How to Account for Contemporaneous and Time Dependence in Business Research?

In the analysis of balanced panel data sets used in Business Research, the commonly used methods are either OLS with adjusted standard errors such as cluster standard errors or Fama-Macbeth. In this paper we show through extensive simulations that Maximum Likelihood Generalized Least Squares (ML-GLS) method outperforms the aforementioned methods. We also present the results from the analysis of a Corporate Finance panel data-set in this context.