[Limdep Nlogit List] Clustering standard errors in a panel data set

[Darren Lee]

Hi,

We are using a panel set and we want to control for firm and time
effects using monthly financial data. There is a paper in the Review of
Financial Studies by Petersen (2009) which suggests correcting for
correlation across these dimensions by clustering the standard errors.
We have found a clustering command in Limdep, i.e. $ Cluster=(cluster
variable) but need to simultaneously cluster on both firm and time
effects. Please advise us on how to code clustering across two variables
(both time and firm).

Also - does the cluster command in Limdep provide the same outcome as
the panel data command such as:

REGRESS;Lhs=BIN;Rhs=ONE,PE_RATIO;Panel;Str=YEAR$

However, it appears that the panel data option only allows for one
stratification (i.e. clustering) variable.

Any help appreciated. Thanks in advance for your advice.

Best Regards,

Darren Lee