Abstract (EN)

This article develops an agent-based model of security market pricing process, capable to capture main stylised facts. It features collective market pricing mechanisms based upon evolving heterogenous expectations that incorporate signals of security issuer fundamental performance over time. Distinctive signaling sources on this performance correspond to institutional mechanisms of information diffusion. These sources differ by duration effect (temporary, persistent, and permanent), confidence, and diffusion degree among investors over space and time. Under full and immediate diffusion and balanced reaction by all the investors, the value content of these sources is expected to be consistently and timely integrated by the market price process, implying efficient pricing. By relaxing these quite heroic conditions, we assess the impact of distinctive information sources over market price dynamics, through financial systemic properties such as market price volatility, exuberance and errancy, as well as market liquidity. Our simulation analysis shows that transient information shocks can have permanent effects through mismatching reactions and self-reinforcing feedbacks, involving mispricing in both value and timing relative to the efficient market price series. This mispricing depends on both the information diffusion process and the ongoing information confidence mood among investors over space and time. We illustrate our results through paradigmatic cases of stochastic news, before generalising them to autocorrelated news. Our results are further corroborated by robustness checks over the parameter space and across several market trading mechanisms.