time variable in unbalanced panel dataset with pooled OLS

[Flor Vsw]

I am working with an unbalanced panel data set and running pooled regression with robust SE

xtset ISSUE QUARTER ==> I have 18 bond issues over 4 years, thus 16 quarters

The problem is that my regression coefficients are strange unless I add the variable 'TIME'.
My question is if it is allowed to add this time variable. Since I defined my quarters from 1 till 15 the time variable is actually the same as the quarter variable. It works as follows:

ISSUE	QUARTER	TIME
1	1	1
1	2	2
1	3	3
1	4	4
1	5	5
1	6	6
1	7	7
1	8	8
1	9	9
1	10	10
1	11	11
1	12	12
1	13	13
1	14	14
1	15	15
2	1	1
2	2	2
2	3	3
2	4	4
2	5	5
2	6	6
2	7	7
2	8	8
2	9	9
2	10	10
2	11	11
2	12	12
2	13	13
2	14	14
2	15	15
3	2	2
3	3	3
3	4	4
3	5	5
3	6	6
3	7	7
3	8	8
3	9	9
3	10	10
3	11	11
3	12	12
3	13	13
3	14	14
3	15	15
4	2	2
4	3	3
4	4	4
4	5	5
4	6	6
4	7	7
4	8	8
4	9	9
4	10	10
4	11	11
4	12	12
4	13	13
4	14	14
4	15	15
5	3	3
5	4	4
5	5	5
5	6	6
5	7	7
5	8	8
5	9	9
5	10	10
5	11	11
5	12	12
5	13	13
5	14	14
5	15	15
6	3	3
6	4	4
6	5	5
6	6	6
6	7	7
6	8	8
6	9	9
6	10	10
6	11	11
6	12	12
6	13	13
6	14	14
6	15	15
7	6	6
7	7	7
7	8	8
7	9	9
7	10	10

The strange thing is that my model coefficients significance and signs are strange, unless I add this variable TIME.

Can someone help / explain this?

Thanks!

[Nick Cox]

I don't think we can say much without seeing the regression command you used and the output it produced. In any case, whether time makes sense as a predictor is a matter for your judgment, not arbitrary oracular opinion.

[Flor Vsw]

Now I add the time variable it is named "KWARTAAL"

So the problem was I wanted to add specific time dummies, but it is not possible since they are correlated with my independent variable 'SIZE'. Therefore did not add Time dummies but this other time variable 'Kwartaal" which works as follows in my unbalanced data set:

[Carlo Lazzaro]

Flor:
set aside the main issue for a while, there's something I'm missing out on your approach:
- you have a panel dataset;
- you started with -xtreg, fe-;
- you switched to a pooled OLS (POLS) without clustered standard errors (SEs) on panelid, but accounting for heteroskedasticity only with robustified SEs. This approach does not take into account that your observations are non idependedent, due to their panel data structure. Please note that while robustified/clustered SEs give the same output under -xtreg-, that feature does not hold under -regress-;
- as a closing-out remark, to justify your switching from -xtreg- to POLS you should have checked that the individual effect, jointly taken, are not different from zero.