Return predictability and social dynamics

The ability to predict stock returns from financial ratios is a long-standing but still controversial topic. There is ongoing debate about the empirical evidence as well as about appropriate theoretical explanations. We provide evidence from a simulated economy that local, social interaction among agents is remarkably successful in matching several established empirical facts. We find significant return predictability at various forecast horizons, absence of dividend growth predictability, high persistence in dividend yields, and absence of significant return autocorrelations. Our results suggest that social dynamics are a simple, intuitively appealing and successful way to explain predictability.