Error: Too many base levels specified?

[Sammiah Javid]

I am trying to compare coefficients before and after the crisis. I am using the SUEST command so I can compare the coefficients of the variables across these two models.

I start running the command, I type:

reg tobinsq L1.(totalforeignnonexovertotalboard) L1.(edshareownership ) L1.( logedcompwithoutshares ) L1.(ceoshareownership)  L1.( logceocompwithoutshares ) L1.( nedshareownership) L1.(logtotalcash) L1.(tenure)  L1.(gender ) L1.(logage) L1.(ceoduality) L1.(boardindependencene) L1.(logofboardsize) L1.(logta) L1.(leverage) L1.(stockreturns) L1.(pricevolatility) i.year i.industry if (crisis==1)

estimates store crisisone

reg tobinsq L1.(totalforeignnonexovertotalboard) L1.(edshareownership ) L1.( logedcompwithoutshares ) L1.(ceoshareownership)  L1.( logceocompwithoutshares ) L1.( nedshareownership) L1.(logtotalcash) L1.(tenure)  L1.(gender ) L1.(logage) L1.(ceoduality) L1.(boardindependencene) L1.(logofboardsize) L1.(logta) L1.(leverage) L1.(stockreturns) L1.(pricevolatility) i.year i.industry if (crisis==3)

estimates store crisisthree

suest crisisone crisisthree, vce(robust)

test [crisisone_mean] totalforeignnonexovertotalboard - [crisisthree_mean] totalforeignnonexovertotalboard =0

When stata gets to the

 suest crisisone crisisthree, vce(robust)

command, I am getting the error code

too many base levels specified r(198);

What does this mean? And, is there anything I can do about it?

[William Lisowski]

Welcome to Statalist.

Without seeing the results of your regress commands it's difficult to know, but my guess is that in your two regressions, different categories were used as the base levels for year and/or industry in the two regressions.

I suggest using

tabulate year crisis
tabulate industry crisis

to find a value of year and a value of industry that is represented for both crisis==1 and crisis==3, and then use the factor variable notation needed (see help factor variables) to select these values as the base levels in both regressions.