Modelling Land Use Changes in GTAP

It is agreed by members in the 7th Conference of the Parties (COP7) to the UNFCCC held in November 2001 to take land-based carbon sequestration into account for the 2008-2012 greenhouse gas (GHG) emissions reduction targets. Activities such as afforestation, reforestation, and deforestation (ARD), forest management, crop management, grazing land management, and re-vegetation affect carbon sequestration. This provision has motivated more research efforts to consider land-use changes in integrated assessment (IA) of climate change issues. A need is elicited to count in greenhouse gas (GHG) emissions from all sources and sinks from land-based resources—land use, land use change and forestry (LULUCF) activities are specially focused. In conventional CGE models, land is normally assumed sectoral-specific and is exogenous. To consider the context of LULUCF in the model, it is important to identify functions of land supply to sectors—especially sources of land supply, as emission coefficients vary to different uses of land. In this paper we introduce the GTAPE-L model (Burniaux, 2002), which recognizes sources of land supply via a "land transition matrix". GTAPE-L is based on the GTAP-E model, which extends the standard GTAP model to accommodate substitution between energy and between capital and energy. GTAPE-L is designed to trace inter-sectoral land transitions and to estimate sectoral net emissions due to land use changes. In GTAPE-L, we treat GHG emissions as part of the CES nested production structure, which takes into account complementary inputs to GHG abatement technologies. We calibrate the CES substitution elasticities to fit marginal abatement costs of engineering estimates. The land transition matrix shows changes of land status (or use) over a given period of time, for example, cropland being transformed into forest land (or afforestation). We derive inter-sectoral land use transitions from the estimation of the RIVM IMAGE 2.2 model (IMAGE Team, 2001). We also calculate net emissions associated with land transitions. For each type of land transition, we assign a specific net emission coefficient measuring the difference in emissions per unit of land between the two types of land uses. In the case of transforming cropland into forest, we can calculate the amount of carbon sequestered by multiplying units of land shifted from the agriculture sector to the forestry sector with the net emission coefficient pertaining to the cropland-forest transition. We associate tax instruments with net emissions due to land transitions so that GHG abatement policies could direct LULUCF activities to some conditionally optimized status. We run illustrative simulations of a 30% reduction in GHG emissions of the US and the European Union under the two sets of scenarios: with and without counting in changes in emissions due to land use changes (or transitions). We analyze how the implementation of this abatement target affects the production and prices of the 11 aggregated sectors—rice, crops, livestock, forestry, coal, oil, gas, petroleum products, electricity, chemicals and rest of the economy—in 5 aggregate countries, including the Unite States (US), European Union (EU), Rest of Annex 1 countries, China and India, and Rest of World. We compare the two sets of results to see how the incorporation of land use changes and associated (net) emissions affect the marginal costs of GHG abatement. The results show that land use transitions do help reduce the marginal abatement costs—a 3% reduction for the US, and a 30% for the EU. Such substantial difference between the US and the EU could be explained by the relatively higher carbon intensity in agricultural production of the US and possible under-estimation of carbon sequestration potential as indicated in the net emissions matrix corresponding to land use changes. In addition, we make an alternative version of the GTAPE-L model where sources of land supply are obscured. This is similar to the approach currently taken by the MIT Joint Program on the Science and Policy of Global Change in their EPPA model (Babiker et. al., 2001). In this case, we sum up the sources of land supply of the land transition matrix so that only sectoral total land values are presented. For GHG emissions associated with land based activities, we are not able to recognize differences in emission intensity between activities (or sectors). Only sectoral gross emissions are presented. Comparing the results of the same simulation by the two versions of GTAPE-L, we find that neglecting the sources of land supply and thus net carbon emission/sequestration will lead to mis-measurement of economic costs and sectoral responses. For the US, the economic cost of GHG abatement is relatively higher due to the positive net emission rate associated with land use change. For the EU, the economic cost of GHG abatement is relatively lower due to the negative net emission rate (i.e., sequestration) associated with land use change.