Simulating economic impacts of droughts on German agriculture using the country-scale Multi-Agent System model DroughtMAS

Given widespread and serious implications of increasingly intense and frequent extreme events such as droughts and heat waves, there is a growing demand for integrated approaches capable of supporting prospective risk reduction through better adaption to a changing climate. While anticipating future impacts is essential to inform policy decisions, there is simultaneous need for projections at higher spatial resolution to forecast the local effects of global change. Here, we present a spatial multi-agent system (MAS) model, DroughtMAS, calibrated using a positive mathematical programming (PMP) approach. The model simulates land-use adaptation to future drought conditions, estimates the economic damages of future droughts, and assesses policy measures aimed at enhancing the drought resilience of German agriculture. It represents the biophysical and agro-economic heterogeneity of German agriculture through 23,396 individually parameterized land-user agents located on a country-wide 4x4km grid. Cropping behavior is calibrated with land-use data from high-resolution remote sensing analyses and public records and validated with independent land-use data. The economic parameters ground the model to a policy-relevant context while the statistical yield functions capture the impacts of biophysical factors on crop production. These yield functions enable the model to respond to soil moisture changes from observed data or projections from hydrological models. DroughtMAS extends the classical PMP model to capture short-run responses to droughts more realistically and analyze how fast the farmers move towards the desirable equilibrium conditions under recurring droughts. The model shows that farmers gradually adapt to prolonged drought conditions, with a lower degree of adaptation in the first drought years only slightly mitigating drought impacts. We present first analysis of future drought scenarios to demonstrate the ability of the model to quantify risks from potential droughts across Germany in monetary terms. The results provide bottom-up estimates of economic damages of droughts accounting for much needed short-run behavioral dynamics of adaptation. This provides valuable and realistic projections of future drought impacts of farm-specific changes aggregated at national scale. The model also presents spatiotemporal pattern of these impacts, showing the potential for such projections to inform targeted policy interventions. DroughtMAS provides a platform that can be extended to capture additional adaptation behaviors (e.g. drought-resilient crops, adapted crop calendars, irrigation systems) and combined with other models that require empirically validated inputs about various agricultural decision-making conditions.