A New Representation of Agricultural Production Technology in GTAP

Objectives This paper seeks to improve the treatment of the agriculture sector in GTAP. It will incorporate into GTAP an agricultural production technology consistent to the maximum degree practicable with that currently employed in the Organization for Economic Co-operation and Development¿s (OECD) Policy Evaluation Matrix (PEM) model. Policy reform simulations in a sector as complex as agriculture often require a variety of different models. Consistency between the different approaches is important. The assumptions about economic behavior should be transparent and similar in spirit across the different models. The focus of this paper will be on the way OECD agricultural policies are represented in GTAP. This is important to preserve consistency relative to existing PEM analysis, in particular with respect to the issue of decoupling. Approach The most important concern is the modeling of land allocation and the influence of policy on those decisions. The base version of GTAP represents land allocation in a nested constant elasticity of transformation (CET) structure similar to, but not as detailed as in PEM. PEM distinguishes different types of land and ¿this allows for detailed representation of land-based policies. In this paper, GTAP is modified to include a three level CET land use structure. The differentiation between different land types according to the crop being farmed and demand for pasture land also implies a supply response that may not always be in line with the supply response generated by a standard GTAP model. The newly developed land allocation system is calibrated to land supply elasticities from the same literature survey (OECD 2001) used to calibrate the PEM model. In the standard GTAP model there is no specificity for agricultural capital or labor. This assumption is defensible for analyses where the assumed adjustment horizon is sufficiently long term. However, for the short/medium term adjustment horizon usually adopted for OECD simulations, it is necessary to modify GTAP in order to recognize that these factors are to some degree sector specific and adjustments are therefore ¿sluggish¿. A common assumption in CGE models, including GTAP, is that intermediate inputs are combined with primary factors in fixed proportions. In PEM, all purchased inputs are assumed substitutable and used in variable proportions with primary factors depending on relative prices. Hertel and Keeney (2003) have modified GTAP to represent this kind of factor substitution, incorporating PEM substitution elasticities in the model. This approach will be adopted for this paper. Preliminary Findings This study should result in a more realistic modeling of the agricultural sector in GTAP. In particular, this study should find that, based on the literature survey, agricultural supply response is more inelastic for specific land using activities than in the standard GTAP model. For example, land used in rice is much less substitutable than land used in wheat production. For an equivalent policy shock, supply response between crops will be more inelastic, and changes in crop prices will be more pronounced. In summary, aggregate supply response for land using sectors (principally agriculture) will be more inelastic (more sector specificity for land use), and rental rate differences will be more pronounced (with implications for welfare distribution).