As our world is becoming increasingly interconnected and globalized, policy modelers and policy makers are faced with complex decisions for challenges like climate change and financial crises. To address such challenges in an efficient and effective way, policy officials require timely and reliable information. Accordingly, the availability of appropriate methods and tools to aggregate scattered pieces of information and stakeholders' expectations is becoming increasingly important. In this paper we focus on the problem of aggregating and interpreting heterogeneous information dispersed among individuals in a timely manner to support public policy making. We propose an approach that involves the creation of belief (Bayesian) networks to model conditional and probabilistic dependencies between policy indices and the use of Information Markets (IMs) for aggregating stakeholder expectations on uncertain policy indices. An example case focusing on predicting Greenhouse Gas emissions shows how our approach can support real life policies and decisions.