The ipshin (Im-Pesaran-Shin panel unit root test) routine of Bornhorst and
Baum has been corrected for a flaw in its calculations. Using the default
cross-sectional demeaning, the ADF-like regressions run with positive lags
were not augmented by the proper lagged values. Thanks to Gene Liang of QMS
for bringing this to our attention, and providing benchmark values.
The revised version of ipshin is available from SSC via ssc install ipshin.