I am running a fixed effects regression on wages and gambling to see what impact gambling has on earnings.
The dataset I have been given has two interesting variables - poverty and unemployment rate. Poverty is a dummy which indicates whether the individual's family was below the poverty line last year. Does this variable make sense in a fixed effects regression of wage?
The unemployment rate variable shows the unemployment rate for the local labour market of the individual. However, the dataset I have is restricted and so instead of showing the unemployment rate of the particular state the individual is in, it shows the average unemployment rate of the wider region (north, east, south and west of the country).
Because of this, I was wondering whether this is a variable that makes sense to use. It comes up as highly significant in the regression BUT I am sceptical because the data covers such a wide geographical area (covering the unemployment rate for the whole region rather than for the individual's state)? I also have regional dummies in my regression if that makes any difference.
Generally speaking, is it the case that any variables that make sense in a cross-sectional OLS regression context also make sense in fixed effects estimation (providing that the variable is not time-invariant)?
The dataset I have been given has two interesting variables - poverty and unemployment rate. Poverty is a dummy which indicates whether the individual's family was below the poverty line last year. Does this variable make sense in a fixed effects regression of wage?
The unemployment rate variable shows the unemployment rate for the local labour market of the individual. However, the dataset I have is restricted and so instead of showing the unemployment rate of the particular state the individual is in, it shows the average unemployment rate of the wider region (north, east, south and west of the country).
Because of this, I was wondering whether this is a variable that makes sense to use. It comes up as highly significant in the regression BUT I am sceptical because the data covers such a wide geographical area (covering the unemployment rate for the whole region rather than for the individual's state)? I also have regional dummies in my regression if that makes any difference.
Generally speaking, is it the case that any variables that make sense in a cross-sectional OLS regression context also make sense in fixed effects estimation (providing that the variable is not time-invariant)?