Converting Unbalanced panel data in Balanced
Posted: Mon Dec 12, 2016 10:03 am
Dear all,
i'm working on a panel dataset (including 11 variables, 16 countries from 1995 to 2013) for my master thesis, and i'm having some trouble with it.
Before estimating the regression for finding if and how fiscal variables affect the long run growth rate in an endogenous growth model, i am considering five different forms of panel data estimator: Pooled OLS, One-way (country dummies) fixed effects model, One-way random effect models, two-way (country and time effects) fixed effects model and two-way random effects model. In order to remove the effects of the business cycle, i proceed by taking a 5 year average. So EViews recognize it as Dated Panel with a frequency: multi-year, 5 year with country as Cross section ID series and year as Date series. I'm working with EViews 9 SV (student version). Please find in attachment my dataset.
The problem is, when i try estimate the two-way random effects model, i'll get the error saying "Two-way random effects not allowed with unbalanced data". This is the first time i'm working with panel data, and i'm not very familiar with it, but it seems that the problem is created by the N.A. values, which unfortunately i have no clue how to estimate, or forecast being at the beginning of the series.
So my question is, how do i convert the unbalanced panel data into balanced?
p.s. i manually tried to fill the n.a. values but in that case for the two-way random effects model i'll still get an error saying: "Random effects estimation requires number of cross sections>number of coefs for between estimator."
Thank you in advanced,
Klea
i'm working on a panel dataset (including 11 variables, 16 countries from 1995 to 2013) for my master thesis, and i'm having some trouble with it.
Before estimating the regression for finding if and how fiscal variables affect the long run growth rate in an endogenous growth model, i am considering five different forms of panel data estimator: Pooled OLS, One-way (country dummies) fixed effects model, One-way random effect models, two-way (country and time effects) fixed effects model and two-way random effects model. In order to remove the effects of the business cycle, i proceed by taking a 5 year average. So EViews recognize it as Dated Panel with a frequency: multi-year, 5 year with country as Cross section ID series and year as Date series. I'm working with EViews 9 SV (student version). Please find in attachment my dataset.
The problem is, when i try estimate the two-way random effects model, i'll get the error saying "Two-way random effects not allowed with unbalanced data". This is the first time i'm working with panel data, and i'm not very familiar with it, but it seems that the problem is created by the N.A. values, which unfortunately i have no clue how to estimate, or forecast being at the beginning of the series.
So my question is, how do i convert the unbalanced panel data into balanced?
p.s. i manually tried to fill the n.a. values but in that case for the two-way random effects model i'll still get an error saying: "Random effects estimation requires number of cross sections>number of coefs for between estimator."
Thank you in advanced,
Klea