problem in panel data analysis: Not enough observations with non-missing model variables for model statement in cross section

[comeon2012]

Hi all,

When I did panel data analysis like below:

proc sort data=al_us_l_r_l_res_sdm_bpres_sa_f;

by ds_code yrmth;

run;

proc panel data=al_us_l_r_l_res_sdm_bpres_sa_f;

id ds_code yrmth;

model wr = wrds_1 wwcor_kz_1 wwcor_kz_1wrds_1 rolling_stdm wwroa_1_d wlnmv_1 lnbm_1 wmomentum /FIXONE;

run;

there is error message like below:

ERROR: Not enough observations with non-missing model variables for model statement  in cross section DS_Code=130286.

NOTE: The SAS System stopped processing this step because of errors.

As I checked the dataset, there was no non-missing values for DS_Code=130286 (see the attached file). Is this the reason that caused the error? Do I need to exclude the observations for DS_Code=130286 to proceed the panel data analysis? If yes, since there are some many identities (DS_Code) in my original dataset, is there any easy way to exclude the observations for these identities (e.g., if there is no non-missing valuse for any variable of an identity, then exclude these identities in the panel analysis)? Will excluding these identities resolve my problem?

Any advice will be appreciated.

[SteveDenham]

The first 177 records have nothing but missing values for the right-hand side variables.  While that may not be the whole of the problem, it is a good place to start.  Also rolling_stdm is missing for every observation.  The latter is probably the main source of your problem.  Restate your model, excluding this variable and see what happens.

Steve Denham

[comeon2012 · Accepted solution]

Thanks SteveDenham.

rolling_stdm is not the only variable that has missing values for every observation. Some other variables also might only have missing values for a certain identity (cross section). So I just delete all the observations that has missing value even only for a variable. By doing this, my problem seems to be resolved.