Estimation of Random Components and Prediction in One and Two-Way Error Component Regression Models

Since one of the main objectives of panel data analysis is to uncover individual and/or time effects, the estimation of these random components is very important. Estimation of individual and time components will also help in predicting the future values of the dependent variable, which has received some attention in the literature. Following the stochastic frontier literature, our contention is that estimation of these random components is akin to the estimation of “firm-specific” efficiency. Thus, considering the conditional distributions of the random components and using the conditional mean and variance, we provide both point and interval estimates of the individual and time effects in the one and two-way error component models. Using these standard errors one can also test the significance of random components. Equations for predictions are also provided for these models. Finally, all our theoretical results are illustrated with an empirical application.