We have been using nested logits and multinomial logits (MNL) to model land use and land use change for modeling greenhouse gas (GHG) emissions from the agriculture, forestry, and land use (AFOLU) sector. The Vietnamese government makes 10-year plans for agriculture, making targets for land dedicated to various crops and other uses. While the targets are included in the plan, the policy instruments used to meet those targets are less clear. Farmers may be pressured to change their own plans to achieve those goals, but the methods used are not very certain.
The underlying household model on which the MNL land use models are based is a random utility model (RUM). A RUM computes utility for various land uses or crop choices. Because these utilities are known with uncertainty (stochastically), the probability of a given piece of land to be put to a given use is computed by comparing the utility of that use to the utility of each of the other choices.
The easiest utility to think about is profit per hectare from cultivating different crops. To effect a reduction in area devoted to one crop, a market-solution would be to raise the cost of cultivating that crop (or equivalently, reduce the profitability of the crop, perhaps by reducing the price). Non-market methods of influencing that decision are difficult or impossible to model directly, but within a model, they might be thought of as a shadow price that could be estimated. If the shadow price is a per unit cost, it can be modeled as a constant. If the shadow price is a per unit of output cost, we can model it as a multiplier on output price.