We analyse the problem of parameter inconsistency in panel data econometrics due to the correlation of exogenous variables with the error term. A common solution in this setting is to use Instrumental-Variable (IV) estimation in the spirit of Hausman-Taylor (1981). However, some potential shortcomings of the latter approach recently gave rise to the use of non-IV two-step estimators. Given their growing number of empirical applications, we aim to systematically compare the performance of IV and non-IV approaches in the presence of time-fixed variables and right hand side endogeneity using Monte Carlo simulations, where we explicitly control for the problem of IV selection in the Hausman-Taylor case. The simulation results show that the Hausman- Taylor model with perfect-knowledge about the underlying data structure (instrument orthogonality) has on average the smallest bias. However, compared to the empirically relevant specification with imperfect-knowledge and instruments chosen by statistical criteria, the non-IV rival performs equally well or even better especially in terms of estimating variable coefficients for time- fixed regressors. Moreover, the non-IV method tends to have a smaller root mean square error (rmse) than both Hausman-Taylor models with perfect and imperfect knowledge about the underlying correlation between r.h.s variables and residual term. This indicates that it is generally more efficient. The results are roughly robust for various combinations in the time and cross-section dimension of the data.
Endogeneity instrumental variables two-step estimators Monte Carlo simulations