On the poverty of theory: natural hazards research in context