Dear statlists,
I am new and have to submit my master thesis soon.
I using a micro survey for 10 countries and two periods (independently drawn)
Sample size for each country is around 2000.
Population projection weights to ensure a national representative sample are given for each obsevation.
Explanatories are unavoidable co-linear.
My regression specification is:
reg y x i.country_n i.year [pweight=wgt], vce(cluster country_n)
With that specification I get robust standard errors.
Question 1:
Running the Breusch-Pagan / Cook-Weisberg-Test I get large heteroscedasticity. Is this normal?
Question 2:
Can I call the specification Pooled Weighted Least squares? Or is it just, pooled ols with robust standard errors?
Tons of thanks,
Michael
MSc. in Economics (soon
University of Heidelberg
I am new and have to submit my master thesis soon.
I using a micro survey for 10 countries and two periods (independently drawn)
Sample size for each country is around 2000.
Population projection weights to ensure a national representative sample are given for each obsevation.
Explanatories are unavoidable co-linear.
My regression specification is:
reg y x i.country_n i.year [pweight=wgt], vce(cluster country_n)
With that specification I get robust standard errors.
Question 1:
Running the Breusch-Pagan / Cook-Weisberg-Test I get large heteroscedasticity. Is this normal?
Question 2:
Can I call the specification Pooled Weighted Least squares? Or is it just, pooled ols with robust standard errors?
Tons of thanks,
Michael
MSc. in Economics (soon
University of Heidelberg
Comment