Efficient Estimation in Dynamic Panel Data Models with Missing Outcomes

Seminari Applied

David Pacini

University of Bristol

6-Nov-2013

Seminar 3 – 14:30

This paper deals with the problem of efficient estimation of the coefficients of a dynamic panel data models when observations on the outcome variable are missing completely at random and the number of individuals is large with respect to the time dimension of the panel. In this context, existing estimators data on individuals observed for at least three consecutive periods. We show that observational units observed for at least three nonconsecutive periods provide information about the coefficients of interest and describe an asymptotically efficient generalized method-of-moments estimator taking such information into account. The properties of the proposed estimator are illustrated in Monte Carlo exercises and in an application to the estimation of the value of a statistical life using data from the Panel Study of Income Dynamics.