Nitrogen (N) pollution has mostly been controlled using command-and-control instruments. However, nitrogen surplus permits (NSPs), which are tradeable, can be more cost-efficient in addressing the problem. To model this instrument, we calculated the individual marginal abatement cost curve for a sample of about 3,400 Swiss farms using farm-optimization models implemented in the agent-based agricultural sector model SWISSland. We also used SWISSland to analyze the effects of two NSP distribution systems (grandfathering and land-based allocation) on different farm types. The results showed that different farm types range in their abatement costs to reduce N surplus from an average of −0.04 CHF kg−1 N on arable farms to 51.06 CHF kg−1 N on special crop farms. We also found that N surpluses hardly explain the level of abatement costs. The biggest differences in effects of the distribution scheme were found in intensive livestock farm types such as pig or poultry farms.
Linking computable general or partial equilibrium models with agent-based models can combine the strengths of both modeling concepts. The linking allows to trace global feedback effects while providing a flexible representation of human behavior and social interactions. This linking would facilitate the simulation of interconnected land-use and economic systems while accounting for actors’ heterogeneity, their specializations, and the representation of alternative decision models. In this paper, we examine the challenges that currently hinder the realization of these potential benefits and present a roadmap outlining possible solutions for successful model linking in the context of land use change. The main challenges include: 1) conceptual misalignment between modeling concepts, 2) differences in scales and resolution, 3) difficulties in validation and calibration of linked models, 4) increased complexity in interpreting and communicating results, 5) high demands on compu-tational infrastructure and computational costs, and 6) limited personnel and financial resources. Successfully linking different model concepts and overcoming these challenges requires the modelling communities and stakeholders to engage in long-term, platform-supported dialogue. This dialogue could facilitate the development of a shared framework for model linking, standards for model documentation, validation procedures, and appropriate approaches for communicating and evaluating results. This process will enhance the benefits and sustain the linked models while helping determine circumstances in which the advantages of linking models may outweigh the costs.