0
$\begingroup$

I have a dataset with 3 variables: 1) "employment growth rate 2017" by country; 2) "employee happiness (1-100) 2017" by country and occupation level.

Hence the variables would be: 1. country 2. employment_growth, 3. occupation_type, 4. employee_happiness.

So I have occupation+country variability for employee_happiness but not for employment growth, is that okay? Where employee happiness is broken down by country and occupation, and employment growth is only broken down by country. All for year 2017

Since there is no time variable (year), Var1 varies only by country, and Var2 varies by country and occupation- would not be a dataset to use in a plm function, with fixed effects?

My intention is to run a FE regression: Var1 ~ Var2, fixed effects(country, occupation)

I have run several PLM regression however with unexpected results, such as all coefficients being zero etc.

I have tried OLS regression with occupation and country dummy variables, PLM regression with specified country and occupation fixed effects but neither of the approaches seem to work.

Would greatly appreciate any insight and assistance.

$\endgroup$

1 Answer 1

0
$\begingroup$

Presumably, your dependent variable is happiness. But you can't use regular regression because your errors are not independent. So, you have to account for that. Two common ways are multilevel models and generalized estimating equations. Both have been discussed here many times and both have entire textbooks devoted to them.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.