Further Monte Carlo Evidence on Seemingly Unrelated Regressions with Unequal Number of Observations

Schmidt's [1977] results on seemingly unrelated regressions with unequal number of observations are replicated. These results are shown to be robust to the type of addtional observations available i. e., whether they are time series or cross-sectional in nature. An important finding is that the extra observations may lead to better estimates of the variance-covariance matrix ? or its inverse, but this does not necessarily lead to better estimates of the regression coefficients.