second generation biofuels; new producing countries) Reilly and Paltsev (2007) Gurgel et al. (2007) Melillo et al. (2009) Disaggregating the SAM Ex-ante inclusion of bioenergy technologies in underlying database Full potential is so far still restricted by data limitations Birur, Hertel and Tyner (2008) Coherence of modelling framework Limitations to model new developments Hertel, Tyner and Birur (2008) Keeney and Hertel (2009) Taheripour et al., 2008 and Taheripour et al., 2009 Table options Theoretically, the most promising approach is to calibrate the model to a SAM that disaggregates bioenergy activities in separate sectors. If the data for this approach were already easily available there would be little need for using the other approaches in the context of first-generation biofuels. However, up till now the accuracy of the SAM approach is limited by insufficient data for the model base year (2001) and the fact that in this year there was still little biofuel production and trade. The GTAP-BIO database which marks the first attempt to disaggregate biofuels in the SAM suffers from these weaknesses and is based on a number of more or less realistic assumptions (e.g. that sugarcane-based ethanol is disaggregated from the ‘chemical, rubber, plastic’ sector). The database GTAP7 published in late 2008 is calibrated to the base year 2004 thus rendering the bioenergy data scarceness for first-generation technologies somewhat less problematic due to their rapidly growing importance over the last years. To disaggregate the SAMs correctly one would need to have detailed information on where biofuel production is included in the national SAMs, which inputs it uses and how biofuel is traded. Gathering these data is probably still not easy and very time consuming. For the research community it will be of great help when the GTAP-BIO database is updated for the 2004 dataset GTAP7, which is obviously underway. In any case, it will be likely that we will soon see further approaches of incorporating bioenergy directly into SAMs and that many models will switch to using these SAMs to model first-generation biofuels. At the same time, we believe that latent technologies will continue to play an important role in the future. Bioenergy production is developing quickly and will expand to regions that are currently not producing on a commercially relevant scale yet. These include Malaysia and Indonesia that are expected to become important biodiesel producers (and possibly also exporters), as well as countries in Latin America and Africa. Furthermore, second-generation biofuels will in the medium to long term play an important role as well. The approach of modelling bioenergy with the help of latent technologies is flexible to account for such new developments and also for modelling long-term scenarios. The downside of this approach is that it is based more or less on mere speculations about cost developments and availability of advanced technologies. The first approach discussed above of implicitly modelling bioenergy for example by assuming a certain input share of feedstocks into refined oil production is in our opinion rather an intermediate step towards more advanced bioenergy modelling. The results that have been cited in this paper highlight the need for further modelling efforts. It has been seen that models that work with different assumptions come to partly greatly diverging results, showing that the assumptions used today need to be constantly checked for their future validity. Part of the problems associated will disappear or at least become less severe over time with the gathering of more reliable data on biofuel production and trade. In addition there are numerous issues for future research. The most important issue not only in the context of bioenergy is modelling land availability and land restrictions in a more sophisticated way. Here, models still experiment with different approaches. Most of these approaches are also used in the models with bioenergy that are described in this paper. The most advanced CGE model in this respect is the adjusted GTAP model that distinguishes between 18 different agro-ecological zones (AEZs) which restrict the possibility of land to move between uses. In order to limit the complexity of single models and to combine advantages of different modelling designs a promising way approach is couple models such as e.g. done in Melillo et al. (2009). Coupling a detailed ecosystem model to the CGE model EPPA allows calculating emission from direct and indirect land use changes, which is a high-priority topic for scientists and policy makers since these emissions determine the overall GHG savings that can be achieved with the help of biofuels. What is still missing is a CGE model that truly endogenizes the calculation of land use emissions so that it is possible to calculate the role of biofuels in an optimal policy mix accounting for direct and indirect emissions. This is clearly an important task for future research. Besides this major task there are a few smaller issues that need to be considered in future CGE analysis of biofuel policies. One is the so-called ‘blending wall’. Conventional engines can only safely take up low-percentage blends of biofuels of up to 10% or even lower, with the precise figure being debated and depending on the exact engine type. Only flex-fuel vehicles can take up higher blends, but these have not penetrated the majority of car markets yet. Their widespread availability would, however, be a prerequisite for imposing targets beyond 10%. One could possibly account for this by introducing transaction costs (representing the costs of replacing parts of the car fleet with flex-fuel vehicles) associated with reaching blends beyond 10%. For modelling current European biofuel policy, this is not an issue. Even if biofuel production in some countries exceeded 10%, the excess production could be thought of as being exported to other countries in order to reach a 10% share of biofuel use there. This brings us directly to the issue of trade in biofuels — which is only partially modelled in the existing models even though this has a major impact on the effects of certain biofuel targets. As trade patterns begin to emerge more clearly, models should be improved to reflect these patterns. A further issue is that so far most models focus on bioethanol and biodiesel only and neglect other possibilities to generate bioenergy such as bioelectricity or direct burning of biomass to generate heat. This is especially important since these bioenergy options are preferable to biofuels in terms of emission reduction costs and potential and are likely to play an important role in a cost-effective policy mix. This as well as a more widespread modelling of by-products should also be on the agendas. Hopefully, this paper will contribute to a better understanding of modelling bioenergy in CGE models and of the results of different studies and will help to improve and extend the models to better capture relevant driving forces of bioenergy and the determinants of their economic and environmental impacts.