Endogenous agricultural land supply: estimation and implementation in the GTAP model

Agricultural economy models

Often the mobility of land is only partly or not taken into consideration at all in economic models. Moreover, when land mobility and availability are considered, land is usually implemented as a homogenous entity, disregarding marginal lands and changes in productivity due to land degradation, water stress and climate change. In order to capture the heterogeneity of land in economic models detailed biophysical information needs to be included. Since, land use modeling has mainly been the domain of biophysically oriented disciplines for a long time, much information on the heterogeneity of land is available, but not yet to a full extent used within economic models. Here, we present a method to include detailed biophysical information on land within an extended GTAP model. In Van Meijl et al., 2005 the land supply curve was conceptually implemented into the GTAP model. It was derived on theoretical considerations (see Abler, 2003) and calibrated using expert knowledge and FAO land use projections. In this paper we show that detailed biophysical data concerning land use and associated land productivity provide an empirical foundation of this curve which is consistent with the previously proposed conceptual model.

Land supply curve

In this methodology the total agricultural land supply is modeled using a land supply curve which specifies the relation between land supply and a land rental rate in each region. The general idea underlying the land supply curve specification is that the most productive land is first taken into production. However, the potential for bringing additional land into agriculture is limited. If the gap between potentially available agricultural land and land used in the agricultural sector is large, the increase in demand for agricultural land will lead to land conversion to agricultural land and a modest increase in rental rates to compensate for the cost to take this land into production. However, when almost all agricultural land is in use, an increase in demand for agricultural land will mainly lead to high increase of the land rental rates (land becomes scarce). In this case land conversion is difficult to achieve and therefore the elasticity of land supply in respect to land rental rates is low as well. First theoretical considerations are taken from Abler (2003).

Modeling framework IMAGE

In this paper the biophysical data underlying the land supply curve is taken from the modeling framework IMAGE (Integrated Model to Assess the Global Environment; Alcamo et al., 1998). IMAGE takes into account marginal lands and changes in the potential land productivity due to changes in land use and climate change. In the IMAGE model, climate and soil conditions determine the crop productivity on a grid scale of 0.5 by 0.5 degrees, allowing the feedback of heterogeneous information of land productivity to the economic framework. From the IMAGE model, a land productivity curve is obtained describing the potential crop productivity (accumulated for all crops) as a function of the accumulated land area. This productivity curve is translated into a land supply curve under the assumption that the land price is a function of the inverse of the land productivity. These curves are generated for each region considered and are limited by an asymptote describing the total availability of agricultural land. This asymptote is provided by IMAGE and equals the available land per world region minus urban area, protected bio-reserves, ice and tundra and so on. When the current agricultural land use is close to the potential land availability, we estimate the asymptote (simultaneously with other parameters of the land supply curve) using only observations concerning the accumulated land area lower then the currently observed agricultural area. This method was used for the EU-15 countries and Japan.

Implementation into GTAP model

The land supply curve is estimated for 25 countries and regions. The estimated land supply function will be implemented into GTAP model and the simulation experiments will be made to compare results obtained for the standard and extended model. We will conduct a systematic sensitivity analysis (SSA) to test the robustness of our results with regard to the assumed potential availability of land. First findings indicate that the estimated land supply functions lead to different regional behaviors. For instance, the current position of Africa on their land supply curve indicates that agricultural land can still be expanded without a high increase in the rental land price in this region. Small expansions in agricultural land in the EU-15 however will lead to a high increase in real land prices, therefore stimulating intensification processes in agricultural practices.