keywords dynamic ordered probit; fixed effects; self-assessed health; reporting bias; panel data; unobserved heterogeneity; incidental parameters; bias correction