You have 15 variables. For each variable, data may be missing for some years. This is what I mean by missing observations. These observations may be missing due to incomplete data reporting, or for some other reason.

Regards 

Dear Rasel

firm-year observations =

[total number of firms(N) multiplied by sample period (t) multiplied by total number of variables(v)] minus number of missing observation.

[N x T x V] - missing observations.

hope this helps. 

Regards

Peterson.

Thank you, Petersen, for your simple formula. This really helps.

One more thing- Do you mean missing variables by missing observations?

Regards,

Md. Ali Rasel.

Yes

You have 15 variables. For each variable, data may be missing for some years. This is what I mean by missing observations. These observations may be missing due to incomplete data reporting, or for some other reason.

Regards 

Thank you again, Peterson. Now it has been super clear to me.

Regards

gthgythyg

what if you have unbalance data, number of firms will not be the same?

My answer is the same as Mr. Peterson (I don't know Mr. Peterson's designation, whether he is a Professor) has given.

Regards.

First check through Pearson correlation coefficient, it will show you the relationship between all the variables. As you are using panel data, check for normality, multicolinearity, heteroskedasticity, auto correlation, than go for multiple regression analysis... check panel effect, if exist than check fixed effect/random effect through hausman test, than run regression as per your model.............you will get the desired result by observing the value of R-square(coefficient of determination or explanatory power of the mode),

F-test (Goodness of fit test) for model, t-test (individual significance test) for every independent variable and also you can observe the effect of independent variables on dependent variable for their respective beta coefficients.