I am estimating an ordinary least squares(OLS) model using the reg
command with the cluster(clustervar) option in Stata 10.1. The problem though
is that the F-model statistic is reported as missing. The help file (help
j_robustsingular) indicates that one reason for this might be because the VCE
is of insufficient rank to preform the model test. However, as the no. of
constraints to be tested (229) is far less than the no. of clusters(9576),
this is unlikely to be the source of the problem. Any suggestions then as to
the cause of the problem?