SURADF: Stata module to conduct series-specific SUR-ADF panel unit root test

suradf implements the series-specific panel unit root test of Breuer, McNown and Wallace (2002, Oxford Bulletin of Economics and Statistics 64(5), 527-546, doi:10.1111/1468-0084.00276). Conventional panel unit root tests - Levin-Lin, Im-Pesaran-Shin, Maddala-Wu, and the multivariate ADF of Sarno and Taylor implemented in your madfuller - test a joint hypothesis. When they reject, the user learns only that at least one series in the panel is stationary, with no indication of how many or which. suradf instead estimates the N augmented Dickey-Fuller equations as a system of seemingly unrelated regressions, with lag orders free to differ across members and errors contemporaneously correlated across them, and tests the unit root null separately for each member. The output names which series are I(0) and which are I(1). Because the null distribution depends on the estimated covariance matrix of the system, no tabulated critical values apply; the paper is explicit that its own values "cannot be applied directly to other data sets". suradf therefore simulates series-specific critical values for the panel at hand, following the procedure in Section V of the paper.