Does information help recovering structural shocks from past observations?