Efficient Probit Estimation With Partially Missing Covariates

A common approach to dealing with missing data is to estimate the model on the common subset of data, by necessity throwing away potentially useful data. The authors derive a new probit type estimator for models with missing covariate data where the dependent variable is binary. For the benchmark case of conditional multinormality they show that their estimator is efficient and provide exact formulae for its asymptotic variance. Simulation results show that their estimator outperforms popular alternatives and is robust to departures from the benchmark case. They illustrate their estimator by examining the portfolio allocation decision of Italian households.