topic Re: Panel regression vs. pooled OLS in SAS Forecasting and Econometrics

Re: Panel regression vs. pooled OLS

raqthesolid — Sat, 07 Oct 2017 06:03:01 GMT

I am trying to run proc panel model but having following issues;

proc sort data=j; by gvkey datadate; run;
proc panel data=j ;
id GVKEY DATADATE;
model invest = TQ cfcon /fixtwo pooled;
run;

ERROR: Data set WORK.J is not sorted in ascending sequence with respect to time series ID. The
current time period has year=2015 and the previous time period has year=2015 in cross
section GVKEY=006506.

2)

proc sort data=j; by datadate; run;
proc panel data=j ;
id GVKEY DATADATE;
model invest = TQ cfcon /fixtwo pooled;
run;

ERROR: There is only one cross section or time series observation. Computations will be
terminated.

any way to correct these issues.

THanks

Re: Panel regression vs. pooled OLS

ChristianMac — Mon, 09 Oct 2017 15:42:29 GMT

Hi Raqthesolid,

Depending on how the data set looks like, there might me a few reasons why you are not getting the answers you are looking for.

Regarding the first issue, could you verify that within the cross section GVKEY=006506 you only have one observation characterized by year=2015?

Concerning the second issue, could determine cross section you have? Essentially how mane distinct values do you have in GVKEY?

Re: Panel regression vs. pooled OLS

raqthesolid — Tue, 10 Oct 2017 06:01:08 GMT

Thanks for the reply. i tried the following code to make sure this thing.

proc sort data=j out=c NODUPKEY; by gvkey year; run;

my first problem has been solved.

Regarding second thing there are about 5000 distinct GVKEY.

i am usning following codes;

proc sort data=j; by gvkey datadate; run;
proc panel data=j ;
id GVKEY DATADATE;
model invest = TQ cfcon /fixtwo pooled;
run;

now i am getting this issue;

ERROR: There is only one cross section or time series observation. Computations will be
terminated.

any way to solve this thing. your efforts are always appreciated.

ABdul