It is becoming increasingly clear that climate change will cause substantial disruption to the socio-economic systems around the world. Climate mitigation and adaptation measures are vital and require careful analysis of economic costs and benefits. While a possibility of crossing thresholds and triggering abrupt irreversible changes in climate-economy system is foreseen, the development of models to study their emergence and effects is challenging. This is especially relevant for economic models, which are designed to study marginal changes only. New information about climate-induced risks, and behavioral changes amplified by social interactions affect economic choices as well as associated potential emissions and damages. In this presentation I will discuss how agent-based computational economics can be useful in exploring the emergence of non-marginal changes in coupled socio-environmental systems.