Female babies and risk aversion: causal evidence from hospital wards