A Study on the Estimation of the Regression Coefficients of the Panel Regression Model with Autocorrelation

In this study, we considered a one-way error component model with first-order autocorrelation for panel data. We conducted a Monte Carlo study to examine the effects of ignoring individual effects or autocorrelation in this model on the estimation of regression coefficients and hypothesis testing. The Monte Carlo results indicated that even when either individual effect or autocorrelation is ignored the bias of the regression coefficient estimates remains close to zero suggesting no significant issues in estimating the regression coefficients. On the other hand, when one of the effects is ignored, the standard errors of the regression coefficient estimates are underestimated or overestimated. As a result, the hypothesis test for the regression coefficients underestimates or overestimates the nominal significance level, which seriously distorts the hypothesis test for the regression coefficients. Finally, we compared the estimation results of the considered models through a case analysis of the Korean Media Panel data. Finally, a case study using the Korean Media Panel Survey data was conducted to compare the estimation results of the considered models. The estimation results from the one-way error component model (ONEAR) with autocorrelation showed that women, individuals without jobs, and those with lower levels of education spent more time watching TV. Additionally, TV viewing time increased with age.